WATER RESEARCH A Journal of the International Water Association
Editor-in-Chief Mark van Loosdrecht Delft University of Technology Department of Biochemical Engineering Julianalaan 67 2628 BC Delft The Netherlands Tel: +31 15 27 81618 E-mail:
[email protected]
Editors J. Block Université H. Poincaré, Nancy I France David Dixon University of Melbourne Australia Hiroaki Furumai The University of Tokyo Japan Xiaodi Hao Beijing University of Civil Engineering and Architecture China
Damien Batstone The University of Queensland Australia G-H. Chen The Hong Kong University of Science & Technology Hong Kong China Tom Curtis Univ. of Newcastle upon Tyne UK Ana Deletic Monash University USA Francis de los Reyes III North Carolina State University USA Rob Eldridge The University of Melbourne Australia Rosina Girones University of Barcelona Spain Stephen Gray Victoria University Australia
Gregory Korshin University of Washington USA
Kate Grudpan Chiang Mai University Thailand
Anna Ledin Technical University of Denmark Denmark
E.E. Herricks University of Illinois - Urbana USA
Eberhard Morgenroth University of Illinois Urbana-Champaign USA
H-Y. Hu Tsinghua University China
W. Rauch University Innsbruck Austria
P.M. Huck University of Waterloo Canada
Maria Reis Universidade Nova de Lisboa/FCT Portugal
Bruce Jefferson Cranfield University UK
Hang-Shik Shin Korea Advanced Institute of Science and Technology Korea Thomas Ternes Bundesanstalt für Gewässerkunde Germany Stefan Wuertz Univ. of California, Davis USA
Associate Editors Andrew Baker University of New South Wales Australia
Ulf Jeppsson Lund University Sweden Sergey Kalyuzhnyi Moscow State University Russian Federation Jaehong Kim Georgia Institute of Technology USA Jes La Cour Jansen Lund Institute of Technology Sweden G. Langergraber BOKU/Univ. of Natural Res. and Applied Life Scs. Austria
S-L. Lo National Taiwan University Taiwan Y. Matsui Hokkaido University Japan A. Maul Université Paul Verlaine-Metz France Max Maurer EAWAG Switzerland How Yong Ng National University of Singapore Singapore Satoshi Okabe Hokkaido University Japan S.L. Ong National University of Singapore Singapore Jong M. Park Pohang University of Science & Technology Korea Susan Richardson U.S. Environmental Protection Agency USA Miguel Salgot University of Barcelona Spain David Sedlak University of California, Berkeley USA Jean-Philippe Steyer LBE-INRA France M. Takahashi Hokkaido University Japan Kai Udert EAWAG Switzerland V.P. Venugopalan BARC Facilities India E. von Sperling Federal University of Minas Gerais Brazil J. Zilles University of Illinois Urbana USA A.I. Zouboulis Aristotle University Greece
Editorial Office Mailing address Water Research Radarweg 29 1043 NX Amsterdam The Netherlands Tel: 31 20 485 2701 Fax: 31 20 485 2521; 31 20 485 3264 E-mail:
[email protected]
Publication information: Water Research (ISSN 0043-1354). For 2010, volume 44 is scheduled for publication. Subscription prices are available upon request from the publisher or from the Elsevier Customer Service Department nearest you or from this journal’s website (http://www.elsevier.com/locate/watres). Further information is available on this journal and other Elsevier products through Elsevier’s website (http://www.elsevier.com). Subscriptions are accepted on a prepaid basis only and are entered on a calendar year basis. Issues are sent by standard mail (surface within Europe, air delivery outside Europe). Priority rates are available upon request. Claims for missing issues should be made within six months of the date of dispatch. Author Enquiries: Please visit this journal’s homepage at http://www.elsevier.com/locate/watres. You can track accepted articles at http://www.elsevier.com/ trackarticle and set up e-mail alerts to inform you of when an article’s status has changed. Also accessible from here is information on copyright, frequently asked questions and more. Contact details for questions arising after acceptance of an article, especially those relating to proofs, will be provided by the publisher. Orders, claims, and journal enquiries: Please contact the Elsevier Customer Service Department nearest you: St. Louis: Elsevier Customer Service Department, 11830 Westline Industrial Drive, St. Louis, MO 63146, USA; phone: (877) 8397126 [toll free within the USA]; (+1) (314) 4537076 [outside the USA]; fax: (+1) (314) 5235153; e-mail:
[email protected] Oxford: Elsevier Customer Service Department, The Boulevard, Langford Lane, Kidlington OX5 1GB, UK; phone: (+44) (1865) 843434; fax: (+44) (1865) 843970; e-mail:
[email protected] Tokyo: Elsevier Customer Service Department, 4F Higashi-Azabu, 1-Chome Bldg, 1-9-15 Higashi-Azabu, Minato-ku, Tokyo 106-0044, Japan; phone: (+81) (3) 5561 5037; fax: (+81) (3) 5561 5047; e-mail:
[email protected] Singapore: Elsevier Customer Service Department, 3 Killiney Road, #08-01 Winsland House I, Singapore 239519; phone: (+65) 63490222; fax: (+65) 67331510; e-mail:
[email protected] Application for membership of International Water Association should be made to: Executive Director, IWA, Alliance House, 12 Caxton Street, London SW1H 0QS, U.K. (Tel.: +44 207 654 5500; Fax: +44 207 654 5555; e-mail:
[email protected]; website: http://www.IWAhq.org.uk). Registered Charity (England) No. 289269. Individual membership is available from £30 upwards. For details contact IWA.
water research 44 (2010) 4271
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Publisher’s note Elsevier would like to take this opportunity to announce a change in the Editorial Team of Water Research. Prof. Mogens Henze has decided to step down as Editor in Chief, in order to be able to devote more time to other activities. Mogens has been associated to the journal since 1985 and has been Editor in Chief for 10 years. He has steered the journal to achieve its present success in terms of reputation, Impact Factor and publication speeds. We want to warmly thank Mogens for his tremendous input in the journal and we are happy that he will now have the time to pursue his other scientific and personal interests.
0043-1354/$ e see front matter doi:10.1016/j.watres.2010.07.040
We are pleased to be able to introduce Prof. Mark van Loosdrecht to you as our new Editor in Chief. Mark has been an editor for Water Research for eight years and has an extensive background in environmental biotechnology at the interface of engineering and microbiology. He is presently Professor at the Department of Biochemical Engineering at Delft University of Technology and Scientific director of the asellus program at KWR Watercycle Research Institute We wish Mark success in his new position.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Activated sludge model (ASM) based modelling of membrane bioreactor (MBR) processes: A critical review with special regard to MBR specificities A. Fenu a,*, G. Guglielmi b,1, J. Jimenez c, M. Spe`randio d,e,f, D. Saroj g, B. Lesjean h, C. Brepols i, C. Thoeye a, I. Nopens j a
Aquafin NV, Department of Research and Products Development, Dijkstraat 8, 2630 Aartselaar, Belgium Department of Civil and Environmental Engineering, University of Trento, Via Mesiano, 77 e 38100, Trento, Italy c Anjou Recherche, Veolia Eau, Chemin de la Digue, BP 76, 78 603 Maisons Laffitte Cedex, France d Universite´ de Toulouse; INSA, UPS, INP; LISBP, 135 Avenue de Rangueil, F-31077 Toulouse, France e INRA, UMR792 Inge´nierie des Syste`mes Biologiques et des Proce´de´s, F-31400 Toulouse, France f CNRS, UMR5504, F-31400 Toulouse, France g Department of Urban Water and Sanitation, UNESCO-IHE Institute for Water Education, Westvest 7, 2611 AX Delft, Netherlands h Kompetenzzentrum Wasser Berlin, Cicerostr. 24, 10709 Berlin, Germany i Erftverband, Am Erftverband 6, D 50126 Bergheim j BIOMATH, Department of Applied Mathematics, Biometrics and Process Control, Ghent University, Coupure Links 653, 9000 Gent, Belgium b
article info
abstract
Article history:
Membrane bioreactors (MBRs) have been increasingly employed for municipal and indus-
Received 7 October 2009
trial wastewater treatment in the last decade. The efforts for modelling of such wastewater
Received in revised form
treatment systems have always targeted either the biological processes (treatment quality
9 April 2010
target) as well as the various aspects of engineering (cost effective design and operation).
Accepted 3 June 2010
The development of Activated Sludge Models (ASM) was an important evolution in the
Available online 11 June 2010
modelling of Conventional Activated Sludge (CAS) processes and their use is now very well established. However, although they were initially developed to describe CAS processes,
Keywords:
they have simply been transferred and applied to MBR processes. Recent studies on MBR
Membrane bioreactors (MBR)
biological processes have reported several crucial specificities: medium to very high sludge
Modelling
retention times, high mixed liquor concentration, accumulation of soluble microbial
Activated sludge models
products (SMP) rejected by the membrane filtration step, and high aeration rates for
Kinetic parameters
scouring purposes. These aspects raise the question as to what extent the ASM framework
Influent fractionation
is applicable to MBR processes. Several studies highlighting some of the aforementioned
Soluble microbial products (SMP)
issues are scattered through the literature. Hence, through a concise and structured
Exo polymeric substances (EPS)
overview of the past developments and current state-of-the-art in biological modelling of MBR, this review explores ASMebased modelling applied to MBR processes. The work aims to synthesize previous studies and differentiates between unmodified and modified applications of ASM to MBR. Particular emphasis is placed on influent fractionation, biokinetics, and soluble microbial products (SMPs)/exo-polymeric substances (EPS) modelling,
* Corresponding author. Tel.: þ32 3450 4511. E-mail addresses:
[email protected] (A. Fenu),
[email protected] (G. Guglielmi),
[email protected] (J. Jimenez),
[email protected] (M. Spe`randio),
[email protected] (D. Saroj),
[email protected] (B. Lesjean),
[email protected] (C. Brepols),
[email protected] (C. Thoeye), ingmar.nopens@ ugent.be (I. Nopens). 1 Present address: E.T.C. Engineering srl, Via Praga, 7, 38121 Trento, Italy. 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.007
4273
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
and suggestions are put forward as to good modelling practice with regard to MBR modelling both for end-users and academia. A last section highlights shortcomings and future needs for improved biological modelling of MBR processes. ª 2010 Elsevier Ltd. All rights reserved.
Contents 1. 2.
3.
4.
5.
1.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1. General overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application of unmodified ASMs to MBR processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. ASMs application to MBRs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Influent fractionation for unmodified ASMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1. Experimental measurement of influent COD fractions in MBR studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.2. “Assessment of influent COD fractions with trial-and-error methods in MBR studies” . . . . . . . . . . . . . . 2.2.3. Outlines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Process kinetics and stoichiometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.1. Parameter sensitivity in MBR vs CAS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.2. Nitrification kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.3. Denitrification kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.4. Phosphorus removal kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.5. Oxygen uptake rate (OUR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.6. Oxygen transfer rate (a-factor) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.7. Sludge production . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.8. Outlines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Application of modified ASMs to MBR processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Integrating membrane rejection studies in ASM models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.1. EPS and SMP definition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.2. Modelling objectives of modified ASMs to MBR processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Influent fractionation for modified ASMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Overview of stand-alone EPS and SMP models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1. Stand-alone EPS models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2. Stand-alone SMP models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Overview of ASM-extensions incorporating EPS/SMP concepts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. Model identification e UAP/BAP kinetics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.1. Outlines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Outlook and future perspectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Influent characterization for the unmodified ASMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Process kinetics for the unmodified ASMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Process kinetics for the modified ASMs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4. Application of ASM models at full-scale, hydraulics and aeration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Introduction
The membrane activated sludge process, commonly referred to as membrane bioreactor (MBR), combines an activated sludge process and membrane micro or ultra-filtration to separate the treated water from the mixed liquor. The first membrane bioreactors were developed in the late 1960s, in “side-stream” configurations, but market penetration became significant only after the commercialization of submerged MBR configurations (Judd, 2006), which proved to be energetically superior. Since the advent of MBR technology, factors such as the need to comply with stringent environmental legislation
4273 4275 4276 4276 4277 4277 4278 4278 4278 4279 4279 4282 4282 4283 4283 4283 4284 4284 4284 4284 4285 4285 4286 4286 4286 4288 4289 4289 4290 4290 4290 4290 4291 4291 4291 4291
and a reduced foot-print were seen as able to mitigate the higher capital and operational cost. Recently, technological advances (i.e., improved configurations) and scientific investigations have led to further significant cost reduction. In early efforts on dynamic modelling of wastewater treatment processes, only two state variables were considered describing degradation of substrate and formation of biomass by first order kinetics (McKinney, 1962). However, the later models incorporated an increasing number of state variables and process descriptions based on widely accepted Monodtype kinetics. The increased understanding of various complex processes further enhanced the models and
4274
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
Nomenclature AOB Ammonia Oxidizing Biomass AS Activated Sludge ASP Activated Sludge Process ASM Activated Sludge Model BAP Biomass Associated Products Autotrophic decay rate bA Heterotrophic decay rate bH BNR Biological Nutrient Removal CAS Conventional Activated Sludge CODEPS Chemical Oxygen Demand Content of EPS CODcell CellularChemical Oxygen Demand CODsol,in Soluble Influent Chemical Oxygen Demand Biodegradable Chemical Oxygen Demand CODb CODtot,IN Total influent Chemical Oxygen Demand DO Dissolved Oxygen DOC Dissolved Organic Carbon EPS Exoecellular Polymeric Substances Fraction of Biomass that ends up as Microbial fB Products Fraction of BAP Produced during Cell Lysis fBAP Fraction of UAP Produced during Substrate fUAP Production fp Fraction of inert COD generated in biomass lysis F/M Food to Micro organisms ratio HRT Hydraulic Retention Time sCOD SolubleChemical Oxygen Demand Half saturation coefficient for BAP KBAP Hydrolysis Rate of BAP kh,BAP EPS formation coefficient Keps Half saturation Coefficient For Microbial Products Kmp Half saturation Coefficient For Ammonia KNH Half saturation Coefficient For Nitrate KNO Autotrophic Saturation Coefficient for Oxygen KOA Heterotrophic Saturation Coefficient for Oxygen KOH Hydrolysis Rate of UAP kh,UAP UAP formation rate coefficient ofHeterotrophs kUAP UAP formation rate coefficient ofAutotrophs kUAPa SMP Half saturation coefficient forHeterotrophs KSMP Ratio between the mg/l of Xtss and the COD of the iTS_X” different fractions LC-OCD Liquid Chromatography - Organic Carbon Detection MBR Membrane Bioreactor MLSS Mixed Liquor Suspended Solids MLVSS Mixed liquor volatile suspended solids, MP Microbial products MW Molecular Weight NOB Nitrite Oxidizing Biomass OM Organic Matter
a steady-state aerobic model was developed (Marais and Ekama, 1976) which later evolved into a well established dynamic model (Dold et al., 1980). The dynamic model included key hypotheses viz. bi-substrate and death regeneration. Such early dynamic models showed their very useful application to design, optimization and control of wastewater
OHO OUR PAO PE PN PHA PS qfe qpp rs rS,BAP RSF SBAP SCOD Si sMBR SMP SNH SNO SPO4 SRT Ss SUAP TKN UAP VSS WWTP Xaut XDNPAO Xi Xhet XPAO Xsto Xs XTSS Xu YH YHD YPAO YSMP YMP Yobs gUAP,H gMP,A gMP,H gUAP,A mH mA mSMP hNO3
Ordinary Heterotrophic Organism Oxygen Uptake Rate Phosphate Accumulating Organisms People Equivalent Proteins Polyhydroxyalkanoate Polysaccharides Maximum Rate For Fermentation Rate Constant for Storage XPP Specific substrate utilization rate Production/Consumption Rate of BAP Relative Sensitivity Function BAP concentration Soluble COD concentration Inert Soluble Fraction Submerged Membrane Bioreactor Soluble Microbial Products Ammonia Concentration Nitrate Concentration Orthophosphate Concentration Sludge Retention Time Biodegradable Soluble Fraction UAP concentration Total Kjeldahl Nitrogen Utilization Associated Products Volatile Suspended Solids Waste Water Treatment Plant Autotrophic Biomass Concentration Fraction of denitrifiers in the PAO biomass Inert particulate fraction Heterotrophic Biomass Concentration PAO Biomass Concentration Organics stored by heterotrophs Biodegradable Particulate Fraction Total Sludge Suspended Solid Concentration Non-biodegradable Organic Particulate Matter Yield Coefficient For Heterotrophic Biomass Yield Coefficient For Heterotrophic Biomass In Anoxic Conditions Yield Coefficient (Biomass/PHA) Yield coefficient for growth on SMP Yield coefficient for growth on MP Observed Yield Heterotrophic UAP Formation Constant Autotrophic MP formation constant Heterotrophic MP formation constant Autotrophic UAP formation constant Heterotrophicmaximum growth rate Autotrophicmaximum growth rate SMP maximum growth rate Anoxic reduction factor
treatment systems with a variety of configurations for carbon, nitrogen and even phosphorous removal. Activated Sludge Modelling or ASM-modelling represented an important milestone in modelling of biological treatment processes. ASMs were initially developed to describe Conventional Activated Sludge (CAS) processes under correspondingly typical
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
operating conditions. Nevertheless, they have been used since the late nineties to simulate biomass kinetics in MBRs systems as well, provided that some necessary adaptations are made to allow for the specific behaviour of these systems. The specificities of the MBR biological process compared to the CAS process have been extensively reported (Ng and Kim, 2007): higher sludge retention times (SRTs), higher mixed liquor concentration and viscosity, accumulation of microbial products rejected by the membrane filtration step, high aeration rates for scouring and good nitrification performance. These all influence and characterize the MBR process behaviour. The logical question one can pose now is whether all current knowledge about ASM based modelling for CAS is simply transferable to MBR systems, or in other words, how the current MBR process understanding can be merged into the ASM framework. MBR literature has been very prolific lately but studies and results on biological modelling are scattered, and a systematic overview of the state-of-the-art of all scientific work performed to date in this area is missing. More specifically, both academics and end-users of biological modelling processes are in need of a concise summary to support their decision making process in terms of MBR design and operations. This review paper aims to (i) provide a concise overview of the past developments and current state-of-the-art in biological modelling of MBR systems and (ii) highlight weaknesses in the current models and propose future needs. For reasons of clarity, the review is broken down into two major sections: unmodified versus modified ASM models for MBR. The former contains the literature studies where the plain ASM models, as described in Henze et al. (2000), are used as such in MBR applications and where only parameter estimations have been performed to improve the model performance (i.e. matching model predictions with measured experimental data). Unmodified ASM models also contain efforts that expand the description of certain biokinetic processes, but without adding state variables to the model state vector. The decision to include those applications here was taken from the perspective that these models are more readily applicable in practice. With modified ASM models, we target all applications where the plain ASM models are extended in terms of (1) biokinetic process models and (2) SMP/EPS models. Here, additional state variables are introduced in the model either for a better description of certain processes already present in ASMs or to allow for the description of processes formerly lacking in ASMs (e.g. EPS/SMP). End-users will most probably be interested in the former section as the models described in that section are intended for use in practice, whereas the latter section focuses on academic work to improve process understanding and, hence, is still in a more basic research phase and further away from practical use. Both sections deal with the typical aspects that are methodologically encountered during a modelling process such as (i) influent characterization and (ii) determination/calibration of kinetics and stoichiometry. A final section highlights shortcomings and the authors’ conception of future needs for improved biological modelling of MBR processes. Finally, to limit the scope of the paper, discussions have been kept to
4275
MBRs treating domestic wastewater. However, applications of models at different scales (lab, pilot and full-scale) are considered.
1.1.
General overview
Since MBRs encompass the Activated Sludge Process (ASP) as their fundamental process, ASMs have been applied for the biological modelling of MBRs as reported in various studies (inter alia Chaize and Huyard, 1991; Wagner and Rosenwinkel, 2000; Wintgens et al., 2003; Delrue et al., 2008). ASMs are robust dynamic models which are widely used for activated sludge based wastewater treatment process understanding, design, optimization and control. There are various versions of ASMs, and one or another is preferred on account of various factors e.g. modelling objective, desired accuracy, calibration effort, ease of use and relevance of process types, etc. Various versions of such dynamic models viz. ASM1, ASM2, ASM2d, ASM3 (Henze et al., 2000) have been developed during the two decades since the introduction of the first version i.e. ASM1 (Henze et al., 1987), by assimilating the developments in process understanding pertaining to wastewater treatment systems. The advent of ASM1 introduced the Gujer matrix form of model presentation, which assimilates all the process descriptions in a condensed and elegant way. ASM1 does not incorporate biological phosphorous modelling although the process had already been established prior to the advent of ASM1. However, most of the plant at that time did not incorporate this process and only required predictions of C and N removal, aeration capacity requirements and sludge production. Later, biological P-removal gained importance and was therefore included in ASM2. The understanding of denitrification in the biological phosphorous removal processes evolved further, resulting in ASM2d, which incorporated the processes pertaining to denitrifying PAOs. ASM2d might not have been considered very important if only carbon and nitrogen removal had been targeted; however, the model played an important role in the understanding of the complexity of combined nitrogen and phosphorous removal processes (Henze et al., 2000). Its practical application was hampered by the large number of parameters in the model. Hence, in the mean time, ASM1 continued to be the state-of-the-art model for dynamic modelling in wastewater engineering, despite certain defects that became apparent in its application, e.g. no nitrogen and alkalinity limitations for heterotrophic bacteria were included, no differentiated decay rates of nitrifiers under aerobic and anoxic conditions were considered and intracellular storage of PHAs was not addressed. The introduction of ASM3 (Gujer et al., 1999) aimed to correct the defects of ASM1 and presented a new standard for ASM based modelling. The original ASM3 did not incorporate biological P-removal (unlike ASM2 or ASM2d), chemical precipitation, growth of filamentous organisms or pH calculations. However, these processes can be connected as add on modules (Henze et al., 2000). One example of this is the extended ASM3 for biological Premoval (Rieger et al., 2001). The process descriptions of anaerobic processes are not part of ASM1/ASM3. Their application is limited to (aerobic)
4276
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
ASP with possible extension to include the anoxic conditions and (partial) anaerobic reactor integration (for Bio-P processes only). For a complete description of anaerobic processes, other models such as the anaerobic digestion model (ADM) are used, which are not the part of current review. The assumptions underlying the ASMs related with influent fractions might no longer be applicable when industrial wastewaters are mixed; various facets of influent fractionation are outlined in the Section 2.2.
2. Application of unmodified ASMs to MBR processes The objective of this section is to give a concise overview of the application of unmodified ASMs. The terminology “unmodified” should be interpreted in terms of modelling scope as well as model structure. Unmodified ASM models are to be used for modelling MBRs when objectives are similar to those originally stipulated for the use of ASM for CAS systems (Henze et al., 2000): process design, effluent characterization, oxygen demand and sludge production. In these cases, plain ASM models are used. The section includes efforts where plain ASMs were calibrated for MBR or where slight modifications to biokinetic processes were performed using solely the state variables defined in the original ASMs. This section includes the models that are closest to being applicable in practice and is, thus, very useful for end-users. It includes an introduction and treats two important aspects: (i) influent characterization and (ii) determination/calibration of kinetics and stoichiometry.
2.1.
ASMs application to MBRs
The application of ASMs are presumably meant for ASP operation in the ranges of conventional ASP operating parameters, e.g. SRT range 3e15 d, HRT range of 3e5 h and MLSS range 1.5e4 g/L for completely mixed systems (Metcalf and Eddy, 2003). A recent study on design and operating experience with municipal MBRs in Europe has reported the ranges of various parameters (Itokawa et al., 2008). The HRT of 13 MBR plants have been reported to be in the range of 2.8e8.1 h, with most of the plant operating at HRT between 4 and 6 h. The MLSS of 11 MBR plants have been reported to be in the range of 7e13.5 g/L, with most of the plant operating at MLSS higher than 10 g/L. Further, the SRT values of 7 plants have been reported to range between 15 and 40 days. For municipal MBR applications, it seems reasonable to define the SRT below 15 days as “low SRT range”, and SRT above 40 days as “high SRT range”. Efforts have been made over the last 15 years towards appropriate application of ASMs for MBRs. While early trials (Chaize and Huyard, 1991) used the very basic form of ASM1, using default parameter values, performing no systematic calibration or influent characterization, recent efforts have presented various aspects of systematic calibration of key and sensitive parameters along with emphasis on the influence of influent wastewater characterization in terms of various ASM based fractions (Delrue et al., 2008; Sperandio and Espinosa, 2008). The early study of Chaize and Huyard (1991), based on
a laboratory scale MBR fed with domestic wastewater, aimed to model effluent COD, TKN and sludge production at two HRT values (viz. 8h and 2h) and very high SRT (nearly 100 d). The MBR system was modelled with ASM1 using default values of parameters (Henze et al., 1987). The predicted effluent COD was reported to be slightly lower than that observed, and the predicted TKN was found quite close to the observed value. However, the major disagreement was reported on solids concentration. The model predicted a lower solids concentration than observed, and the solids concentration prediction was relatively better at higher HRT. The probable reason was thought to be the very high SRT (i.e. 100 days). These outcomes illustrate that a non-calibrated ASM1 is able to give a reasonable estimate of effluent COD and TKN, but is insufficient for very low HRT and very high SRT systems. Hence, this imposes care in the application of those models and in the investigation of appropriate parameter sets valid for these systems under variable operational conditions. This sets the scene for investigating the whereabouts of the encountered limitations. The application of ASM1 moved towards better understanding of model parameters and, hence, a more systematic calibration, taking into account the nature of the MBR biology and specific operating conditions. The ASM1 application on side-stream MBR by Jiang et al. (2005) stressed the importance of various sensitive biokinetic parameters and influent wastewater characterization. More recently, Delrue et al. (2008) commented that despite some difficulties, ASM1 is suitable for modelling MBR plants if influent characterization and systematic calibration of aeration can be taken care of. The incorporation of storage phenomena (Krishna and van Loosdrecht, 1999; Gujer et al., 1999) is a unique feature of ASM3 and might play a role in the case of MBRs on account of possibilities of low organic load conditions (Wintgens et al., 2003). Nevertheless, ASM1 has been shown to be sufficient where conditions are not favourable to storage phenomena (Delrue et al., 2008). In the aim of modelling MBRs over a large range of SRTs, Sperandio and Espinosa (2008) used ASM1 and ASM3 and commented that ASM models could provide satisfactory prediction of aerobic biological processes in submerged MBRs, although these could be improved for high SRT conditions. Studies so far are not conclusive as to whether ASM1 or ASM3 is better for MBRs. It appears that the application of ASMs, in their original forms, often needs careful calibration of parameters, especially for sludge production and nitrification modelling. The issue of the significance of high SRT, which was a matter of further attention even in early MBR modelling studies (Chaize and Huyard, 1991), remains a relevant point. It has been reasoned in recent research (inter alia Masse´ et al., 2006; Sperandio and Espinosa, 2008) that high SRT operation of MBRs is linked with corresponding influence on MBR specific sludge production and autotrophic biology. Throughout, it can generally be observed that all the recent efforts aiming at an accurate biological modelling of MBRs focus on MBR specificities (e.g. high SRT operation, membrane retained microbial metabolites etc.) and the corresponding parameter adjustment and modifications required in ASMs.
4277
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
2.2.
Influent fractionation for unmodified ASMs
The monitoring practice for municipal WWTPs typically relies on collecting influent and effluent wastewater samples to assess the inlet organic and nutrient loading and the impact on the receiving water body. However, when process modelling is carried out, a more detailed characterization of feedwater COD is needed, which supplies the modeller with information about the degree of biodegradability (readily, slowly and inert) and the physical aggregation state (soluble, particulate) of the influent substrate. Generally, two different approaches are reported in the MBR-modelling literature for COD characterization: (i) integration of chemical/physical/ biological methods and (ii) application of “trial-and-error” procedures which aim to fit experimental observations (i.e. MLSS in the biotank) with model simulation by tuning the different COD fractions in a reasonable range of values. In the following subsections inferences from these two different approaches are presented and discussed.
2.2.1. Experimental measurement of influent COD fractions in MBR studies Jiang et al. (2005) compared two methods for wastewater characterization to calibrate ASM1 in a side-stream membrane bioreactor (sludge age w20 days, HRT e8 h). The first method (also referred to as “chemical-biological”) combined the respirometric technique proposed by Vanrolleghem et al. (1999) and physical separation (filtration on 0.45 mm fibreglass paper). First, the 0.45 mm filtration was carried out to quantify soluble versus particulate COD. Once assessed, the soluble and particulate biodegradable COD (SS and XS respectively) were determined with a respirometric test (S0/X0 ¼ 1/200). By assuming the soluble inert fraction of influent COD (SI) to correspond to 90% of effluent COD (according to Vanrolleghem et al., 2003), the authors determined the inert particulate fraction (XI) as the difference between CODtot,IN and (SS þ SI þ XS). In order to evaluate the effectiveness of this method, a trial-and-error procedure was also used to determine XI and XS, by comparing a steady-state COD mass balance with measured and simulated MLSS concentration in the biotank. The two methods provided very close results (e2.7% on XS value). The second approach applied by the authors (“physical-chemical method”) used the STOWA protocol for wastewater characterization (STOWA, 1996; STOWA, 1999, Roeleveld and van Loosdrecht, 2002) and coupled physical separation (0.45 mm) with BOD analysis. The soluble inert COD (SII) in the influent was calculated as 90% of soluble effluent COD and soluble biodegradable COD (SS) was given by the difference (CODsol,in e SI). Then, biodegradable COD (CODb ¼ SS þ XS) was calculated as a function of BOD5 and XS was calculated from CODb and SS; finally, XI was determined as the difference between CODtot,IN, SI, SS and XS. The mean results of the wastewater fractionation on 16 influent samples are summarized in Table 1. The most significant difference between the two approaches occurred in the XI fraction, which was significantly higher when determined with the “physical-chemical” method compared to the “chemical-biological” one. This discrepancy reflected a relevant overestimation of the MLSS concentration in the
Table 1 e Comparative evaluation of the two methods for wastewater characterization. Total COD: 579 g mL3 (adapted from Jiang et al., 2005). Parameter SI SS XI XS
Chemicalebiological (g m3)
PhysicaleChemical (g m3)
33 (5.7%) 214 (37.0%) 58 (10.0%) 274 (47.3%)
33 228 141 177
biotank for the “physical-chemical” method under steadystate conditions (13245 g m3 for the physical-chemical method versus 10020 g m3 experimentally measured). A bench-scale MBR (0.016 m3) was modelled by Sperandio and Espinosa (2008) with ASM1 and ASM3 over a wide range of sludge ages (10O110 days), with special focus on wastewater characterization and excess sludge. During the four experimental periods, a combined physical-biological method was used for influent wastewater fractionation: the nonbiodegradable soluble COD (SI) was assumed to be equal to the effluent COD, the inert particulate COD (XI) was measured at the end of long-term BOD measurement (30 days), the slowly biodegradable (XS) and readily biodegradable COD (SS) were obtained by combining respirometry (Spe`randio and Paul, 2000) and 0.45 mm filtration. The simulated MLVSS trend indicated that ASM1 was able to predict sludge production at SRTs shorter than 50 days, while an overestimation was observed at SRT ¼ 110 days. In contrast, an underestimation is reported by the authors in the application of ASM3 at SRT values of 10 days and 30 days, and a slightly better prediction of the MLVSS concentration at 110 days. According to the authors, most of the discrepancy between the ASM1 model and the experimental observations is due to the large quantity of non-biodegradable organic particulate matter (XU) related to the death-regeneration concept, which is replaced by the endogenous decay concept in ASM3. In terms of surplus sludge production (expressed as Yobs) the work points out that, the longer the SRT, the more relevant the impact of the inert fraction in the influent feedwater (XI) on the sludge composition. Therefore, the authors recommend introducing a slow hydrolysis mechanism in the standard ASM1 and ASM3 to correctly predict sludge production at very long SRT. This also matches the conclusion of other groups (Rosenberger et al., 2000; Rosenberger, 2003), who demonstrated that, for SRTs above 80d, an hydrolysis factor could be used to better simulate the MBR sludge production. Witzig et al. (2002) later demonstrated that there was actually no hydrolysis at high sludge age, but that the bacteria went into maintenance mode and did not grow any more, mathematically impacting the sludge yield in a similar way to hydrolysis (Drews et al., 2005). This fact, known for pure cultures, was shown to be also applicable for mixed cultures in activated sludge. In a recent work, a small pilot-scale MBR (sludge age e36 days) was modelled by Sarioglu et al. (2009) with a modified ASM1 to describe the simultaneous nitrification-denitrification process. All the growth and decay processes were coupled with switching functions defining the diffusion limits of oxygen and substrate. A combined physical-chemical-
4278
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
biological method was used for COD characterization; the biodegradable COD (CODb) was determined according to Roeleveld and van Loosdrecht (2002), while the SI and XI fractions were measured according to Germirli et al. (1991) and Orhon et al. (1999) respectively. Once SS was calculated as the difference between influent soluble COD and SI, the XS fraction was determined by mass balance. This calibration resulted in a fairly good fit of MLVSS and MLSS in the biotank. A respirometry-based approach for wastewater fractionation has been reported by Guglielmi et al. (2009), who calibrate an extended version of ASM3 that includes the simultaneous growth of heterotrophs on both storage products (XSTO) and readily biodegradable substrate (SS), originally proposed by Sin et al. (2005). The model has been successfully validated in terms of sludge production under dynamic conditions on a large pilot-scale MBR run under an SRT of 20O25 days. In detail, the total biodegradable COD was determined according to Ekama and Marais (1986), whereas the soluble biodegradable COD was estimated according to the single-OUR method proposed by Ziglio et al. (2001), with sodium acetate as standard. Afterwards, SI was calculated as the difference between filtered COD in the influent (0.45 mm) and the soluble biodegradable COD, being usually equal to 90e95% of COD concentration in the permeate. Finally, XI was quantified with a mass balance.
2.2.2. “Assessment of influent COD fractions with trial-anderror methods in MBR studies” When ASM-modelling tools are used for process optimization in a full-scale installation, there is typically a general tendency towards “trial and-error” procedures. In the application of ASM1 to a full-scale MBR in Gue´thary, France (10,000 PE, 1600 m3 d1; SRT ¼ 30O60 days; F/ M ¼ 0.02O0.05 kgBOD5 kgMLVSS1 d1), Delrue et al. (2008) compared three different methods for soluble/particulate fractionation, namely filtration on fibreglass filter, 0.1 mm membrane filtration and coagulation-flocculation followed by a 0.1 mm filtration. Then, in order to divide COD into biodegradable/non-biodegradable organic matter, the method proposed by Roeleveld and van Loosdrecht (2002) was compared with a trial-and-error procedure based on MLVSS steady-state mass balance (fitting method). In their conclusion, the combination of fibreglass filtration and MLVSS fitting was chosen as the most reliable protocol for influent COD characterization. A model-based optimization of a full-scale MBR was carried out at Ro¨dingen, Germany (Erftverband, 2001), in which the influence of the weather conditions on the COD fractionation was described. The plant (3000 PE, SRTe25 days; combined sewer) was modelled in ASM1 and calibrated by means of a trial-and-error procedure on the observed sludge production over two different periods (57 and 28 days). A comparative evaluation of inert fractions (SI and XI) is shown in Table 2, with an increase of both non-biodegradable components during rainy events. The catchment area of the MBR was small but widely ramified and the sewer had only a mild slope so that effects of run-off, the flushing out of deposits and dilution by the rain water were very pronounced and produced a visible effect on the plant. Unlike what is observed in CAS modelling, the need for a more frequent
Table 2 e Results of the COD fractionation of a large scale MBR working on combined sewer influent (Erftverband, 2001).
XI, % of COD SI, % of COD
Dry weather
Rain weather
20 8
40 15
calibration of influent COD fractions was reported by the authors, due to the shorter HRT and reduced buffering volume.
2.2.3.
Outlines
Summing up the inferences for ASM applications to membrane bioreactors, the available literature highlights the crucial role that design/operational SRT plays in influent COD characterization. Particularly, the actual operational sludge age can influence the choice of a proper method for the determination of XS and XI fractions since at very long SRT a slow but significant hydrolysis of the “inert” fraction takes place. From the modelling point of view, this results in a general overestimation of the sludge production. This issue is strictly related to the biodegradability concept itself, which depends not only on the substrate characteristics and the substrate/biomass affinity but also on the time available for biological degradation. In terms of practical process design and operation, this aspect has great relevance for MBR installations with noticeable seasonal fluctuations of the influent organic loading (e.g. tourist areas), where SRT can increase up to 50e60 days or more. In this case, a trial-anderror method tuning XS and XI seems more appropriate. On the other hand, when operating at more conventional sludge ages (<30O40 days), respirometry-based fractionation seems to be able to provide satisfactory characterization.
2.3.
Process kinetics and stoichiometry
Next to influent characterization (previous section) and module structure (unaltered in this case), kinetic and stoichiometric parameters provide degrees of freedom for matching ASM model predictions with experimentally collected data. The standard ASM models come with default parameter values, obtained from wide experience at both lab-scale and full-scale. However, the inclusion of membranes in activated sludge systems may cause associated changes in these default kinetic parameters compared to CAS systems. The main effects on process kinetics are possibly due to (i) specific biomass selection (high SRTs, free bacteria retention), (ii) high biomass concentration, and (iii) high hydrodynamic constraints imposed by continuous or cyclic air scouring of the membrane element instead of quiescent conditions in the settler. The crucial parameters for design and control of MBRs are: the sludge suspended solids (XTSS) impacting the excess sludge production and the oxygen transfer rate (a-factor), the removed and residual nitrogen species (SNH, SNO), the residual phosphorus concentration (SPO4) and the oxygen consumption rate (OUR, and SO).
4279
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
In the next subsection, the parameter sensitivity will be primarily discussed. The following subsections will look more deeply at the impact of MBR on the kinetics of the different processes.
2.3.1.
moderately to very influential, depending on process operational conditions. In fact, Jiang et al. (2005) studied the local parameter sensitivity in a model calibrated for a specific case, with specific operating conditions (SRT ¼ 20 days, DO range ¼ 3e8 mg/l). At different operating conditions, the sensitivity of the parameters might change. That is why sensitivity analysis is systematic and necessary in every calibration exercise. An alternative to this approach is the use of global sensitivity analysis which explores the entire parameter space and yields average sensitivities (and not just locally around a single point in this space), as performed by Benedetti et al. (2008), with an ASM2 model for BSM2 (Benchmark Simulation Model no.2). Global methods are, however, always more computationally expensive and interpretation of their results also needs scrupulous care.
Parameter sensitivity in MBR vs CAS
Given the similar nature of the process, it would be reasonable to think that the influence of most kinetic constants is comparable in CAS and MBR processes. The reasons pointed out above might, however, lead to different system behaviours. During the calibration exercise, sensitivity analysis can help by assessing the most influential parameters for a given process and biological reaction (sludge production, uptake of COD, denitrification and nitrification). The results of such a sensitivity analysis for ASM1 applied to an MBR are given in Table 3. Relative sensitivity functions (RSF) were calculated by Jiang et al. (2005) for several model variables (Y) towards all model parameters (q), as described by the equation (1). RSF ¼
q dY Y dq
2.3.2.
Nitrification kinetics
As autotrophic bacteria are very sensitive to the environmental conditions, the impact of including membranes for the solid-liquid separation on the kinetics of nitrogen removal has received considerable attention. Ammonium consumption and residual concentration is sensitive to nitrification parameters, in decreasing order of influence: mA, bA, KNH, KOA and YA (Jiang et al., 2005). In ASMs, the nitrification rate is controlled by the concentration of active autotrophic bacteria (Xaut) stabilized in the process, which is imposed by the conversion yield YA, the influent nitrifiable nitrogen, and the decay rate bA. The nitrification rate is basically linked to the product mAXaut. For this reason, the parameters bA and mA are highly correlated and it is impossible to identify them simultaneously if the active biomass is not stabilized at different levels, i.e. data should be collected at different SRTs in a continuous process (Sperandio and Espinosa, 2008). Munz et al. (2008) highlighted the difference between the ammonium-nitrogen concentration in CAS and MBR effluents (both with 50 days SRT): for MBRs the nitrification was more stable and complete. In the same way Parco et al. (2006) found a specific autotrophic uptake rate for MBR 1.8 times higher than for conventional Biological Nutrient removal (BNR). By operating CAS and MBR in the same conditions (SRT), Manser et al. (2005) did not observe any significant difference in the
(1)
RSFs rescale the sensitivity function in order to allow easy comparison and fair ranking accounting for different orders of magnitude in parameters and variables. Ranking is done according to the absolute values of RSF as shown in Table 3. A finite difference approach is typically used to compute the derivative. The influent wastewater characterization, with particular regard to XI, XS, SS and SNH fractions, appears to be extremely influential on MLSS concentration and effluent quality, which confirms the importance of the previous discussion (Section 2.2). The stoechiometric parameter (YH) and kinetic parameters (bH, bA, mmaxH and mmaxA) are very to moderately influential on the MLSS concentration and effluent quality, whereas the remaining parameters do not seem to be candidates for change when performing a calibration. Petersen et al. (2002) used the same sensitivity function applied to CAS (with an SRT of 8.6 days). The most influential parameters were the same as for MBR but with different degrees of importance. Kinetics parameters (KOA, KNH and kh) which had no influence in Jiang et al.’s (2005) model became
Table 3 e Results of sensitivity analysis with the calibrated ASM1 model. RSF
RSF<0.25 (Not influential) 0.25 < RSF<1 (Moderately influential) 1 < RSF<2 (Very influential) RSF>2 (Extremely influential)
Steady-state calibration XTSS
SS
SNO
SNH
fp, bA, ka, kh, KNH, KNO, KOA, KOH, KS, KX, mmaxA, mmaxH, YA, SNO bH - (YH, XI)*-
fp, bA, ka, kh, KNH, KNO, KOA, KOH, KX, mmaxA, YA, YH SS, SNH,SNO bH KS - XS*-
fp, bA, ka, kh, KNH, KNO, KOA, KOH, KS, KX, mmaxA, mmaxH, YA, SNO bH Y H - (YH, bA, KNH, KOA, SS)*-
fp, ka, kh, KNO, KOA, KOH, KS, KX, mmaxH, YA, YH,, SNO
YH, SNH
mmaxH - (YH,, mmaxH, Ks, kh)* XS
(SNH, XS)*
XI, XS, SS
XS, SS, SNH - mmaxA*-
RSF results with the calibrated ASM1 model (Jiang et al., 2005): MBR with SRT ¼ 20 (In standard character). RSF results with calibrated ASM1 model (Petersen et al., 2002) CAS: with SRT 8.6 days (In bold and with *).
bA, KNH, SNH - (bA, KNH, SNH, KOH, KOA, kh, SS, XS)*mmaxA, SS YH, * XS
4280
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
maximum specific ammonium uptake rate but authors showed that the ammonium uptake rate became higher in MBR compared to the CAS during transient shock loads, especially at low temperature and relatively low DO (dissolved oxygen). The difference between nitrifiers in MBRs and CAS may depend on cumulative factors such as: (i) the different microorganism selection (not demonstrated); (ii) the higher bioavailability of substrates which can be due to the smaller size of flocs (Manser et al., 2005); and (iii) the tendency of nitrifiers clusters to grow at different places in the flocs or with lower density (presence of Ammonia Oxidizing Bacteria aggregates on the surfaces of MBR flocs was observed only by Munz et al. 2008). These findings might however also depend on a different mass balance, i.e. higher autotrophic biomass concentration can be maintained by a better retention of solids (for example, it is clear that solids loss in the activated sludge can contribute to a more or less significant decrease of nitrifiers depending on the SRT). In this sense, to better understand this section, the reader must consider that lower affinity constants or higher growth rates do not necessary mean higher removal rates, since the nitrification mass balance must be taken into account entirely. Several authors have found discrepancies when modelling MBR processes with ASM1 using the initial default values of Henze et al. (1987) for nitrification (Jiang, 2007; Sperandio and Espinosa, 2008). Sperandio and Espinosa (2008) used ASM1 and ASM3 to calibrate an MBR working at a large range of SRT, through the parameters mA and bA. The authors reported that ASM1 default values (0.8; 0.04d1) overestimated ammonia removal for all the SRTs studied, whereas ASM3 (1, 0.15d1) gave better results but minimized the SRT influence. The data obtained from this study led to a mA ¼ 0.45 d1 and bA ¼ 0.04 d1. Autotrophic growth rate is known to be variable from one process to another even for activated sludge processes. This means that more data will be necessary to conclude on the best set of parameters for MBR, i.e. mA should be measured properly in future works on MBR. Since this parameter is bacteria-specific, an analysis of microbial species (AOB, NOB) by molecular techniques in MBRs would probably help in interpreting the observed changes in mA. Concerning the better behaviour of MBRs reported by Munz et al. (2008), Parco et al. (2006), and Manser et al. (2005), an explanation may be offered by the reduction of the halfsaturation constants KNH and KOA, which directly influence the residual ammonia concentration model predictions. Regarding the KNH, experiments on MBRs in the literature report values from 0.15 to 1 mgN/l and tend, in some cases, to be lower than the ASM1 default value (ASM default values is 1 mgN/l), leading to an improvement of ammonia transfer (Delrue et al., 2008; Spe´randio et al., 2005; Jiang et al., 2009; Jimenez et al., 2008; Erftverband, 2001, 2004). A parameter set from the different studies on MBRs in urban wastewater treatment is reported in Table 4. On the other hand, Manser et al. (2005) reported observed KNH values lower than ASM1 default values but similar for CAS (0.14 mg N/l) and MBR (0.13 mg N/l), concluding that the limitations induced by larger floc density were not that significant for ammonia transfer. It should also be noted that KNH also varied significantly from one activated sludge to another (Sperandio and Espinosa, 2008).
The KNH value seemed to be linked to internal transfer resistance but also to external mass transfer resistance. KNH was reported to change from 0.3 to 0.6 when MLSS increases from 3 to 8 g/l, pointing to an increase of viscosity (Spe`randio et al., 2005; Sarioglu et al., 2009) and an increase of external transfer resistance. When MLSS increased, ammonia uptake rate (AUR) and mA decrease indicating a concentration effect (Pollice et al., 2008; Ramphao et al., 2006). Parco et al. (2006) explained that the MLSS effect is related to ammonia diffusion limitations and not to DO transfer limitation. Regarding KOA, ASM1 models fitted to respirometric data resulted in a KOA range of 0.18e0.4 mgO2/L in MBR (Spe´randio et al., 2005; Jiang, 2007; Jimenez et al., 2008) and tended to be lower than the ASM1 default value of 0.4 mgO2/l. A parameter set from the different studies on MBRs in urban wastewater treatment is reported in Table 4. Manser et al. (2005) showed that values differed significantly from MBR to CAS. In detail, he observed lower KOA values for MBR (0.18 mg O2/L) in comparison with the CAS (0.79 mg O2/l) and explained this variation with the lower transfer resistance induced by MBR smaller floc size (which became negligible for floc sizes under 100 mm). Similar conclusions have been drawn in other studies (Jiang, 2007; Jimenez et al., 2008) as shown in Tables 4 and 5. Jimenez et al., (2008) observed that KOA had a very high sensitivity to nitrification (based on RSF results) when DO concentration was lower than 2 mg/l and average floc size was lower than 35 mm. However, it is not always obvious that floc sizes are lower in MBRs than in CAS, since they depend on the hydrodynamic conditions and on the type of aeration (intermittent or continuous). An inverse tendency was obtained by Spe`randio et al. (2005), who reported results indicating that an sMBR can have larger flocs and larger values for KNH, if aeration and shear stress are limited. Similar results were recently found by Sarioglu et al. (2008, 2009). On the other hand, floc size is also influenced by SRT (Masse´, 2004). This author showed that an SRT increase from 9 to 106 days led to a diminution of the average floc size in the MBR from 240 to 70 mm. Deflocculation of biomass could be due to the increase in aeration when MLSS increases, or to the reduction of F/M ratio, leading to the decrease in the active fraction of the sludge (Spe`randio et al., 2005). Therefore, although there is a tendency in MBRs to find a value for the half-saturation constants, KOA and KNH, lower than the ASM1 default values, these parameters depend on the operating conditions (SRT, MLSS concentration, viscosity, oxygen concentration, floc size distribution). Concerning the autotrophic decay rate (ba), Manser et al. (2006) did not find significant differences between CAS and MBR concerning the ammonia oxidizing biomass (AOB), but a slight difference concerning the nitrite oxidizing biomass (NOB): AOB ba was found to be 0.13d1, both in CAS and MBR, while the NOB ba was 0.28d1 and 0.17d1 for CAS and MBR respectively. The conversion yield YA does not seem to be influential (see Table 3) but can become relevant when dissolved oxygen sensitivity is studied under dynamic conditions (Jiang et al., 2005). Moreover, this parameter was measured by Jiang et al., (2005), yielding the value (0.25 gN/gCOD), close to the ASM default value (0.24gN/gCOD).
Table 4 e Parameter set from the different studies on MBR in urban wastewater treatment. Model
Nitrification
d d-1
ba
d-1
KNH
mgN -NH4/l mgO2/l
YA Denitrification %XI COD oxidation YH Sludge prod. bH KO,NOB KOH
Experimental Default Jiang et al., Spe´randio Manser Jiang methods ASM1 2005 et al., 2005 et al. (2005) 2007 references ASM1 ASM1 ASM3 ASM1 ASM2d
20 [2], [5]; [7], [8], [6],[10] [1], [5], [7], [8], [10] [8], [10]
0.8 0.05e0.15
0.08
1
[4], [8], [9], [10] gCOD/gN [3], [8]
0.4 0.24
0.25
% COD gCOD /gCOD d-1 mgN/l mgO2/l
[3], [8]
15 0.67
0.72
[2], [4], [8] [9] [4], [8]
0.62 0.5 0.2
10e110 0.45
20
Sarioglu et al., 2008 ASM1 endogenous decay model 38 1
Delrue, Jimenez Erftverband 2008 et al., 2008 2001, 2004 ASM1
30e60
ASM1 modified 15 0.8
0.04
0.055
0.06
0.25e0.6
0.2
2
1
1
0.2
1.25
1
0.25
0.18
ASM1
RWTH 2008
Range of values
ASM1
0.45e1.00
0.15
0.04e0.15 0.1
0.10e2.00 0.18e1.25 0.24
17.5
15
0.66
0.25
0.4 0.13 0.05
0.24 2 1
0.67
1 0.22
0.1
e 0.52e0.92 0.63e0.67 0.03e0.47 0.24e0.4 0.13e2 0.05e1
Jiang et al., 2005: YH with acetate addition overestimated, inducing an underestimation of fp Spe´randio et al., 2005: Assessment of (ma, ba) valuable in a large range of SRT Manser et al. (2005): Accurate estimation of KOA: Correlation with floc size distribution Jiang, 2007: Biological P-removal calibration (cf. paragraph P-removal) Sarioglu et al., 2008: Specifically adapted for modelling simultaneous nitrification-denitrification Jimenez et al., 2008: Correlation with floc size distribution Erftverband, 2001, 2004: Assessment for simulation of timeline over several weeks
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
SRT mmaxA
KOA
Comment
Units
[1] Henze et al. (1987), [2] Ekama et Marais (1986), [3] Solfrank et Gujer (1991), [4] Kappeler et Gujer (1992), [5] Lesouef et al. (1992), [6] Nowak (1994), [7] Spanjers et al. (1995), [8] Vanrolleghem et al. (1999), [9] Ficara et al. (2000) [10] Van Haandel (2007).
4281
4282
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
Table 5 e Biokinetic parameters and size distribution found in MBR modelling literature. Mean floc size (mm) ASM1 Manser et al. (2005) CAS (Manser et al., 2005) Jimenez et al. (2008) Jiang et al. (2009) Sarioglu et al. (2008) SNdN Sarioglu et al. (2009) SNdN
35 307 35 30e50 N.d. N.d High viscosity (SVI ¼ 800e1000 ml/g)
K_OA gO2/m3
K_OH gO2/m3
K_O,NOB gO2/m3
0.40 0.18 0.79 0.25 0.20 1.25 2.00
0.20 0.05 0.16 0.10 0.20 1.00 1.75
0.50 0.13 0.47 e 0.17 e
N.d.: not determined. SNdN: simultaneous nitrification and denitrification.
In order to obtain a clear independent estimation of nitrification parameters, tailored batch tests are suggested. Substrate excess or substrate absence conditions would be necessary to respectively determine mA and bA, (Henze, 1992; Vanrolleghem et al., 1999; Lesouef et al., 1992) while batch tests with substrate addition at various oxygen concentrations would be necessary to estimate the oxygen half-saturation constant (Kappeler and Gujer, 1992; Ficara et al., 2000). YA and KNH could be estimated with ammonia addition in batch tests, as in methods described by Vanrolleghem et al. (1999), Van Haandel and Van Der Lubbev (2007).
2.3.3.
an influence on oxygen transfer and consequently on oxygen transfer in the denitrification zone. In fact, some MBR configurations present sludge recirculation from the aerated membrane tank to the denitrification zone and, hence, the level of dissolved oxygen has an important effect on denitrification potential (Sarioglu et al., 2008). Dissolved oxygen level can inhibit denitrification in anoxic growth reaction modelling, through the parameter KOH. Manser et al. (2005) found in MBRs, a KOH value of 0.05 mgO2/l, lower than the 0.16 mgO2/l for a CAS process working at the same conditions. However, this parameter also depends on system hydrodynamics and configuration.
Denitrification kinetics
Nitrate concentration in the bioreactors and in the effluent depends on both nitrification and denitrification. It is dependent on the nitrified nitrogen, which is linked to the amount of nitrogen entering the system and indirectly to the amount of nitrogen assimilated into the sludge. But more significantly, removal of nitrate depends on the amount of easily and slowly biodegradable substrate used for denitrification (Xs, Ss), and also the heterotrophic yield YH (Jiang et al., 2005). Obviously residual nitrate also depends on denitrification kinetic parameters. However, the higher the SRT, the more sensitive the denitrification rate will be to total biodegradable COD (Xs þ Ss) and to the endogenous respiration. Specific denitrification rates are extremely dependent on SRT via mass organic load (denitrification rates measured in MBRs are often very low at high SRT) and consequentially design guidelines (Pinnekamp, 2006) recommend a ratio VAE/VAX ¼ 50/50%, whereas it could be at least 75/25% in CAS. Denitrification rates were measured by Parco et al. (2007) for an MBR in UCT configuration (20 days SRT and a mass load of 0.14 gCOD/gMLSS.d). Conventional denitrification rates of 0.25 mgNO3/mgSS.d (similar to CAS) were obtained. The authors concluded that kinetic parameters for denitrification could be applied directly to MBR BNR systems. From these results, the conclusion is that the reduction factor for anoxic growth (hg), and the anoxic heterotrophic yield (YHD) are probably not different in MBR compared to CAS. In contrast to the nitrification process, denitrification is apparently less modified by the membrane configuration. However, halfsaturation constants (KNO, KOH) which control the effect of low concentrations of nitrate or oxygen on the kinetics have not been specifically determined yet for MBR. As for the nitrification, Manser et al. (2005) showed that floc size distribution had
2.3.4.
Phosphorus removal kinetics
From an extensive study on an MBR UCT process, Parco et al. (2007) concluded that kinetic parameters for biological P-removal are comparable in MBR and CAS. The anaerobic P-release and acetate consumption rates, and the anoxic and aerobic P-uptake rates were very close to the range of values in the literature for conventional BNR systems with mixed cultures. Moreover, the rates obtained for different concentrations of volatile suspended solids (VSS) indicated no effect of the sludge concentration on these rates. Acetate consumption rates were zero order with respect to acetate concentration in agreement with the studies in the literature. Aerobic and anoxic P-uptake rates indicate a relatively low fraction of denitrifiers in the PAO biomass (XDNPAO/XPAO was around 15 to 36%). Jiang et al. (2008) used ASM2d to predict phosphorous removal. With default parameters, the model overestimated nitrate concentration while underestimating phosphorous concentration. The authors calibrated the model simultaneously reducing SA production in the anaerobic compartment and the aerobic/anoxic phosphorous uptake rate (qfe ¼ 1d1, qpp ¼ 1.1 d1 and hNO3,PAO ¼ 0.4), by trial and error. However, it should be noted that, after including the SMP concept, these parameters could be restored to default ASM2d values. Recently, MBR research at full-scale (Silva et al., 2009) showed that unexpected high biological phosphorous removal was obtained in MBRs not designed for EBPR. In the full-scale MBR located in Schilde (Belgium), significant biological P removal was reported though the process scheme did not contain anaerobic compartments, effluent nitrate concentration was generally higher than 3 mg/l, (Bixio et al.,
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
2006; Fenu et al., in press). Silva et al., (2009) explained this phenomenon by the high floc compactness and density of EPS inducing anaerobic micro-niches. The higher sludge viscosities in MBR systems (when compared to CAS systems), might also impede mixing and hydrodynamic homogeneity, creating artificial local anaerobic zones. These assumptions have not been fully demonstrated and need further investigation.
2.3.5.
Oxygen uptake rate (OUR)
The dynamics of the oxygen uptake rate (OUR) are generally considered sensitive to kinetic parameters. For example, a highly loaded period can result in Xs accumulation, whereas during low loaded periods hydrolysis of these slowly biodegradable compounds controls the OUR dynamics (Ekama and Marais, 1986). However, in MBR processes this conclusion is no longer valid during very low-loaded operation (high SRT). In that condition, the actual oxygen uptake rate (instantaneous oxygen demand) varies mainly with the flux of biodegradable substrate (Ss, Xs and SNH) entering the system. The stoichiometric parameters (YH, YA) are then more important than the kinetics, as no Xs, Ss or SNH accumulation is normally observed in that condition.
2.3.6.
Oxygen transfer rate (a-factor)
The a-factor is defined as the ratio of the volumetric transfer coefficient under process conditions (i.e. with mixed liquor) to the clean water transfer coefficient. This is therefore a normalized parameter depending primarily on the mixed liquor characteristics. Together with the hydrodynamic conditions (such as aeration types and powers), it determines the standard aeration efficiency (SAE) of aeration systems (Krause and Cornel, 2007). In these respects, the energy efficiency of activated systems can be described as inversely proportional to the a-factor, and the modelling of the biological aeration demand in MBRs should account for the effect of specific process conditions on the a-factor. The a-factor is mainly influenced by the viscosity of the activated sludge which is associated with the MLSS concentration (Krause et al., 2003; Germain et al., 2007). As the viscosity is unknown in many cases, Krause (2005) gives an approximate function of the a-factor versus the MLSS concentration of a ¼ e0.056MLSS for the design of municipal MBRs. As a result, a then equals 0.5 for an MLSS concentration of 12 g/L. Actually, a bandwidth of values can be found. However, the lower a-factor in MBR applications constitutes a significant difference compared with the CAS processes: while for an MLSS concentration lower than 5 g/L the a-factor may be in the range 0.7e1, it drops to a value of about 0.5 for 10 gMLSS/L and may go down as low as 0.2 for 20 gMLSS/L. Despite this, other factors also contribute to this behaviour. Results from Germain et al. (2007) suggested that bound carbohydrates and soluble COD also had a secondary impact on the a-factor: a -factor increased with increasing bound carbohydrates (facilitating the formation of large flocs), and decreased with increasing soluble COD (probably due to the presence of surfactants), although with specific impact about half of that generated by the MLSS concentration. Note that the parameters of bound proteins, bound COD, soluble carbohydrate, soluble protein, and mass median diameter of
4283
the floc were monitored but not identified as variables having a significant influence on a-factor within the range of parameters studied.
2.3.7.
Sludge production
Total sludge suspended solid concentration and excess sludge production are clearly of great importance in WWTPs. A realistic prediction of the concentration of active components (XHET, XAUT, XPAO) is in fact crucial for dynamic simulations. MBRs systems are run at average to very high SRTs. In these operating conditions, suspended solids are mainly composed of inert particulate matter, originating from both influent and biomass decay. High SRTs increase the amount of energy spent on maintenance rather than on growth, and average biomass concentration notoriously increases, due to the reduced quantity of biomass wasted. In aerobic zones, specific OUR will decrease as a result of inert compounds accumulation and reduction in the active fraction of biomass (Tan et al., 2008). Moreover, in MBR systems operated at very high SRTs, biodegradation of COD fractions, considered inert in ASMs as described in the previous sections, induces an overestimation of sludge production. This is definitely a major drawback of ASM models for modelling MBR in a large range of SRTs (Sperandio and Espinosa, 2008).Changes in the parameters YH, bH and fp can severely modify the predicted XTSS (Jiang et al., 2005; Sperandio and Espinosa, 2008). The experimental estimation of these parameters has been attempted by several research groups and the main outcomes are reported below. Jiang et al., (2005) measured the conversion yield of heterotrophs (YH) in MBR processes, through respirometric measurement with acetate. The obtained YH (0.72 gCOD/gCOD at 23 C) was higher than the default value in ASM1 (0.67 gCOD/gCOD at 20 C), but this higher value is not specific to MBRs and it is probably due to the storage phenomenon with acetate which is easily converted to PHA (Majone et al., 1999; van Loosdrecht and Heijnen, 2002). YH values obtained with a single carbon source should be considered with precaution as they are substrate-specific. A similar value can be obtained with acetate as a pure substrate in the activated sludge process. The mean value of 0.63e0.67 gCOD/gCOD is then more valuable for domestic wastewater considering the large variety of carbon source (carbohydrates, proteins, alcohols, carboxylic acids.). Heterotrophic decay rate (bh) seems to be close to the ASM default value. In a first study, Jiang et al. (2005) measured bH with respirometric tests (Vanrolleghem et al., 1999) and found a value of 0.25 d1, lower than the default value (0.4 d1). The authors explained this lower value by the decrease of predation in the MBR. However, in a subsequent study using ASM2, Jiang (2007) used the default value for bH in ASM2 (0.4 d1) with a good fit of the mixed liquor COD. Sarioglu et al. (2008) estimated bH ¼ 0.24 d1 (endogenous respiration concept), similar to the value in ASM3. Modifications of YH, bH, and fp parameters need to be carefully examined as they play a major role in other processes. For example an increase of YH would reduce the electron consumption, i.e. oxygen consumption or nitrate removal. For this reason, it is considered preferable to fit XI, which is logically wastewater specific, rather than other parameters.
4284
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
Table 4 summarizes the parameter set found by the authors who have calibrated MBR systems in urban wastewater treatment. It includes some comments and references to methods.
2.3.8.
Outlines
Nitrification parameters seem to be the most affected by the differences between CAS and MBR. However, they depend on both hydrodynamic and operational conditions. It is relatively risky to choose mA, bA, YA, and KNH and KOA by combining values obtained from different studies, as the coherence of each set of parameters is necessary. It is thus recommended to give special attention to these parameters, measuring them with proper methods if necessary. Denitrification rates seem to be similar to CAS ones, and consequently parameters are not different. An exception is the oxygen half-saturation constant KOH (used as inhibition parameter of oxygen for denitrifiers). This parameter could be affected, depending on the oxygen level in the denitrification zone and the small floc size which could ease the oxygen mass transfer. Moreover, a detailed characterization of the anoxic hydrolysis rate and the endogenous respiration rate is still needed, controlling both the slow and endogenous denitrification rate. Concerning kinetic parameters for biological phosphorus removal, data and literature are still insufficient on this topic. Phosphorus removal performance in MBRs seems to be slightly better than in CAS in some cases, but data are still needed to demonstrate whether these differences are explained by kinetic parameters of PAO species or different operating conditions (local anaerobic zone in biological tank, absence of anaerobic zone of settlers,.).
3. Application of modified ASMs to MBR processes The objective of this section is to give a concise overview of the application of modified ASMs. The terminology “modified” should be interpreted in terms of modelling scope as well as model structure. This section discusses the following aspect: (i) The impact of the phase separation mechanism on the ASMs (ii) The influent fractionation for modified ASMs (iii) An overview of the stand-alone EPS and SMP models (iv) An overview of ASM extensions with EPS/SMP concepts (v) The over-parametrization of modified ASMs.
3.1. Integrating membrane rejection studies in ASM models The phase separation mechanism is the main technological distinction between CAS and MBR systems. The mechanism is essentially a sieving effect performed by membranes with a nominal pore size of typically 0.02 up to 0.2 mm: all particles whose size is larger than the membrane pore size are retained, whereas the smaller dissolved fractions are not. Hence, flocs, bacteria, biopolymers such as polysaccharides and proteins, and organic colloids are in great extent retained. Humic and low molecular weight substances can instead pass the membrane (Drews et al., 2006). The retention of the
biopolymers is in fact a specificity of MBR systems compared with conventional ASP. The above mentioned fractions accumulate in the mixed liquor (inter alia, Drews et al., 2007; Malamis and Andreadakis, 2009), and are susceptible of biodegradation. Shin and Kang (2003) report an initiation of the reduction of microbial products at SRTs longer than 10e20 days. This reduction is commonly agreed to be more significant for polysaccharides than for proteins (inter alia, Al-Halbouni et al., 2008). With increasing SRTs, a reduction in the molecular weight distribution of the mixed liquor supernatant particles has also been reported (Shin and Kang, 2003; Ng et al., 2006). In contrast, the accumulation of microbial products did not appear to trigger any deviation from the expected biomass metabolic activity (Shin and Kang, 2003). Current ASM models fail to account for many of these specific MBR features. They neither distinguish between protein and polysaccharide fractions, nor account for shifts in the molecular weight distribution. But it is important to realize that, in systems with low organic loads (such as MBRs), the retained molecules may have a significant impact on the metabolic path, allowing further use of carbon based metabolites (Furumai and Rittmann, 1992). These shortcomings in the current ASM models could be overcome by expanding the models with EPS/SMP concepts.
3.1.1.
EPS and SMP definition
EPS are a complex mixture of Proteins (PN), acid, polysaccharides (PS), lipids, DNA and humic acids. They surround cells, create a matrix for microbial flocs and films (Liao et al., 2001) and allow micro-organisms to live continuously at high cell densities in stable mixed population communities (Wisniewski, 1996; Laspidou and Rittmann, 2002a,b). They are further differentiated into bound EPS, the fraction bound to the sludge flocs, and soluble EPS, the fraction able to move freely between sludge flocs and the surrounding liquor (Rosenberger and Kraume, 2002). As it is difficult to differentiate between SMP and soluble EPS, the latter are commonly denominated as SMP. It must also be inferred that EPS analysis relies on its extraction from the sludge flocs. So far, the scientific community has not agreed on a standard procedure, and a comparison of data set generated from different extraction methods becomes complex (Judd, 2006; Liu and Fang, 2002). SMPs are defined as soluble cellular components or debris that are released during cell lysis, diffuse through the cell membrane, are lost during synthesis, or are excreted for some purpose. Substrate utilization, biomass decay, and EPS hydrolysis are believed to be the major processes contributing to SMP formation. With respect to “SMP formation through substrate utilization”, if intermediate products of metabolic processes are included in the SMP definition, substances that do not have a microbial origin but come directly from the substrate tend to be wrongly included (Noguera et al., 1994). The problem has been taken into serious consideration by a number of researchers, but within the scopes of this paper, it is sufficient to refer to SMPs as “any soluble material that leaves the effluent from a biological system that was not present in the influent”, (Barker and Stuckey, 1999).
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
3.1.2. Modelling objectives of modified ASMs to MBR processes From a end-user point of view, expanding the models with EPS/SMP concepts is suggested only if the following modelling objectives are pursued:
3.1.2.1. Linking biology with fouling. The extension of ASM models with soluble microbial product (SMP) and Exo Polymeric Substances (EPS) concepts is undoubtedly crucial when predicting membrane fouling. SMP/EPS have been proved to accumulate in MBR systems as a consequence of a high membrane rejection and low biodegradability (Shin and Kang, 2003; Drews et al., 2007; Liang et al., 2007). Their influence on fouling, or their use as indirect indicators of fouling propensity through biomass deflocculation (De la Torre et al., 2009), has been evaluated and acknowledged by numerous researchers (inter alia Rosenberger et al., 2005, 2006; Jarusutthirak and Amy, 2007; Le-Clech et al., 2006; Zhang et al., 2006; Wu et al., 2007). Based on these considerations, the SMP/EPS would be a necessary input for a filtration model and, hence, needs to be incorporated in the biological model in order to explain or predict the filtration performance of the system.
3.1.2.2. Soluble COD prediction. The supernatant of settled mixed liquor is mainly composed of living cells and microbial products. So far it is certain that, in municipal applications, effluent soluble organic matter of activated sludge water from conventional activated sludge systems is mainly SMP (Namkung and Rittmann, 1986; Jarusutthirak and Amy, 2007) and the same can be said for MBR processes (Lu et al., 2002; LeClech et al., 2006; Rosenberger et al., 2006; Aquino and Stuckey, 2008). SMP and EPS are microbial products, not active cells, and represent energy that is not invested in cell growth (Laspidou and Rittmann, 2002a,b). Ignoring SMP and EPS formation could lead to a general overestimation of true cellular growth rates and this would severely under predict the COD effluent (Jiang et al., 2008). This is commonly artificially corrected by an overestimation of Si in the conventional ASM1 model. Finally, it should be pointed out that the COD effluent predictions are not a real concern in municipal MBR systems since values are generally low and stable. Model high SRT processes Furumai and Rittmann (1992), investigated the interactions between nitrifiers and heterotrophs. They reported that, in a CAS at high SRT, SMP formed by nitrifiers promoted heterotrophic growth and reduced the minimum substrate concentration necessary for heterotrophs, allowing heterotrophic growth at very low influent organic concentration. MBR processes generally run with average to high SRTs and the above considerations could be relevant at low F/M ratio (in MBRs, easily as low as 0.01 g COD/g MLSS.d). Bound EPS has been experimentally shown to accumulate with decreasing SRTs (Ng and Hermanowicz, 2005; Masse´ et al., 2006), with a corresponding trend on SMP (Rosenberger et al., 2006; Al-Halbouni et al., 2008). Beyond the prediction on membrane filtration, a proper modelling of this fraction is beneficial for sludge production prediction.
4285
In this work, efforts at ASM model extensions with SMP and EPS concepts in the literature are reviewed along with their relevance for MBRs.
3.2.
Influent fractionation for modified ASMs
When introducing the SMP concentration as a state variable in the model, one of the key issues for COD fractionation is the determination of inert components (Si, Xi) with proper methods, considering the fact that new “inert” substances can be produced in the process as by-products of the microbial metabolism. In this sense, an early modification of ASM1 was proposed by Lu et al. (2001) by modelling a bench-scale MBR fed with synthetic sewage. Wastewater was characterized by means of batch tests aimed at quantifying the inert soluble fraction according to the method proposed by Henze et al. (1987). Activated sludge from the bioreactor (MLSSe13,000 g m3) was washed by ultrafiltered effluent and, after washing, it was diluted to about 2000 O 3000 g m3. The filtered feedwater was afterwards fed to the sludge with an initial COD concentration of 200 g m3. The steady value of soluble COD concentration reached after approximately 4e5 h was assumed to represent the inert soluble COD in the influent (SII). This concentration was slightly reduced in order to take the presence of SMP into account. Similarly, the XI fraction was determined by the same method using substrate pretreated by sonication and by subtracting the SI contribution. Finally, SS and XS were given by the difference between soluble COD and SI and particulate COD and XI respectively. More recently, an extension of ASM1 to the SMP concept has been proposed by Di Bella et al. (2008) in an integrated model for physical-biological wastewater organic removal in MBRs. The model couples the SMP-extended ASM1 firstly proposed by Lee et al. (2002) with a physical model accounting for organics removal in both the membrane and the cakelayer. Here, the COD fractionation comes from a multi-step process including the trial-and-error on MLSS and effluent COD concentrations and an automatic calibration with 10,000 Monte Carlo simulations intended to define the values for the most sensitive parameters of the whole model (including the physical sub-model). Interestingly, in this work, a relevant fraction of influent COD is attributed to the heterotrophic active biomass in the wastewater (Xhet), which is usually assumed to be negligible in the above mentioned research. In the successful application of a modified ASM3 including SMP to a pilot-scale MBR treating pre-settled municipal wastewater working at almost 50 days, Oliveira-Esquerre et al. (2005) used OUR profiles to quantify the SS and XS fractions in the influent feed. In detail, they measured the inert soluble fraction as COD concentration in the permeate after 12 h’ aeration, thus degrading the SMP contribution. To sum up, the implementation of additional processes for specific MBR modelling purposes in the ASM framework is not reflected in a different COD fractionation. As for unmodified activated sludge models, a general trend towards the trialand-error fractionation is observed when longer sludge ages are operated, whereas the respirometry-based approach is generally preferred at more conventional values of SRT.
4286 3.3.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
Overview of stand-alone EPS and SMP models
Some models have been developed as stand-alone descriptions of the concepts of production and degradation of EPS and SMP. Others have focused on integrating the latter concepts into the ASM type of models. This distinction is used in the overview given below.
3.3.1.
Stand-alone EPS models
An early model to characterize microbial products formation was proposed by Luedeking and Piret (1959) for the fermentation of lactic acid (Eq (1) in Table 6). In this equation, the first term accounts for EPS formation associated with a first order growth (with K1 as the fraction of substrate electron shunted to EPS formation) and the second term represents EPS formation associated with a non-growth term. The main objections to this simple model were (i) the inconsistent production rate values proved by Luong and Muchaldani (1988); (ii) the fact that a mechanism of EPS dissolution was not included although theoretical evidence supported this hypothesis (Laspidou and Rittmann, 2002a,b). Laspidou and Rittmann, 2002a,b studied the EPS mass balance in a continuous flow reactor (Eq (2) in Table 6). In this equation, the first term is the product of the substrate utilization rate rs, the active biomass Xa and the part of substrate electrons shunted to EPS formation, Keps. The second term quantifies the rate of EPS loss due to hydrolysis, using a firstorder relationship with respect to EPS. The hydrolyzed EPS would then become soluble EPS or SMPs. The Laspidou and Rittmann model was criticized on 2 major points: (i) The formation of bound EPS is said to be growth-associated, and produced in direct proportion to substrate utilization. During transient conditions of organic shock loads this would theoretically lead to very high EPS concentrations. Aquino and Stuckey (2008) did not observe this, and proposed to model the formation of EPS as a non-growth-associated product (eq. (3) in Table 6). This entails that the EPS formation rate can only be high at high concentrations of biomass. (ii) The EPS hydrolysis/ dissolution rate was 0.17 d1 in Laspidou and Rittmann’s works, but was recently reduced to 0.02e0.03 d1 by Jang et al. (2006), and Aquino and Stuckey, 2008. In this respect, a valuable hypothesis is that the high concentration of “hydrolysis end products” would reduce the hydrolysis rate. (Jang et al., 2006). It is interesting to note that, regarding EPS formation kinetics, Aquino and Stuckey (2008) proposed to calculate the EPS formation rate Keps, so that the steady-state model would
yield a ratio of bound EPS to VSS (bEPS/VSS) close to the experimental value of PN-like-EPS to biomass. Keps resulted in 0.03 mgCODEPS/mgCODcell/d, which is in accordance with Robinson et al. (1984).
3.3.2.
Stand-alone SMP models
SMP models have been developed since the late eighties. Namkung and Rittmann (1986) put forward the following SMP subdividision: 1. UAP, i.e. SMP that are associated with substrate metabolism and biomass growth and are produced at a rate proportional to the rate of substrate utilization. 2. BAP, i.e. SMP that are associated with biomass decay and are produced at a rate proportional to the concentration of biomass. This method of subdivision has been widely accepted. There seems to be a general consensus on UAP formation and degradation mechanisms. UAP formation results from substrate utilization and is proportional to the rate of substrate utilization and biomass concentration (Luedeking and Piret, 1959; Namkung and Rittmann, 1986; Lu et al., 2001; Laspidou and Rittmann, 2002a,b). Contrary to Laspidou and Rittmann’s approach, Lu et al. (2001) differentiated the active biomass into Xhet and Xaut eqs. (3e6) in Table 7) and UAP cannot be consumed but only produced by Xaut (Fig. 1). Recently, a modelling study conducted with aid of LC-OCD and batch tests characterization (Jiang et al., 2008) further hypothesized two types of UAPs: the UAPs produced during storage formation of readily biodegradable COD would have a lower MW and would be biodegradable; the UAPs produced during the utilization of storage products would instead have a higher MW and would be more refractory. However the study was limited to the modelling of the first proposed UAP type. Unlike for the UAP case, there is no consensus on BAP production and degradation mechanisms. With regard to the BAP production, Laspidou and Rittmann (2002a,b) suggested that BAP were produced solely by EPS hydrolysis (Eq.(7) in Table 7). This hypothesis was shown to be weak. (i) Hydrolyzed (soluble) EPS and BAP revealed different physicochemical properties (Ramesh et al., 2006). Aquino and Stuckey (2008) demonstrated that both soluble EPS and cell lysis products were the sources of BAP (Eq.(8) in Table 7). (ii) Regarding the kinetics, Jang et al. (2006) found that the BAP/ UAP kinetics of the Laspidou and Rittmann model could not be applied in their specific tests. When employing a maximum utilization rate for UAP/BAP, they found that the UAP/BAP utilization exceeded the formation. It was therefore assumed that UAP/BAP accumulation inhibited the degradation rate. In the ASM models experiences, EPS are generally not considered and BAP are not produced from EPS hydrolysis but from cell lysis processes (Jiang et al., 2008) or additionally from
Table 6 e EPS model rates in literature. Leudeking and Piret., 1959 (Eq. (1))
Laspidou and Rittmann, 2002a,b (Eq. (2))
Aquino and Stuckey, 2008 (Eq. (3))
reps ¼ K1 m X þ K2 X
rEPS ¼ Keps rs Xa Khyd EPS
rEPS ¼ Keps Xa Khyd EPS
4287
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
Table 7 e Summary of main formation and degradation rates for UAP and BAP. UAP formation rate Laspidou and Rittmann, 2002a,b (Eq. (4))
rUAP ¼ K1 rquap
S Xbm Ks þ S
UAP degradation rate Laspidou and Rittmann, 2002a,b (Eq. (6))
rUAP ¼ rquap
UAP Xbm Kuap þ UAP
Lu et al., 2001 (Eq. (5))
rUAP ¼ mhet Xhet or rUAP ¼ maut Xaut
Lu et al., 2001 (Eq. (7))
rUAP ¼ mSMP
SSMP Xhet KSMP þ SSMP
BAP formation rate Laspidou and Rittmann, 2002a,b (Eq. (8))
Aquino and Stuckey, 2008 (Eq. (9))
rBAP ¼ Khyd EPS
rBAP ¼ K2 X þ Khyd EPS
BAP degradation rate Laspidou and Rittmann, 2002a,b (Eq. (10))
Jiang et al., 2008 (Eq. (11))
rbap ¼ qbap
BAP Xbm Kbap þ BAP
rbap ¼ Kh;bap Sbap XH
(*) Switching functions are not reported since their introduction depends on the process where UAP is formed or degraded.
hydrolysis of particulate biodegradable organic matter (Lu et al., 2001). The BAP degradation has been considered by many researchers (Lu et al., 2001; Laspidou and Rittmann, 2002a,b; Aquino and Stuckey, 2008; Oliveira-Esquerre et al., 2006) as a direct degradation and only Jiang et al., (2008) assumed hydrolysis (Eq. (12) at Table 7). Jiang reported that 63% of the BAP had a molecular weight larger than 20 kDa, which makes it unlikely that such large molecules pass the cell membranes directly. They therefore concluded that BAP were hydrolyzed yielding fermentable COD. Overall, Jiang et al. (2008) proposed 3 steps for each biomass type to determine the complete BAP dynamics, (formation, hydrolysis, degradation). The SMP equilibrium concentration results from production/degradation mechanisms but also depends on SMP retention by the membrane. The latter aspect is given little consideration in MBR modelling. Authors have mainly measured or fitted it as a steady value, independently of the process specificities. The percentage of SMP permeating through the membrane ranges from 0 to 100% (Jang et al., 2006; Jiang
Fig. 1 e Alpha-value in dependence of MLSS (Krause and Cornel, 2007).
et al., 2008; Zarragoitia-Gonza´lez et al., 2008; Silva et al., 1998; Lu et al., 2001, 2002). Drews et al. (2007) studied the impact of ambient conditions on the rejection of MBRs. According to the latter authors, the rejection of SMP components appears to decrease at higher temperatures and higher nitrification activity. The modelled rejection factors should thus take experimental observations in considerations. The predominance of UAP over BAP or vice versa needs to be discussed for two main reasons (i) One of the two fractions could have more impact in terms of soluble COD, and an SMP fraction could be neglected leading to a simplification of the model. The exclusion of a fraction would be of great benefit in terms of modelling since it would reduce the number of variables. In fact ASM models in the early nineties used to consider only BAP fractions. However, the process conditions may determine the predominant fraction case by case. A rule of thumb is that the BAP fraction tends to predominate at high SRTs (Furumai and Rittmann, 1992) or in steady-state conditions (Aquino and Stuckey, 2008), while UAP predominate in the SMP production when the rate of substrate degradation is high (Aquino and Stuckey, 2008; Laspidou and Rittmann, 2002a,b); (ii) One of the two fractions could have more impact in terms of fouling predictions. Zhang et al., (2007) operating an MBR fed with external carbon source proved that the concentration of large molecule OM was greater in BAP than that in UAP, (over 18% of molecules with more than 100 K), being the main cause of the increasing resistance. Conversely, Rosenberger et al. (2006) observed that the SMP fraction produced at different sludge retention times in MBR units operated with municipal wastewater led to a specific long-term fouling rate (per SMP mass unit) higher with 15d SRT than with 8d SRT, even though much more SMP was produced at 8d SRT than at 15d SRT. This demonstrates that the SMP fraction produced by the biological system differs not only in quantity but also in quality depending on the environmental conditions, with crossed impact of both criteria on the fouling propensity of the mixed liquor.
4288
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
3.4. Overview of ASM-extensions incorporating EPS/ SMP concepts The SMP concept was incorporated into activated sludge model No. 1 (ASM1) in the early 90s (Orhon et al., 1989; Artan et al., 1990). First, a very simple SMP model including only BAP was developed (Orhon et al., 1989). So-called SP (equivalent to BAP) is produced proportional to the hydrolysis of particulate COD (Xs) and they are assumed non-biodegradable. The model was further developed to include UAP (Artan et al., 1990). However, this model combines the concepts and degradation kinetics of UAP and BAP resulting in strong parameter correlations. Moreover, the model lacks experimental support. Lu and coworkers have incorporated a very complex SMP model into ASM1 (Lu et al., 2001) and ASM3 (Lu et al., 2002) in MBR studies (Fig. 2). However, the COD of their SMP model is not balanced, i.e. the loss of substrate COD is not equal to the sum of formed UAP COD, formed biomass COD and consumed oxygen. In addition, 8 SMP-related parameters are introduced, but the experimental results available for model calibration are limited to steady state soluble COD (SCOD) measurements (Jiang et al., 2008). The kinetic parameters are estimated by calibration but strong parameter correlation may hamper correct determination. Thus, the fitting does not convincingly demonstrate the validity of the model structure and parameter values. Other authors have also adopted the Lu model. ZarragoitiaGonza´lez et al. (2008), adopted the Lu ASM1 model (with the exception of nitrification processes) and linked it with a membrane fouling model where SMPs were used to estimate the bound EPS concentration in the sludge supernatant according to the equation SUAPþSBAP/0.8XTSS. Kinetic parameters of Lu’s model were partly adopted. This work did not include SMP modelling results. Di Bella’s (2008) work was also based on Lu’s model. The work was important since it focused on describing the fate of COD through the pilot MBR by distinguishing the COD removal contribution of (i) the physical membrane and (ii) the cake layer according to its depth. COD effluent was fairly well predicted. However, the model parameters were not in line with other experiments (Table 8). SMP kinetic parameters
were calibrated by fitting but the SMP concentrations were not shown. Sensitivity tests report that YSMP, and gUAP,A have a strong influence on the majority of output variables while bBAP,h and mSMP mainly affect NH4 and NO3. The sensitivity tests confirm that, once SMP are embedded in the ASM model, their intermediate role in decay and substrate metabolism is very relevant. Identification of the kinetic parameters introduced is thus crucial for the prediction of the effluent quality. Ahn et al. (2006) tried to include EPS in the ASM-SMP models, though their use in ASM works has generally been avoided. EPS were modelled in eq.(1) in Table 6, thus excluding an explicit EPS loss term, and BAP were produced uniquely by EPS hydrolysis. For the modelling process, 5 processes were added: hydrolysis of EPS, EPS formation by heterotrophs, EPS formation by autotrophs, UAP formation by heterotrophs, and UAP formation by autotrophs (Fig. 2). In order to describe the newly added processes, 8 new parameters were introduced (mSMP, kUAP, kUAPa, kBAP, KSMP, kEPS, kEPSa, and KXEPS), yielding an over parameterized model. In this model, the effect of SRT was not observed in SMP production but rather in EPS production, and the experimental results showed good agreement with the simulation results. EPS concentration was sensitive to Keps, Kbap and SRT, while Kuap sensitivity was low. However, kinetic parameters are not reported, SMP behaviour not described, and the model lacks an appropriate calibration (Jiang et al., 2008). Oliveira-Esquerre et al., (2006) introduced SMPs in ASM3. UAP and BAP were lumped into a general term MP. Overall, the model comprised 5 new SMP kinetic parameters (gMP,H, gMP,A, kMP, fb, Ymp) whose values were adopted from Lu et al. (2001), and 2 new processes. The work focused on understanding the peculiarities of an SMP model in ASM3 when compared to the ASM1. In the ASM1 model it is assumed that slowly biodegradable substrates are hydrolyzed before their use for growth. In the ASM3 model it is assumed that all organic substrates are directly converted into stored material and that stored compounds are subsequently used as a carbon and energy source for growth purposes. Consequently, as the specific rate of hydrolysis of MPs in ASM1 is considerably lower than the specific rate of storage in ASM3, it becomes a rate-limiting factor in the uptake of MPs. This explains why ASM3 gave a markedly low MP concentration (0.75 gCOD m3),
Fig. 2 e Concepts of the Lu et al. model (2001) (left) and the Ahn et al. model (2006) (right).
4289
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
Table 8 e SMP/EPS Kinetic parameters in recent literature works. EPS/SMP modelling Affinity constants KUAP
KBAP
[gCODSMP/l] Furumai and Rittmann, 1992 Laspidou and Rittmann, 2002a,b Jang et al., 2006 Aquino and Stuckey, 2008
Formation SMP
*** 0.5
100 100 500
85 85 500
UAP
Degradation BAP 1
**
[d ]
0.2 0.05 0.05 0.2
0.1* 0.17 0.02 0.0034
KBAP
KUAP
EPS fSMP
KEPS
KHYD [d1]
[mgCODproduct/mgCODcell/d] 1 0.07 0.07 0.03
1.27 0.4 1.2
0.18**** 0.18**** 0.03
0.17 0.02 0.03
ASM related works Affinity constants KSMP [gCODSMP/l] Lu et al., 2001 Lu et al., 2002 Lee et al., 2002 Oliveira-Esquerre et al., 2006 Zarragoitia-Gonza´lez et al., 2008 Di Bella et al., 2008 Jiang et al., 2008
30 60 30 30 30 133
Formation
Degradation rates
mSMP
KBAP
fSMP
fUAP
fBAP
KSMP
Kh,BAP
Kh,UAP
[d1]
[d1]
[]
[]
[]
[d1]
[d1]
[d1]
0.7 2.5 0.7
0.4 0.01
0.7 8.3
0.22 0.33
7.4E-7
0.0102
0.38 0.3 0.4 0.4
0.1 0.38 0.82 0.0963
0.0215
(*) in [mgCODproduct/mgCODcell/d]; (**) in [mgCODuap/mgCODsub]; (***) in [mgCODcell/mgCODsmp]; (****) in [mgCODeps/mgCODsub].
while a value of 80 gCOD m3 was obtained with ASM1. Both models would, however, give similar values of MP concentration if no storage was considered in ASM3 and MPs were directly used for bacterial growth, as assumed in ASM1. Jiang et al. (2008) introduced the SMP concept in an ASM2d model. This work differed considerably from the previous ASM-SMP models. BAP and UAP were degraded by hydrolysis steps, creating 3 new processes and imposing variations in 13 other processes. SMP kinetic parameters, production and hydrolysis rate of SMPs (subdivided by UAPs and BAPs) were investigated through specially designed experiments. The experiments were associated with an LC-OCD characterization, offering interesting outcomes. Jiang et al.’s (2008) UAP and BAP model peculiarities are in this review discussed in the stand-alone models section.
3.5.
Model identification e UAP/BAP kinetics
Two problems have been identified with SMP/EPS models. The kinetic parameters are not easily determinable experimentally and the models are usually over parameterized. Saroj et al., (2008) infer that incorporation of EPS/SMP in ASM would tend to worsen the model identification process, which is crucial in any ASM calibration exercise. As can be seen in Table 8, in order to overcome these problems several recent models made use of the Lu et al. model identification values, but this is not a good practise if SMP dynamics are related to specific processes or influent composition. The lack of validation campaigns confirming the theoretical models is striking. Few researchers have made efforts in the experimental determination of the new kinetic parameters. Jiang et al. (2008) calibrated their model once BAP and UAP kinetic parameters had been experimentally derived. BAP parameters were calculated by sCOD
monitoring of a batch system in famine conditions. UAP parameters were calculated by PN and PS concentration monitoring of a batch system in external substrate excess conditions. The degradation rate equations were chosen in order to minimize the number of unknown parameters. The determination of the unknown parameters was reported by the authors as complex, especially in the case of UAP, because BAP and acetate could interfere in the UAP measurements. The main criticism with regard to this batch approach could reside in: (i) The BAP batch test is run up to SCOD values of 200 mg COD/l. This value is very high when compared to WWTP supernatant COD; (ii) UAP tests are substrate specific; (iii) Both BAP and UAP tests are dilution dependent as shown by Laspidou and Rittmann; (iv) The washing of biomass prior experiments is arguable because the initial soluble COD of the supernatant would consequently be very low. But it is questionable whether the rate of SMP production/degradation are concentration dependent. This issue deserves further attention.
3.5.1.
Outlines
The use of an ASM expansion with the EPS/SMP concept is encouraged only if the following modelling objectives are to be pursued: (i) linking biology with fouling, (ii) soluble COD predictions (iii) Model high SRT processes. The ASM extension with EPS/SMPs concept creates difficulty in identifying the newly proposed parameters. Modellers have to implement strategies to reduce these parameters according to experiences reported in the literature and the process specificities. Parameter reduction strategies have been adopted by (i) coupling UAP/BAP into SMPs (OliveiraEsquerre et al., 2006); (ii) excluding EPS modelling (Lu et al., Oliveira-Esquerre et al., Lee et al. Jiang et al.); (iii) modelling the EPS (as sole AS foulants) with equations external to the
4290
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
ASM model (Saroj et al., 2008); (iv) excluding a water fraction UAP/BAP when negligible (Orhon et al., 1989). It would always be preferable to determine the kinetics parameters by specific batch tests, as suggested by Jiang et al., (2008), despite the limitation of this approach as discussed above.
4.
Outlook and future perspectives
In this final section, the authors would like to point out what they feel as the main shortcomings of the current state-ofthe-art of and future challenges in biological modelling of MBRs. The section is structured similarly to the previous sections.
4.1.
Influent characterization for the unmodified ASMs
Notwithstanding the fact that a large number of ASM applications have been reported in MBRs, some key factors still have to be further investigated and considered in wastewater characterization. One of these relates to the assessment of the active heterotrophic biomass Xhet in the influent wastewater that, although usually neglected in CAS modelling, needs to be better addressed when modelling membrane bioreactors. In fact, from a theoretical point of view, longer sludge ages lead to decreases in the percentage of active biomass in the MLVSS. Therefore, the higher the SRT, the less negligible the new biomass entering the plant via the influent becomes. This contribution is influenced by the characteristics of the sewage system (separate/combined), the presence and typology of pre-treatment units, the residence time in the sewage pipelines and the possible presence of toxic compounds due to industrial discharges. Possibly, a combination of well-known respirometric methods for Xhet determination (Kappeler and Gujer, 1992) and advanced biomass characterization tools such as flow-cytometry (Ziglio et al., 2002; Foladori et al., 2007) can be adopted at experimental level to actually understand the need for further calibration efforts. Another open issue in the application of ASMs to long-SRT membrane bioreactors is the fate of inorganic compounds in the influent (mineral suspended solids), whose concentration clearly depends on the upstream operation units (presence/ absence of sand removal and/or primary settling tank, fine screening meshes, run off, infiltration into the sewers). Some experimental evidence reports solubilization of inorganic solids entering the system: Laera et al. (2005) showed the loss of a significant amount of inorganic particulate matter in a bench-scale MBR operated with no sludge wasting, and they suggested that hydrolysis and solubilization produce molecules with a smaller size than the membrane pores, which leave the reactor with the effluent permeate. In the above mentioned work, Sperandio and Espinosa (2008) found that up to 50% of influent mineral solids were converted into soluble inorganics at a sludge age of 110 days. Therefore, in the specific case of ASM3 application to MBRs, this would mean at very long SRT an adjustment of the absolute values of composition parameters for MLSS (iTS_XI, iTS_XS, iTS_XBM, iTS_XSTO). An alternative option could be given by the introduction of the model proposed by Ekama and Wentzel (2004) for
Inorganic Suspended Solids (ISS) in the ASM framework. This model assumes ISS in the sludge to be due to the accumulation of influent mineral solids and the uptake of inorganic dissolved solids (IDS) by ordinary heterotrophic organism (OHO) and phosphate accumulation organisms (PAO). In this case, a suitable calibration of parameters ”fOHO” and “fPAO” (inorganic solids content for OHO and PAO respectively) could “mask” the gradual solubilization of influent inorganics at long sludge ages. In addition, a solubilization mechanism should be mathematically introduced in the ASM matrix.
4.2.
Process kinetics for the unmodified ASMs
Concerning the biokinetic parameters of unmodified ASM1, most of the biological processes have been widely investigated. Little, on the other hand, is known about the possible specificities of biokinetics for phosphorus removal in MBR. The available literature does not show significant differences between MBR and CAS calibration parameters. However, recent microbial studies have shown the presence of PAOs in several full-scale MBR not specifically designed for biological phosphorous removal. Considering that a volume percentage of the anoxic reactor is anaerobic is not an applicable solution in all cases (since nitrate supernatant concentration may not allow it) and predicting phosphate effluent concentrations in these conditions remains rather problematic. This aspect needs further investigation.
4.3.
Process kinetics for the modified ASMs
EPS and SMP concentrations are sensitive to biomass production, hydrolysis and degradation rate. Few detailed experimental data are available for calibrating the rates of each process independently. Models including the EPS/SMP concept show that EPS constitute a large amount of the organic reserve in a bioreactor and their hydrolysis can impact the SMP release in the sludge water very significantly. Slow hydrolysis of microbial products (in aerobic, anoxic or anaerobic conditions) is a poorly understood process which definitely needs more attention. The SMP equilibrium concentration results from production/degradation mechanisms but also depends on SMP retention by the membrane. The latter aspect is not sufficiently considered in the works reviewed and authors have mainly measured or fitted it as a steady value, independent of the process variations. The percentage of SMP permeating through the membrane ranges from 0 to 100% (Jang et al., 2006; Jiang et al., 2008; Zarragoitia-Gonza´lez et al., 2008; Lu et al., 2001, 2002). Quantification of this transport depending on membrane type, process conditions and fouling characteristics needs further investigation. The subdivision among UAP and BAPs has been widely acknowledged and modelling efforts for their production and consumption are under way. However, recent observations tend to show how the SMP/bound EPS ratio (i.e. the state of flocculation of the biomass) also depends on process disturbances such as shock loads, shear stress, temperature, pH and oxygen shocks (inter alia Drews et al., 2007). An important question is whether the current models are valid for inclusion of these process disturbance effects. Most probably they are
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
not and model extensions are required in order to predict SMP dynamics with regard to process operation and process disturbances. Finally, as sufficiently noted in the text, the determination of kinetic parameters of the SMP/EPS models is a major concern. Over-parametrization or lack of identifiability appears as a typical problem of ASM model extensions. Is it possible to mathematically overcome the problem or do we need to systematically characterize each process experimentally?
4.4. Application of ASM models at full-scale, hydraulics and aeration Most modelling exercises and conclusions are based on labscale or pilot-scale experiments. However, in order to take the step to usage and full exploitation of the benefits of biokinetic models for plant design and operation, more full-scale studies are required. More attention should be paid to full-scale model applications in order to confirm the applicability of the current findings. This problem is not new, however, as it is a known issue in CAS modelling as well. In principle, the models described in chapter 2 (unmodified ASM) should be suitable for full-scale use, bearing in mind some points requiring attention as stipulated. Full-scale plant aeration as well as hydraulic characterization are key issues in modelling MBR. However, they have not received much attention to date (especially hydraulics). The contribution of coarse bubble aeration (membrane scouring) to the overall oxygen supply for oxic processes needs to be better characterized. In the case of sMBR systems, there is a need to couple valid kinetic ASM models with more accurate hydraulic models (e.g. CFD). Intensive hydraulic characterization of aerobic/anoxic basins is still insufficient. When membranes are directly submerged in the bioreactor, it becomes difficult to correctly predict the influence of membrane aeration on the overall performance. Fine bubble aeration (for biological needs) is generally regulated by on/off control, for minimising energy consumption and promoting denitrification through increased anoxic reactor volume. But membrane coarse bubbles aeration rate is more or less constant and contributes to the oxygen supply, influencing nitrification in a complex hydraulic system with heterogeneous oxygen concentration. A better balance between the different sub-models (biokinetic model, aeration model, hydraulic model) should, therefore, be investigated.
5.
Conclusions
A concise overview of the most recent literature on ASMbased MBR modelling is given in this review work. Clarity was sought by categorizing models as unmodified or modified ASMs, underlining the strong relationship with the modeller’s target: modelling to improve process performance or modelling for further understanding of the process. The paper has aimed to extract relevant information that seems commonly agreed on within the scientific community in this particular area, and also to highlight contradictory observations and conclusions present in scientific papers. The
4291
summarizing paragraphs at the end of each section are intended to guide end-users in the current availability and applicability of specific models and techniques, their limitations and possible pitfalls. This should assist in modelling municipal laboratory, pilot and full-scale MBRs, either for performance optimization or for decision support. The experience reported in this review proves that, when modelling purposes do not differ from effluent characterization, oxygen demand and sludge production, ASMs are very relevant to MBR applications. However, particular care needs to be taken since the specific conditions present in MBRs are reflected in some important discrepancies when compared to ASM default parameter values. In terms of biokinetic studies of the different processes, the authors have preferred not to propose a general set of values, since, as research results show, the values are very dependent on the operating conditions. Instead, researchers’ work has been brought together in concise tables that can, indicate common directions for each process. In cases of ASMs developed for specific purposes related to MBR operations, and with particular care to fouling prediction modelling, the knowledge on SMP and EPS modelling has been critically reviewed on the basis of the literature, including the most recent developments. Progress in tackling the overparametrization of the extended models has been highlighted, as this paper intends to serve as an incentive to bring scientific efforts into practice. Finally, the outlook and future perspectives have been systematically highlighted and proposed for each section. The paper provides a guide for different end-users of mathematical models for MBR, be they people active in process design and operations, or academics who want to learn about the current state-of-the-art or detect current shortcomings in MBR modelling.
Acknowledgements This review paper is an outcome of the “Liaison Group on MBR Modelling” that was established within the coalition of European projects, MBR-Network (http://www.mbr-network.eu/). The authors would like to thank the European Commission for its financial support through the EUROMBRA project (Contract No. 018480, FP 6thdGlobal Change and Ecosystems), the AMEDEUS project (Contract No. 018328, FP 6thdGlobal Change and Ecosystems), and the MBR-TRAIN project (Marie Curie Host Fellowship for Early Stage Research Training supported by the European Commission under the 6th Framework Programme with Contract No. MEST-CT-2005-021050).
references
Ahn, Y.T., Choi, Y.K., Jeong, H.S., Shin, S.R., 2006. Modeling of extracellular polymeric substances and soluble microbial products production in a submerged MBR at various SRTs. Water Sci. Technol. 53 (7), 209e216. Al-Halbouni, D., Traber, J., Lyko, S., Wintgens, T., Melin, T., Tacke, D., Janot, A., Dott, W., Hollender, J., 2008. Correlation of EPS content in activated sludge at different sludge retention
4292
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
times with membrane fouling phenomena. Water Res. 42 (6e7), 1475e1488. Aquino, S.F., Stuckey, D.C., 2008. Integrated model of the production of soluble microbial products (SMP) and extracellular polymeric substances (EPS) in anaerobic chemostats during transient conditions. Biochem. Eng. J. 38, 138e146. Artan, N., Orhon, D., Baykal, B.B., 1990. Implications of the task group model I. the effect of the initial substrate concentration. Water Res. 24 (10), 1259e1268. Barker, D.J., Stuckey, D.C., 1999. A review of soluble microbial products (SMP) in wastewater treatment. Water Res. 33 (14), 3062e3082. Benedetti, L., Batstone, D.J., Debaets, B., Nopens, I., Vanrolleghem, P., 2008. Global Sensitivity Analysis of the Benchmark Simulation Model No. 2. IEMSs Fourth Biennial Meeting: International Congress on Environmental Modelling and Software. Spain, Barcelona. 7e10 July. Bixio, D., De Wilde, W., Lesjean, B., 2006. Model-based Evaluation of Flow Repartition Control Strategies for the CAS/MBR Dual 1 Concept, Including Determination of the Sludge Characteristics Amedeus Project, D52 (task 91). Chaize, S., Huyard, A., 1991. Membrane bioreactors on domestic wastewater treatment sludge production and modeling approach. Water Sci. Technol. 23, 1591e1600. De la Torre T., Iversen V., Meng F., Stu¨ber J., Drews A., Lesjean B., Kraume M, 2009. Searching for a universal fouling indicator for MBR. 5th IWA Specialised Membrane Technology Conference for Water & Wastewater Treatment, 1e3 September 2009. Beijing, China. Delrue F., Racault Y., Choubert J.M., Spe`randio M., (2008), Modelling a full scale Membrane Bioreactor using Activated Sludge Model n 1: challenges and solutions. IWA Regional Conference “Membrane Technologies in Water and Waste Water Treatment” (2e4 June 2008 Moscow, Russia). Di Bella, G., Mannina, G., Viviani, G., 2008. An integrated model for physical-biological wastewater organic removal in a sMBR: model development and parameter estimation. J. Membr. Sci. 322, 1e12. Dold, P.L., Ekama, G.A., Marais, G.v.R., 1980. A general model for the activated sludge process. Prog. Water Technol. 12, 47e77. Drews, A., Evenblij, H., Rosenberger, S., 2005. Potential and drawbacks of microbiology-membrane interaction in membrane bioreactors. Environ. Prog. 24 (4), 426e433. Drews, A., Vocks, M., Iversen, V., Lesjean, B., Kraume, M., 2006. Influence of unsteady membrane bioreactor operation on EPS formation and filtration resistance. Desalination 192 (1e3), 1e9. Drews, A., Mante, J., Iversen, V., Vocks, M., Lesjean, B., Kraume, M. , 2007. Impact of ambient conditions on SMP elimination and rejection in MBR. Water Res. 41 (17), 3850e3858. Erftverband, 2001. Weitergehende Optimierung einer Belebungsanlage mit Membranfiltration e Zwischenbericht. (Advanced optimisation of an activated sludge plant with membrane filtration e mid-term-report), Report to the Ministry of Environment North e Rhine Westpahlia, Germany, Erftverband Bergheim, pp. 73e93. Erftverband, 2004. Optimisation of an Activated Sludge Plant with Membrane Filtration e, Report to the Ministry of Environment North e Rhine Westpahlia, Germany, Erftverband Bergheim, pp. 2e16. Optimierung einer Belebungsanlage mit Membranfiltration e Band 3, vol. 3. Ekama, G.A., Wentzel, M.C., 2004. A predictive model for the reactor inorganic suspended solids concentration in activated sludge systems. Water Res. 38 (19), 4093e4106. Ekama, G.A., Marais, G.R., 1986. Procedures for determining influent COD fractions and the maximum specific growth rate of heterotrophs in activated sludge systems. Water Sci. Technol. 18 (6), 63e89.
Fenu A., Roels J., Wambecq T., De Gussem K., Thoeye C., De Gueldre G., Van De Steene B. Energy audit of a full scale MBR system. Desalination, in press. Ficara, E., Rocco, A., Rozzi, A., 2000. Determination of nitrification kinetics by the ANITA-DOstat biosensor. Water Sci. Technol. 41 (12), 121e128. Foladori, P., Bruni, L., Andreottola, G., Ziglio, G., 2007. Effects of sonication on bacteria viability in wastewater treatment plants evaluated by flow-cytometry: faecal indicators, wastewater and activated sludge. Water Res. 41 (1), 235e243. Furumai, H., Rittmann, B.E., 1992. Advanced modelling of mixed population of heterotrophs and nitrifiers. Water Sci. Technol. 26 (3e4), 493e502. Germain, E., Nelles, F., Drews, A., Pearce, P., Kraume, M., Reid, E., Judd, S.J., Stephenson, T., 2007. Biomass effects on oxygen transfer in membrane bioreactors. Water Res. 41 (5), 1038e1044. Germirli, F., Orhon, D., Artan, N., 1991. Assessment of the initial inert soluble COD in industrial wastewater. Water Sci. Technol. 23, 1077e1086. Guglielmi G., Avesani D., Brepols C., Foxon K., Brouckaert C., Buckley C., 2009. Sludge production in aerobic and anaerobic membrane bioreactors, Proc. of WISA Membrane technology Conference. 13e15 May, Stellenbosch, South Africa. Gujer, W., Henze, M., Mino, T., Loosdrecht, M., 1999. Activated sludge model No.3. Water Sci. Technol. 39 (1), 183e193. Henze, M., Grady, C.P.L., Gujer, W., Marais, G.v.R., Matsuo, T., 1987. Activated Sludge Model No.1 IAWPRC Scientific and Technical Report No.1, London (GB). Henze, M., 1992. Characterization of wastewater for modelling of activated sludge systems. Water Sci. Technol. 18 (6), 91e114. Henze, M., Gujer, W., Mino, T., van Loosdrecht, M., 2000. Activated Sludge Models ASM1, ASM2, ASM2d and ASM3 IWA Scientific and Technical Report No.9, London (GB). Itokawa, H., Thiemig, C., Pinnekamp, J., 2008. Design and operating experiences of municipal MBRs in Europe. Water Sci. Technol. 58 (12), 2319e2327. Jang, N., Ren, X., Cho, J., Kim, I.S., 2006. Steady-state modeling of bio-fouling potentials with respect to the biological kinetics in the sMBR. J. Memb. Sci. 284, 352e360. Jarusutthirak, C., Amy, G., 2007. Understanding soluble microbial products (SMP) as a component of effluent organic matter (EfOM). Water Res. 41 (12), 2787e2793. Jiang, T., Liu, X., Kennedy, M.D., Schippers, J.C., Vanrolleghem, P. A., 2005. Calibrating a side-stream membrane bioreactor using Activated Sludge Model No. 1. Water Sci. Technol. 52 (10e11), 359e367. Jiang T., 2007. Characterization and modelling of soluble microbial products in membrane bioreactors. PhD thesis, Ghent University, Belgium pp. 241. Jiang, T., Myngheer, S., De Pauw, D.J.W., Spanjers, H., Nopens, I., Kennedy, M.D., Amy, G., Vanrolleghem, P.A., 2008. Modelling the production and degradation of soluble microbial products (SMP) in membrane bioreactors (MBR). Water Res. 42 (20), 4955e4964. Jiang, T., Sin, G., Spanjers, H., Nopens, I., Kennedy, M., van der Meer, W., Futselaar, H., Amy, G., Vanrolleghem, P., 2009. Comparison of the modeling approach between membrane bioreactor and conventional activated sludge processes. Water Environ. Res. 81 (4), 432e440. Jimenez J., Grelier P., Meinhold J., Tazi-Pain A., 2008. Biological modelling of MBR and impact of primary sedimentation. MIDW-EDS conference, Toulouse, 20e22 October 2008. Judd, S., 2006. The MBR Book: Principles and Applications of Membrane Bioreactors for Water and Wastewater Treatment. EN, July 2006. Elsevier, ISBN 978-1-85617-481-7. Kappeler, J., Gujer, W., 1992. Estimation of kinetic parameters of heterotrophic biomass under aerobic conditions and
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
characterization of wastewater for activated sludge modelling. Water Sci. Technol. 25 (6), 125e139. Krishna, C., Van Loosdrecht, M.C.M., 1999. Substrate flux into storage and growth in relation to activated sludge modelling. Water Res. 33 (14), 3149e3161. Krause, S., 2005. Untersuchungen zum Energiebedarf von Membranbelebungsanlagen. Schriftenreihe WAR 166, Darmstadt, Techn. Universita¨t, Diss., 2005. In: Research into the Energy Consumption of Membrane Bioreactor Plants, ISBN 3-932518-62-4. Krause, S., Cornel, P., Wagner, M., 2003. Comparison of different oxygen transfer testing procedures in full scale membrane bioreactors. Water Sci. Technol. 47 (12), 169e176. Krause S., Cornel P., 2007. Membrane Bioreactor’s Energy Demand in Wastewater Treatment. 4th IWA International Membrane Conference, 15the17th May 2007, Harrogate, UK Laera, G., Pollice, A., Saturno, D., Giordano, C., Lopez, A., 2005. Zero net growth in a membrane bioreactor with complete sludge retention. Water Res. 39 (20) 5241e4249. Laspidou, C.S., Rittmann, B.E., 2002a. Non-steady state modeling of EPS, SMPs, and active and inert biomass. Water Res. 36 (8), 1983e1992. Laspidou, C.S., Rittmann, B.E., 2002b. A unified theory for EPS, SMPs, and active and inert biomass. Water Res. 36 (11), 2711e2720. Le-Clech, P., Chen, V., Fane, T.A.G., 2006. Fouling in membrane bioreactors used in wastewater treatment. J. Membr. Sci. 284 (1e2), 17e53. Lee, Y., Cho, J., Seo, Y., Lee, J.W., Ahn K.-, H., 2002. Modelling of submerged membrane bioreactor process for wastewater treatment. Desalination 146, 451e457. Lesouef, A., Payraudeau, M., Rogalla, F., Kleiber, B., 1992. Optimizing nitrogen removal reactor configurations by onsite calibration of the IAWPRC Activated sludge model. Water Sci. Technol. 25 (6), 105e123. Liang, S., Liu, C., Song, L., 2007. Soluble microbial products in membrane bioreactor operation: behaviors, characteristics, and fouling potential. Water Res. 41 (1), 95e101. Liu, H., Fang, H.H.P., 2002. Extraction of extracellular polymeric substances (EPS) of sludges. J. Biotechnol. 95, 249e256. Lu, S.G., Imai, T., Ukita, M., Sekine, M., Higuchi, T., Fukagawa, M., 2001. A model for membrane bioreactor process based on the concept of formation and degradation of soluble microbial products. Water Res. 35 (8), 2038e2048. Lu, S.G., Imai, T., Ukita, M., Sekine, M., Higuchi, T., 2002. Modeling prediction of membrane bioreactor process with the concept of soluble microbial product. Water Sci. Technol. 46 (11e12), 63e70. Luedeking, R., Piret, E.C., 1959. A kinetic study of lactic acid fermentation batch process at controlled pH. J. Biochem. Microbiol. Technol. Eng. 1 (4), 393e412. Luong, J.H., Muchaldani, A., 1988. Kinetic of bio polymers synthesis: a revisit. Enzym. Microb. Technol. 10, 326e329. McKinney, R.E., 1962. Mathematics of complete mixing activated sludge. J. Sanitary Eng. ASCE 88 (3), 87e113. Majone, M., Dircks, K., Beun, J.J., 1999. Aerobic storage under dynamic conditions in activated sludge processes. The state of the art. Water Sci. Technol. 39 (1), 61e73. Malamis, S., Andreadakis, A., 2009. Fractionation of proteins and carbohydrates of EPS in a MBR. Bioresour. Technol. 100 (13), 3350e3357. Manser, R., Gujer, W., Siegrist, H., 2005. Consequences of mass transfer effects on the kinetics of nitrifiers. Water Res. 39 (19), 4633e4642. Manser, R., Gujer, W., Siegrist, H., 2006. Decay processes of nitrifying bacteria in biological wastewater treatment systems. Water Res. 40 (12), 2416e2426. Marais, G.V.R., Ekama, G.A., 1976. The activated sludge process. Part 1 e steady-state behaviour. Water SA 2, 163e199.
4293
Masse´ A., 2004. Biore´acteurs a` membranes immerge´es pour le traitement des eaux re´siduaires urbaines: spe´cificite´s physicochimiques du milieu biologique et colmatage. Ph.D., INSA Toulouse, no. 759, 15, de´cembre 2004. Masse´, A., Spe´randio, M., Cabassud, C., 2006. Comparison of sludge characteristics and performance of a submerged membrane bioreactor and an activated sludge process at high solids retention time. Water Res. 40 (12), 2405e2415. Metcalf, Eddy, 2003. Wastewater Engineering: Treatment and Reuse, fourth ed.. Mc Graw Hill. Munz, G., De Angelis, D., Goria, R., Moric, G., Casarci, M., Lubello, C., 2008. Process efficiency and microbial monitoring in MBR and conventional activated sludge process treatment of tannery wastewater. Bioresour. Technol. 99 (18), 8559e8564. Namkung, E., Rittmann, B.E., 1986. Soluble microbial products (SMP) formation kinetics by biofilms. Water Res. 20 (6), 795e806. Ng, H.Y., Hermanowicz, S.W., 2005. Specific resistance to filtration of biomass from membrane bioreactor reactor and activated sludge: effects of exo-cellular polymeric substances and dispersed micro-organisms. Water Environ. Res. 77 (2), 187e192. Ng, H.Y., Tan, T.W., Ong, S.L., 2006. Membrane fouling of submerged membrane bioreactors: impact of the mean cell residence time and the contributing factors. Environ. Sci. Tecnol. 40, 2706e2713. Ng, N.L.A., Kim, A.S., 2007. A mini-review of modelling studies on MBR treatment for municipal wastewaters. Desalination 212 (1e3), 261e281. Noguera, D.R., Araki, N., Rittmann, B.E., 1994. Soluble microbial products (SMP) in anaerobic chemostats. Biotechnol. Bioeng. 44, 1040e1047. Oliveira-Esquerre, K.P., Narita, H., Yamato, N., Funamizu, N., Watanabe, Y., 2005. Modelling of a Conventional MBR for Wastewater Treatment. 21st Center of Excellence Program Publication. Department of Urban and Environmental Engineering, Hokkaido University, Japan. Oliveira-Esquerre, K.P., Narita, H., Yamato, N., Funamizu, N., Watanabe, Y., 2006. Incorporation of the concept of microbial products formation into ASM3 and the modelling of a membrane bioreactor for a wastewater treatment. Braz. J. Chem. Eng. 23 (4), 461e471. Orhon, D., Artan, N., Cimcit, Y., 1989. The concept of soluble residual product formation in the modelling of activated sludge. Water Sci. Technol. 21, 339e350. ¨ , So¨zen, S., 1999. The effect of residual Orhon, D., Karahan, O microbial products on the assessment of the particulate inert COD in wastewaters. Water Res. 33, 3191e3203. Parco, V., Wentzel, M., Ekama, G., 2006. Kinetics of nitrogen removal in a MBR nutrient removal activated sludge system. Desalination 199, 89e91. Parco, V., du Toit, G., Wentzel, M., Ekama, G., 2007. Biological nutrient removal in membrane bioreactors: denitrification and phosphorus removal kinetics. Water Sci. Technol. 56 (6), 125e134. Petersen, B., Gernaey, K., Henze, M., Vanrolleghem, P., 2002. Evaluation of an ASM1 procedure on a municipal-industrial wastewater treatment plant. J. Hydroinf. 04 (1), 38. Pinnekamp, J., 2006. Municipal Water and Waste Management: Membrane Technology for Waste Water Treatment. In: Friedrich, H. (Ed.). FIW Verlag, Aachen, ISBN 3-939377-01-5. Pollice, A., Laera, G., Saturno, D., Giordano, C., 2008. Effects of sludge retention time on the performance of a membrane bioreactor treating municipal sewage. J. Membr. Sci. 317, 65e70. Ramesh, A., Duu-Jong, L., Hong, S.G., 2006. Soluble microbial products (SMP) and soluble extracellular polymeric substances (EPS) from wastewater sludge. Environ. Biotechnol. 73, 219e225. Ramphao, M.C., Wentzel, M.C., Ekama, G.A., Alexander, W.V., 2006. A comparison of BNR activated sludge systems with
4294
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 7 2 e4 2 9 4
membrane and settling tank solid-liquid separation. Water Sci. Technol. 53 (12), 295e303. Rieger, L., Koch, G., Kuehni, M., Gujer, W., Siegrist, H., 2001. The EAWAG bio-P module for activated sludge model no.3. Water Res. 35 (16), 3887e3903. Robinson, J.A., Trulear, M.G., Characklis, W.G., 1984. Cellular reproduction and extracellular polymer formation by Pseudomonas aeruginosa in continuous culture. Biotechnol. Bioeng. 26, 1409e1417. Roeleveld, P.J., van Loosdrecht, M.C.M., 2002. Experience with guidelines for wastewater characterization in The Netherlands. Water Sci. Technol. 45 (6), 77e87. Rosenberger, S., Witzig, R., Manz, W., Szewzyk, U., Kraume, M., 2000. Operation of different membrane bioreactors: experimental results and physiological state of the microorganisms. Water Sci. Technol. 41 (10e11), 269e277. Rosenberger, S., Kraume, M., 2002. Filterability of activated sludge in membrane bioreactors. Desalination 146, 373e379. Rosenberger S., (2003), Charakterisierung von belebtem Schlamm in Membranbelebungsanlagen. Dissertation, Forschrittsberichte VDI-Verlag, Reihe 3, Nr. 769 ISBN 3-18376903-4, 160 S. Rosenberger, S., Evenblij, H., Poele te, S., Wintgens, T., Laabs, C., 2005. The importance of liquid phase analyses to understand fouling in membrane assisted activated sludge processesdsix case studies of different European research groups. J. Membr. Sci. 263 (1e2), 113e126. Rosenberger, S., Laabs, C., Lesjean, B., Gnirss, R., Amy, G., Jekel, M. , Schrotter, J.-C., 2006. Impact of colloidal and soluble organic material on membrane performance in membrane bioreactors for municipal wastewater treatment. Water Res. 40 (4), 710e720. Sarioglu, M., Insel, G., Artan, N., Orhon, D., 2008. Modelling of long-term simultaneous nitrification and denitrification (SNDN) performance of a pilot scale membrane bioreactor. Water Sci. Technol. 57 (11), 1825e1833. Sarioglu, M., Insel, G., Artan, N., Orhon, D., 2009. Model evaluation of simultaneous nitrification and denitrification in a membrane bioreactor operated without an anoxic reactor. J. Membr. Sci. 337, 17e27. Saroj, D.P., Guglielmi, G., Chiarani, D., Andreottola, G., 2008. Modeling and simulation of membrane bioreactors by incorporating simultaneous storage and growth concept: a special attention to fouling while modelling the biological process. Desalination 221, 475e482. Shin, H.S., Kang, S.T., 2003. Characteristics and fates of soluble microbial products in ceramic membrane bioreactor at various sludge retention times. Water Res. 37 (1), 121e127. Silva, D.G.V., de, Urbain, V., Abeysinghe, D.H., Rittmann, B.E., 1998. Advanced analysis of membrane-bioreactor performance with aerobic-anoxic cycling. Water Sci. Technol. 38 (4e5), 505e512. Silva A.F., Carvalho G., Lousada Ferreira M., Nieuwenhuijzen van A., Guglielmi G., Crespo J.G., Reis M.A.M., Crespo M.T.B., 2009. Microbial population structure of pilot and full scale membrane bioreactors. Book of Proceedings of final MBR Network workshop, 31 Marche1 April, Berlin (Germany). Sin, G., Guisasola, A., De Pauw, D.J.W., Baeza, J.A., Carrera, J., Vanrolleghem, P.A., 2005. A new approach for modelling simultaneous storage and growth processes for activated sludge systems under aerobic conditions. Biotechnol. Bioeng. 92 (5), 600e613. Spe`randio, M., Masse´, A., Espinosa, M.C., Cabassud, C., 2005. Characterization of sludge structure and activity in submerged membrane bioreactor. Water Sci. Technol. 52 (10e11), 401e408.
Sperandio, M., Espinosa, M.C., 2008. Modelling an aerobic submerged membrane bioreactor with ASM models on a large range of sludge retention time. Desalination 231 (1e3), 82e90. Spe´randio, M., Paul, E., 2000. Estimation of wastewater biodegradable COD fractions by combining respirometric experiments in various S0/X0 ratios. Water Res. 34 (4), 1233e1246. STOWA, 1996. Methods for Wastewater Characterization, Inventory and Guidelines. STOWA-report 97-23, Hageman Fulfilment, Zwijndrecht, The Netherlands. (in Dutch). STOWA, 1999. Wastewater Characterization of Raw and Pretreated Wastewater, the Influence of Primary Sedimentation and Pre-Precipitation. STOWA-report 99-13, Hageman Fulfilment, Zwijndrecht, The Netherlands (in Dutch). Tan, T.W., Ng Yong, H., Leong Ong, S., 2008. Effect of mean cell residence time on the performance and microbial activity of pre-denitrification submerged membrane bioreactors. Chemosphere 70, 387e396. Van Haandel, A.C., Van Der Lubbev, J.G.M., 2007. Handbook Biological Waste Water Treatment e Design and Optimization of Activated Sludge Systems. Quist Publishing, Leidschendam, The Netherlands, ISBN 978-90-77983-22-5. EN. Vanrolleghem, P.A., Spanjers, H., Petersen, B., Ginestet, P., Takacs, I., 1999. Estimating (combinations of) activated sludge model n 1 parameters and components by respirometry. Water Sci. Technol. 39 (1), 195e214. Vanrolleghem, P.A., Insel, G., Petersen, B., Sin, G., De Pauw, D., Nopens, I., Dovermann, H., Weijers, S., Gernaey, K., 2003. A comprehensive model calibration procedure for activated sludge models. Proc.. of WEFTEC 03, October 11-15 2003, Los Angeles, CA, USA. Van Loosdrecht, M.C.M., Heijnen, J.J., 2002. Modelling of activated sludge processes with structured biomass. Water Sci. Technol. 45 (6), 13e23. Wagner, J., Rosenwinkel, K.H., 2000. Sludge production in membrane bioreactors under different conditions. Water Sci. Technol. 41 (10e11), 251e258. Wintgens, T., Rosen, J., Melin, T., Brepols, C., Drensla, K., Engelhardt, N., 2003. Modelling of a membrane bioreactor system for municipal wastewater treatment. J. Membr. Sci. 216 (1e2), 55e65. Wisniewski C., (1996), Etude du comportement de cultures mixtes en bioreacteur a` membrane. Cine´tiques re´actionnelles et filtrabilite´. PhD-Thesis. Univeristy Montpellier II. Witzig, R., Manz, W., Rosenberger, S., Kru¨ger, U., Kraume, M., Szewzyk, U., 2002. Microbiological aspects of a bioreactor with submerged membranes for aerobic treatment of municipal wastewater. Water Res. 36 (2), 394e402. Wu, Z., Wang, Z., Zhou, Z., Yu, G., Gu, G., 2007. Sludge rheological and physiological characteristics in a pilot-scale submerged membrane bioreactor. Desalination 212, (1e3), 152e164. Zarragoitia-Gonza´lez, A., Schetrite, S., Alliet, M., Ja´uregui-Haza, U. , Albasi, C., 2008. Modeling of SMBRs: conceptual study about link between activated sludge biokinetics, aeration and fouling process. J. Membr. Sci. 325, 612e625. Zhang, J., Chua, H.C., Zhou, J., Fane, A.G., 2006. Factors affecting the membrane performance in submerged membrane bioreactors. J. Membr. Sci. 284, (1e2), 54e66. Zhang, H.F., Sun, B.S., Zhao, X.H., Qi, G.S., 2007. Generation of soluble microbial products in membrane bioreactor. China Environ. Sci. 27 (4), 539e542. Ziglio, G., Andreottola, G., Foladori, P., Ragazzi, M., 2001. Experimental validation of a single-OUR method for wastewater RBCOD characterisation. Water Sci. Technol. 43 (11), 119e126. Ziglio, G., Andreottola, G., Barbesti, S., Boschetti, G., Bruni, L., Foladori, P., Villa, R., 2002. Assessment of activated sludge viability with flow cytometry. Water Res. 36 (2), 460e468.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 9 e4 4 1 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
A biofiltration model for tertiary nitrification of municipal wastewaters Emmanuelle Vigne a,b, Jean-Marc Choubert b, Jean-Pierre Canler b, Alain He´duit c, Kim Sorensen d, Paul Lessard a,* a
De´partement de ge´nie civil et ge´nie des eaux, Pavillon Adrien-Pouliot, 1065 avenue de la Me´decine, Universite´ Laval, Que´bec (Qc), Canada, G1V 0A6 b Cemagref, UR MALY, F-69336 Lyon cedex 9, France c Cemagref, UR HBAN, F-92163 Antony cedex, France d VE´OLIA, Immeuble Giovanni Battista B, 1, rue Giovanni Battista Pirelli, 94410 Saint-Maurice, France
article info
abstract
Article history:
The main objective of this work concerns the evaluation of the biological aerated filtration
Received 5 February 2010
model found in GPS-X, which had never been evaluated with adequate data. This model is
Received in revised form
interesting since it integrates the physical and biological phenomena involved during
20 May 2010
filtration with a low complexity of use. The validation of the model parameters combines
Accepted 2 June 2010
experimental and theoretical approaches. Experimental data were recorded at a semi-
Available online 9 June 2010
industrial pilot scale submerged biofilter operated at a tertiary nitrification stage, receiving the effluent of a medium loaded activated sludge process for municipal wastewater. Also,
Keywords:
several protocols were regularly applied to characterize the biofilm and the nitrogen
Dynamic modelling
removal performances: dry density and thickness of biofilm, nitrification rates and corre-
Parameter estimation
sponding quantity of autotrophic biomass accumulated inside the filtering media, quantity
Submerged biofiltration
of extracted autotrophic bacteria in the backwash water, nitrification capacity along the
Tertiary nitrification
biofilter, as well as nitrogen compounds in the effluent. For short-term dynamic conditions, a set of reliable parameter values has been used to predict nitrogen removal for different data sets. For long-term dynamic periods, the need to adapt some of the parameters from one set of data to another is demonstrated. It is shown that the hydraulic loading rate and the backwashing frequency are the main parameters responsible for these modifications. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Since the elaboration of the European Directive 91/271/EEC in the early nineties, a concentration lower than 10 g N m3 total nitrogen is a common discharge objective for municipal wastewater treatment plants. To upgrade medium and highly loaded activated sludge plants built in areas with strong land
pressures, practitioners can choose the submerged aerated biofiltration process as a solution to meet low concentration of ammonium in treated water. This technology can nitrify high loads of ammonium in a limited space due to the development of an important amount of autotrophic biomass on the filtering media (Payraudeau et al., 2000; Canler et al., 2003). Several studies dealing with tertiary nitrification biofiltration
* Corresponding author. E-mail addresses:
[email protected] (E. Vigne),
[email protected] (J.-M. Choubert), jean-pierre.canler@ cemagref.fr (J.-P. Canler),
[email protected] (A. He´duit),
[email protected] (K. Sorensen),
[email protected] (P. Lessard). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.005
4400
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 9 e4 4 1 0
are reported in literature. They were carried out on pilot-scale plants (Tschui et al., 1994; Han et al., 2001; Jeong et al., 2006) as well as at full-scale industrial plants (Payraudeau et al., 2000; Canler et al., 2003; Rocher et al., 2007). In order to design and optimize the operation of wastewater treatment processes, mathematical models are useful tools for engineers and researchers. For the activated sludge process, a simple and robust model was proposed two decades ago, ASM1 (Henze et al., 1987). The relative simplicity of this model led to a widespread use by researchers and practitioners. Biofilm models are more complex and difficult to use. They have, thus, found little application in engineering practice (Plattes et al., 2006). None the less there seems to be an emerging interest in using biofilm models in practice, e.g. the German Association for Water, Wastewater and Waste has had a working group running on biofilm modelling since December 2005 (Horn et al., 2008). For submerged biofilter systems, few studies on modelling have been reported in the literature. Some numerical biofilter models were found in the literature (Lessard et al., 2008) and these were mainly one-dimensional. They were set up and verified by means of experiments carried out at different scales: lab-scale (Mann and Stephenson, 1997; Hidaka and Tsuno, 2004), semi-industrial scale (Sanz et al., 1996; Farabegoli et al., 2004), and even at industrial full-scale (Viotti et al., 2002). The main differences consist in the mechanisms that are simulated. Most of them include a transport model along the filter, a conversion model (biological reactions) in the bulk liquid and the biofilm. Few of them include a filtration or a head loss prediction model. The Biological Aerated Filter (BAF) model found in GPS-X, a simulation software (Hydromantis, 2003), is a comprehensive model for submerged up-flow biological filters. Its content consists of five components (Vigne et al., 2007; Vigne, 2007): 1) the hydraulic representation of the filtering media, 2) the transfer and the transport of soluble and particulate substrates inside the biofilm (Spengel and Dzombak, 1992), 3) the growth and the decay of heterotrophic and autotrophic biomass due to biological conversion of substrates (e.g. Henze et al., 1987), 4) the head loss evolution and the TSS filtration and attachment (Horner et al., 1986) and 5) the backwash mechanism. These five modules are linked by mass transfer equations such that the model can be used to predict the nitrification and the TSS removal performances for engineering projects. However, one of the backlashes in using such models is the number of parameters involved: a total of 45 parameters are included. A thorough sensitivity analysis of these, according to the procedure described by Petersen et al. (2002), has previously been carried out to classify the BAF model parameters into two classes, high and low sensitivity (Vigne et al., 2007). The parameters which held the strongest influence on the simulated output of the BAF model, were the physical characteristics of the biofilm, the kinetics properties of autotrophic biomass and the concentration of autotrophic biomass entering the system. These parameters required a specific determination either by an experimental measurement or by a mathematical estimation, whereas the other ones (not measurable) were obtained by calibration of the BAF model on data. A protocol showing the succession of calibration steps
(particular and then soluble components) was proposed to correctly adjust the BAF parameters (Vigne et al., 2007). The main objective of this paper is, thus, to evaluate the reliability of the BAF model found in GPS-X with semiindustrial pilot-scale data, and to point out for future modellers where priority should be given during calibration of this model. Experiments were carried out for over a year to collect the appropriate data set at both steady-states and dynamic conditions of a submerged biofilter operated in tertiary nitrification.
2.
Material and methods
The present section is divided in two parts. First, the data collection for model verification is described. Second, the procedures to determine or estimate the sensitive parameters are given.
2.1.
Data collection: biofilter experimental plant
A sampling campaign was carried out at a submerged up-flow biofilter pilot plant (Biostyr type) treating an average flow of 200 m3/d and located at the effluent of a medium loaded activated sludge plant in France (COD removal, continuous aeration, SRT of 2e3 days, 3 gMLSS/L in the activated sludge). The cylindrical shaped reactor was 6.65 m high with a diameter of 0.907 m. The reactor was filled with 3.5 m of a floating type media (polystyrene, 4 mm of diameter, average specific area of 955 m2 m3, porosity of 0.33). Part of the treated water was stored for backwashing. The pilot plant was fully automated. Process parameters such as hydraulic loading rate (CH), air flow (Vair), dissolved oxygen (DO), pH, temperature ( C) and filtration-backwashing runs were controlled, measured and recorded in an on-line data acquisition system. DO over 6 mgO2 L1 in treated water and a CODfiltered/NH4eN ratio lower than 2.5 in the influent were target conditions. The sampling campaign started in December 2005 and was completed in February 2007. After a 2 month start-up period from natural seeding of the filtering media by autotrophic bacteria, three runs were performed: three different volumetric nitrogen loading rates (0.5, 1.0 and 1.5 kg NH4eN (m3_media.d)1) were applied for three periods of seven, four and two months respectively. For each run, pseudo-steady state and dynamic conditions of inflow rates were carried out alternately, and the effect on nitrification performances were studied.
2.1.1.
Pseudo-steady state conditions and data collection
At pseudo-steady state conditions, influent ammonium concentration variations depended directly on the upstream plant performances. The hydraulic loading rate (CH) was maintained at two different constant values as shown in Table 1. Between the first and the second nitrogen loading rate periods, a CH increase from 3 to 6 m3 (m2.h)1 was carried out. Between the second and the third nitrogen loading rate periods, an alkali solution (20.5%) was added to the influent composition in order to increase the influent ammonium concentration.
4401
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 9 e4 4 1 0
Table 1 e Operating conditions during the sampling campaign. Event
Seeding
Run 1: CV ¼ 0.5 0.2 kgNH4 eN.(m3.d)1
Run 2:CV ¼ 1.0 0.2 kgNH4 eN.(m3.d)1
Run 3:CV ¼ 1.5 0.2 kgNH4 eN.(m3.d)1
Period
day 0eday 60 (Dec. 05eJan. 06)
day 61eday 270 (Feb.eAug. 06)
day 271eday 390 (Sep.eDec. 06)
day 391eday 450 (Jan.eFeb. 07)
3 12 25e30 13e15
3 8e15 15e30 12e22
6 8e15 15e30 14e22
6 15e19 30e36 14e18
CH (m3/m2.h) Vair (Nm/h) NH4eN (mgN/L) T ( C)
For the data collection, 24 h flow proportional composite samples were taken at the input and the output of the pilot plant. The following analyses, TSS, total COD, filtered COD, NH4eN, NO2eN, NO3eN and alkalinity were done according to the Standard Methods (APHA et al., 1998). Furthermore, five batch experiments were carried out to determine the COD fractionation of the influent. The combination of a biological method with BOD-tests (Roeleveld and Van Loosdrecht, 2002) and a physicochemical one (Mamais et al., 1993) were used.
2.1.2.
Dynamic conditions and data collection
At the end of each of the three steady-state periods, hydraulic peak-loads of different duration and magnitude were applied during filtration. For example, hydraulic peak-loads of 3 h were applied at the same period of the day, for several days with the same (or different) magnitudes. A rain period was also considered as a dynamic event and was used for the evaluation of the BAF model. The nitrogen concentrations were recorded every 15 min with on-line ammonium and nitrate analysers (Datalink instruments, AM200 and NT200) in the influent and in the effluent, respectively. Furthermore, 1 h flow proportional composite samples were taken during the hydraulic dynamic peak-loading period in order to check the correlation between the on-line analyser data and the laboratory data.
2.2.
Procedures for some parameters determination
The determination of some parameters is done through an experimental methodology carried out on a grab sample of media taken at the same height of the biofilter at different operational conditions. The following protocols have been chosen for the evaluation of the different parameters.
2.2.1.
Characteristics of biofilm
2.2.1.1. Mean density of dry biofilm (r). The protocol to measure the density of the biofilm (r) was adapted for spherical polystyrene beads from the one proposed by Horn and Hempel (1997). The ratio between the dry biofilm mass (M ) and the wet biofilm mass (MW) is computed as shown in equation (1) on grab samples of the biofilter media (triplicate measurements). r¼
M ðMW =rW Þ
(1)
The polystyrene beads were first weighed, dried and weighed again. The mass of the same amount of clean polystyrene beads
was deducted to obtain Mw and M. A wet density of the submerged biofilm of 1 g/cm3 (i.e. density of water) is assumed. The test was carried out after the application of three different nitrogen loading rates and with three different volumes of polystyrene beads (30, 70 and 100 mL) in order to check the protocol reproducibility.
2.2.1.2. Maximum attached liquid film thickness of the biofilm (Lf,lim). The maximum attached liquid film thickness of the biofilm (Lf,lim) is defined as the layer between the bulk liquid phase and the biofilm. Lf,lim (equation (2)) is calculated as the diffusion coefficient of water soluble components (DW, m2/s) divided by the mass transfer coefficient KM (m/s). Lf;lim ¼ DW =KM
(2)
Several equations relating Lf,lim to the Sherwood number (Sh), the Reynolds number (Re, defining hydraulic flow but not including the air velocity) and the Schmidt number (Sc, defining water properties) have been given in the literature (e. g. Horn and Hempel, 1995; Wanner et al., 2006). Ohashi et al. (1981) proposed a specific equation for fixed bed with spherical particles, which was used by Christiansen et al. (1995) on a submerged denitrifying biofilter (3 mm biostyren spheres). As no equation relating both the hydraulic flow and the air flow rate to the Reynolds number was found in the literature, Ohashi et al. (1981) correlation was used on the Biostyr pilot plant of this study, with inflow rate between 3 and 6 m3 (m2. h)1 (Re between 8 and 15) in order to evaluate Lf,lim. Sc, Sh and Lf,lim were deduced and adapted with the diffusion coefficients of soluble components inside the biofilm (DW).
2.2.1.3. Maximum and actual biofilm thickness (Lf,max and Lf). The maximum biofilm thickness (Lf,max) is defined as the ultimate (maximum) volume fraction of solids which could be fixed around the media. This parameter controls the quantity of solids and the growth of biomass that can be stored in the biofiltration column. Its direct determination is not possible as it depends on local hydraulic conditions. It is the evolution of the biofilm thickness with time that can be directly simulated for given operating conditions. The actual value (Lf(t), equation (3)) increases as a function of time according to the quantity of solids in the biofilm for a given density of the biofilm (r) and a maximum biofilm thickness (Lf,max), which is the highest value that can be taken by Lf. Lf ðtÞ ¼
Solids in biofilmðtÞ Lf;max Lf;max r
(3)
Lf is calculated (equation (4)) as the ratio between the volume of the biofilm (Va) and the specific surface area of the
4402
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 9 e4 4 1 0
biofilm (Aa). This last parameter is assumed to be the product of specific media surface area aG (m2/m3) and media sample volume. The volume of the biofilm is the ratio between biofilm wet mass (MW) and the biofilm wet density (rW). Lf ¼
Va MW ¼ Aa rW aG Media sample volume
(4)
Lf is a state variable and its value can be obtained by simulation and be compared to literature values obtained with microscopic scale measurement for equivalent operating conditions (Zhang and Bishop, 1994; Boller et al., 1994).
2.2.2. Concentrations of autotrophic biomass (input, inside, output) 2.2.2.1. Input with the influent XB,A_in (mass balance on the upstream plant). The autotrophic biomass concentration (XB, A_in) contained in the influent of the pilot plant was estimated from a nitrogen mass balance applied to the activated sludge located upstream. The methodology uses the daily nitrified nitrogen fluxes calculated from the concentration of nitrogen forms measured on the daily average flow proportional composite samples of raw and treated waters (Nowak et al., 1994; Choubert et al., 2005). The amount of the autotrophic biomass produced each day by nitrification (MXB,A(t)) is estimated by the sum of daily nitrified nitrogen (4Nnit) and autotrophic biomass load input with the influent (4XB,A). The fraction that disappears each day is estimated from decay process (assuming day rate, bA of 0.13 d1 at 10 C (Dold, 2002; Marquot, 2006; Choubert et al., 2008)) and from the evaluation of the mass removed by excess sludge extraction (Dx, withdrawal rate). The difference provides the daily net growth of the autotrophic biomass that accumulates in the system (equation (5)).
DMXBA ðtÞ ¼ YA $4Nnit þ 4XB;A;input ðbA þ DX Þ$MXB;A ðtÞ Dt
(5)
The concentration of the autotrophic biomass released in the effluent was obtained assuming a similar proportion of autotrophic biomass in the mixed liquor and in the effluent TSS.
2.2.2.2. Release with the backwash XB,A_out (batch-test). Ohashi et al. (1995) and Boller et al. (1997) have demonstrated that once a nitrifying biomass is attached to the granular material, it is not easily removed even by intense backwash scouring. In order to verify this result and to evaluate/calibrate the BAF model, a nitrification batch-test was carried out on the water sampled during a backwashing period (3 tests). NH4eN and NO3eN were recorded in an aerated reactor for 4 h where non-limiting conditions of ammonium and oxygen were applied. This test provides the nitrate production rate, and the quantity of available autotrophic biomass MXB,A_out in the backwashing water assuming that the maximum autotrophic growth rate of the biomass is the same as the one determined in the point 2.2.3.1.
2.2.2.3. Storage in the reactor XB,A_BAF (mass balance on the BAF). A similar mass balance method to the one presented for the upstream activated sludge was applied for the BAF pilot plant to estimate the amount of biomass stored in it.
Nitrogen concentration in the influent and the effluent of the pilot plant were used to compute the nitrified flux. The amount of biomass produced and eliminated (decay and backwashing water) was used to estimate the amount of autotrophic biomass stored in the BAF. Finally, the calculated mass was divided by the volume of the biofilm to obtain the concentration of the autotrophic biomass (XB,A, gCOD/m3_biofilm) accumulated in the biofilter column.
2.2.3. k a)
Kinetic parameters of autotrophic biomass (mA,MAX, bA,
The kinetic parameters related to the autotrophic biomass (mA, bA, ka) were found to be very sensitive (Vigne et al., 2007). The decay rate (bA, d1) and the ammonification rate (ka, m3/ gCOD/d), were first given literature values, as they are not easily measurable. The maximum specific growth rate (mA,MAX, d1) was first evaluated by calculation of the biomass quantity stored in the reactor (nitrogen mass balance (Nowak et al., 1994)), and second with a measure of the nitrification rate (Tschui et al., 1994). The nitrification rate was investigated with two protocols: one with grab samples of filtering media to determine the maximum nitrification rate; the other one requires ammonium sampling at different heights of the biofilter column to record the effective nitrification rate. MAX,
2.2.3.1. Maximum nitrification rate. The maximum nitrification rate (RV, MAX) of the filtering media is observed for nonlimiting aeration and substrates conditions, through recording NH4eN and NO3eN every 10 min for 1 h. The nitrate production rate is expressed as kgN.(m3_media.d)1 or kgN. (m3_biofilm.d)1 (Tschui et al., 1994) for samples taken at two heights of the biofilter. This measurement represents the product of the maximum autotrophic growth rate (mA,MAX) and the concentration of the corresponding biomass (XB,A, gCOD/ m3_biofilm) divided by the autotrophic yield coefficient YA (gCOD/gN) (equation (6)). RV;MAX ¼
mA;MAX XB;A YA
(6)
Finally, the maximum autotrophic growth rate coefficient is deduced from the combination of the autotrophic biomass concentration calculated with the nitrogen mass balance in the biofilter (XB,A), and the nitrification rate measurement carried out in batch-tests on media grab samples (RV,MAx).
2.2.3.2. NH4 bed profiles. The effective nitrification rate is determined by monitoring the ammonium uptake rate along the filtering media (bed profiles). This technique investigates the distribution of the biomass in a biofilter. The ammonium concentration was measured in samples taken at three different heights of the biofilter column (0.5 m, 1.5 m and 2.5 m from the bottom of the floating media). Sampling times were delayed according to the hydraulic retention time between each media height. The samples were taken for average nitrogen loading rates (steady-states conditions), and also under hydraulic peak-loads in order to select low and high ammonium conditions.
4403
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 9 e4 4 1 0
Table 2 e COD Fractioning of pilot plant influent from 5 tests. COD Fractionsa
Literature values on a primary effluent from Pasztor et al. (2009) (%)
Final chosen values for BAF influent model (%)
9 (3e15) 29 (14e57) 41 (21e54) 13 (8e19) 50e70
25 9 32 34 41
Si Ss Xs (¼XBH þ XBA) Xi Xs þ Ss
a COD values for the 5 tests are: 54, 67, 72, 80 and 136 gCOD/m3.
3.
Results and discussion
3.1.
Specific inputs for BAF model use
This section presents the results of the experimental methodologies aiming to determine the BAF influent fractionation (XBA and COD fractions), and biofilm characteristics (physical, concentrations and kinetic parameters).
3.1.1.
Tertiary influent characteristics
Besides conventional parameters (like COD, NH4eN), two additional influent variables were needed to be specified for the BAF model. This section reports the obtained values following the method described above. The concentration of the autotrophic biomass entering the system has been calculated with an accurate nitrogen mass balance on the upstream activated sludge process. As almost no nitrification occurred (continuous aeration, COD removal, NO3eN < 2 mgN/L) in the process, only 11 and 16 g CODauto3 were computed in the upstream activated trophic biomass.m sludge mixed liquor. As a result, the autotrophic biomass concentration in tertiary influent (XB,A_in) was estimated at 0.11 0.10 g COD_biomass.m3 for the summer period, and 0.20 0.10 g COD_biomass.m3 for the winter period. This difference is mainly caused by a smaller gap between autotrophic growth rate and decay rate in winter than in summer (Marquot, 2006; Choubert et al., 2009). XB,A_in was considered close to zero for the influent applied to the BAF model. The COD of the influent entering the pilot-scale plant has been split into the four usual fractions (Table 2). All Xs was expected to be in the form of XBH and XBA because of the upstream activated sludge. The results were compared with COD fractionations found in literature for primary effluents (Pasztor et al., 2009). Uncertainties related to higher analytical error at small values of COD might influence test results. The sensitivity analysis of the COD fractionation has always shown an influence lower than 5% on tertiary nitrification performance (Vigne et al., 2007). No competition on autotrophic biomass was observed probably due to the applied non-limiting conditions of oxygen coupled with the low content of biodegradable COD in the influent.
and 1.03 d1. Considering a temperature change coefficient (q) of 1.059 (Marquot et al., 2005), the mA,MAX converted to 20 C is in the range of 0.83 and 0.90 d1 (average: 0.86 0.03 d1) which is in an agreement with literature (Copp and Murphy, 1995). Considering the nitrogen mass balance, the instant average concentration of XB,A contained in the BAF was computed and found to be between 0.6 and 1.2 kgCOD/ m3_media seeded (Fig. 1), or 3580 and 7150 gCOD/m3_biofilm if considering the volume occupied by the biofilm volume. A 17% increase in XB,A between runs 1 and 2, and a 27% increase between runs 2 and 3 were observed. Maximum nitrification rates measured on backwashing water with batch-tests (0.014e0.034 kgN (m3_reactor.d)1) are presented in Table 3. Values found, corresponding to a autotrophic biomass concentrations of 4e11 gCODbiomass/m3, are 10 times lower than the ones observed for a low loaded activated sludge process (N removal with SRT of 12 days, Monti and Hall (2008)). Loss of autotrophic biomass during backwashing accounts for approximately 2e7% of the amount of autotrophic biomass stored in the biofilter (Boller et al., 1997). The measured concentrations values of XB,A, in the filtering media and in the backwashing, are very useful values to evaluate the results of simulations and to make the calibration more robust.
3.1.3.
Physical characteristics of the attached biofilm
Physical characteristics of the attached biofilm determined with Equations (1), (2) and (3), for the three periods of operations are shown in Table 3. Between run 1 and run 2, the hydraulic loading rate was increased from 3 to 6 m3 (m2.h)1, resulting in a pollutant loads increase. In both, similar values have been obtained for
3.1.2. Autotrophic biomass in the filtering media and in backwash water RV,MAX measured in batch-tests with filtering media (temperature between 17 and 23 C) lie between 2.5 and 3.7 kgN. (m3_media.d)1. The corresponding mA,MAX was between 0.71
Fig. 1 e Kinetic parameters and mean concentration of autotrophic biomass in the filtering media for bA(20 C) [ 0.17 dL1 and q [ 1.059 (period 0e450 days).
4404
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 9 e4 4 1 0
Table 3 e Main operating conditions of the BAF, nitrification characteristics (maximum rate (RV,MAX) and XB,A released by the backwashing water at 20 C) and physical characteristics of the biofilm for media sampled after backwash, at 0.5 m from the bottom of the filter.
Operating conditions
Autotrophic biomass
Biofilm
Run Period
1 day 61eday 270
2 day 271eday 390
3 day 391eday 450
Seeded height (m) Hydraulic loading rate CH (m3.(m2.h)1) Reynolds Re (-) Backwashing frequency (period.d1) TSS volumetric loading rate (kgTSS.(m3.d)1) COD volumetric loading rate (kgCOD.(m3.d)1) Nitrogen volumetric loading rate (kgNH4eN.(m3.d)1) RV,MAX in the biofilm (kgN/m3_media/d) RV,MAX in the backwashing water (kgN/m3_media/d) XB,A in the backwashing water(gCOD/m3) Removed fraction by backwashing/autotrophic biomass stock (%) Dry mean density r (kgTSS/m3_biofilm) Mean biofilm thickness Lf (mm) Attached liquid film thickness Lf,lim (mm)
2.5 3 8 1 0.8 0.2 1.9 0.2 0.6 0.2 2.5 0.014 4 0.5 21
3.5 6 15 1 1.1 0.2 2.6 0.2 1 0.2 3.0 0.017 6 0.5 31
3.5 6 15 3 1.5 0.2 4 0.2 1.3 0.2 3.67 0.034 11 1 71
70 5 85 5 120 20
75 5 100 5 90 15
120 5 120 5 90 15
the mean density (r) of the biofilm, 70 and 75 kg TSS/m3_biofilm (7% difference). But, the mean biofilm thickness (Lf) were different: 85 and 100 mm (18% higher in run 2). The attached liquid film thickness of the biofilm (Lf,lim) were also different: 120 and 90 mm (25% lower in run 2). In run 3, the pollutants loading rates were increased and the daily backwash frequency increased by a factor three. The hydraulic loading rate was similar to Run 2 (6 m3(m2.h)1). The mean density of the biofilm (120 kg TSS/m3_biofilm) and the mean thickness (120 mm) are higher than in run 2 by 60% and 20% respectively (5% variation of the measurement). The same value (90 mm) is obtained for the attached liquid film thickness of the biofilm (Lf,lim). These individual values might be preferred by users of the BAF model instead of the default values. From our observations, the characteristics of the biofilm seemed to be governed more by the interval time separating backwash periods and by the pollutants loading rates, than by the hydraulic loading rate itself, as is often considered. The turbulence (higher Reynolds number) is known to increase both the advective transport inside the biofilm layers, which leads to increase biofilm thickness (Tschui et al., 1994), and the hydraulic channelling, which results in a thinner liquid film (Lf,lim). Both modifications result in an improved nitrification efficiency (Zhu and Chen, 2001). Furthermore, it was shown that the biofilm dry density measured after backwash increases with time (Ohashi et al., 1995) and along biofilm depth (Ohashi and Harada, 1994). Until now, no mathematical relation was determined to link the characteristics of the biofilm and give the model more accuracy.
3.2.
Parameters determination for BAF model
performance filtration-backwash). For this simulation, the default values of the parameters given in the BAF model were used. Particular concentrations inside the biofilm (XB,A, XB,H and Xi) account for 5%, 24% and 60% respectively of the particulate COD. The XB,A concentration ranges between 4500 and 5500 gCOD/m3_biofilm depending on the applied volumetric nitrogen loading rate. The heterotrophic biomass concentration XB,H is five times higher. According to the evaluation strategy proposed in Vigne et al. (2007), the following steps were pursued for each run defined in Table 1: (Step 1) a preliminary simulation carried out at pseudo-steady states with actual operating conditions, and (Step 2) simulation under dynamic condition (event like an inflow rate increase). For a correct calibration of the BAF model, the best fit between simulated and observed values of several variables were needed to obtain: maximum nitrification rate (RV,MAX) of the filtering media and of backwashing water, nitrogen forms in effluent, characteristic of the biofilm. The previously presented values of the measured biofilm characteristics (Table 3) have been put into the model for the density (r), the maximum biofilm thickness (Lf,max) and the liquidebiofilm interface thickness (Lf,lim), as the experimental values were obtained with 5% relative standard deviation and are similar to literature values. The attached liquid film thickness (Lf,lim) of the biofilm applied was linked with the hydraulic loading and mixing intensity with the Ohashi et al. (1981) equation (120 mm for Run 1 with CH ¼ 3 m3(m2.h)1 and 90 mm for Run 2 with CH ¼ 6 m3(m2.h)1). Moreover, the kinetic parameters values were modified according to experimental values measured for the maximum growth rate (mA, max) and for the decay rate (bA) (Marquot, 2006; Manser et al., 2006).
3.2.1.
Initialization and strategy
3.2.2.
An initialization period was first applied to estimate the values of the particular concentrations inside the biofilm (XB,A, XB,H and Xi) after the seeding period. The simulation was done using average operating conditions applied to the pilot plant (constant inflow rate and concentrations, treatment
Calibrated set of parameters
The obtained parameter values are gathered in Table 4 (temperature of 20 C): default and modified values are shown for each parameter and for each step. It is interesting to note that most parameters stay constant even though the operating conditions change.
E E M E M M M E E E Biological conversion ASM1
a M ¼ value measured or taken from literature. E ¼ value estimated from measured data compared with simulated data.
0.00002 0.001 5.7 0.004 0.24 0.86 0.17 3 0.04 1 0.00002 0.001 5.7 0.004 0.24 0.86 0.17 3 0.04 1 0.00002 0.001 5.7 0.004 0.24 0.86 0.17 3 0.04 1 0.00002 0.001 5.7 0.01 0.24 0.8 0.04 3 0.08 1 Filtration and Backwash
70 170 0.5 0.00012 0.001 100 170 0.5 0.00005 0.001
Dry mean density r (kgTSS/m3 biofilm) Maximum thickness Lf,max (mm) Reduction coefficient of diffusivity fD () Maximum attached liquid film thickness Lf,lim (m) Surface solids detachment coefficient rate kdetach (kg/m2.d) Internal solids exchange coefficient rate (m/d) Clean bed filter coefficient lo () Packing factor () Solids removal rate during backwash kextr (d1) Autotrophic yield YA (gCOD/gN) Maximum specific autotrophic growth rate mA,MAX (d1) Autotrophic decay rate bA (d1) Maximum specific hydrolysis rate kh (d1) Ammonification rate ka (m3/gCOD/d) Half-saturation coefficient for ammonium KNH (g/m3) Physical characteristics of biofilm Mass transfer of biofilm
4405
Default values of kinetic parameters were not modified for the maximum specific hydrolysis rate (kh) and half-saturation coefficient for ammonium (KNH). Concerning the packing factor and the autotrophic yield (YA), their values were voluntarily maintained at the default value because they are thought to not vary significantly from one biological treatment process to another and from one operating and environmental condition to another. The clean bed filter coefficient (lo), the surface solids detachment coefficient rate (kdetach) and the internal solids exchange coefficient rate were also maintained at the default value as it provided TSS in treated water close to the observed value. The following parameters modifications have been carried out after the simulations of pseudo-steady state (step 1) and dynamic conditions (Step 2) of Run 1, 2 and 3.
0.00002 0.001 5.7 0.004 0.24 0.86 0.17 3 0.04 1
M E E M E 120 170 0.55 0.00009 0.001 70 170 0.55 0.00009 0.001 70 170 0.5 0.00012 0.001
Run 3 Run 2 Run 1
Default values (initial step) Parameters of the biofilm Modules
Table 4 e Parameters values under different conditions (at 20 C).
Values in pseudo-steady state conditions (step 1)
Values in dynamic hourly nitrogen concentrations (step 2)
Determination techniquea
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 9 e4 4 1 0
3.2.2.1. Under pseudo-steady state (step 1). The simulations have shown that the ammonification rate (ka) and the solids removal rate during backwash (kextr) needed to be modified from 0.08 to 0.04 d1, and from 0.01 to 0.004 d1 respectively. These modifications increase the predicted autotrophic biomass stored in the BAF and lower the difference between observed and predicted values. The use of the default value for the ammonification rate (ka) provided predicted TKN concentrations in the effluent below 2 mgN/L, whereas usual values between 2 and 3 mgN/L were measured. Lowering ka by 50% brought a decrease of the ammonification rate, which is plausible because of low organic nitrogen at tertiary nitrification stage. Simulation results with a default kextr value predicted 50% too high autotrophic biomass concentration in the backwashing effluent in comparison with the observed value from batch-tests (Table 3). kextr was lowered by 60% to adjust the amount of the autotrophic biomass inside the filtering media and the autotrophic concentration in the backwashing water. 3.2.2.2. Under dynamic conditions (step 2). To lower the difference between simulated and observed values, the simulations carried out under dynamic conditions (with measurement of hourly nitrogen concentrations) were used. Seven parameters were modified (underlined bold in Table 4). The increase of the reduction coefficient of diffusivity ( fD) from 0.5 to 0.55 was carried out between Run 1 and Run 2 to increase the diffusion of ammonium in the biofilm (the increase of growth rate of autotrophic was inefficient due to substrate-rate limitation). It can be noted that the maximum biofilm thickness given by the user (Lf,max) was voluntarily maintained at the default value which gave good results between predicted and observed biofilm thickness for the three runs.
3.3.
BAF model performance
The capacity of the BAF model to predict nitrogen removal in submerged biofiltration systems for short-term and long-term periods are presented. The data recorded during two periods [14-days (day 129e142) and 21-days (day 273e294) long period] are used to this end. The parameters used for each run are then compared. Dynamic variations of nitrogen forms were observed at different hourly nitrogen feed rates, 3-h inflow
4406
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 9 e4 4 1 0
rate increase periods, and rainfall events. The predicted results obtained with the new set of parameters proposed in Table 4 are compared to the observed data in Figs. 2 and 3.
3.3.1. days)
Dynamic modelling for short-term periods (15e21
For Run 1 (Fig. 2), the NO3eN and the NH4eN concentrations in treated water were predicted with a difference of about 1.0 0.5 mgN/L (þ6%) and 1.5 0.5 mgN/L (8%) respectively compared to the experimental measurements. Inside the filtering media (bed profiles), the differences between simulated and actual NH4eN concentrations is between 10 and 15%. The predicted autotrophic biomass concentration and the maximum nitrification rate are 2 to 3% higher than the observed ones. For Run 2 (Fig. 3), the difference between simulated and observed biological variables (XB,A and RV,MAX) was 7%; during hydraulic peak-loads and pseudo-steady state periods, the simulated nitrate concentrations were 5e10% and 2% lower than the experimental ones respectively. The
simulated XB,A concentration in the backwashing water range between 4 and 6 g COD.m3 for the two sets of data and fits the experimental ones. Finally, the predicted biofilm thickness is 105 mm. This value is in the range of 85 and 105 mm and corresponds to previously reported values (Tschui et al., 1994; Deront et al., 1998). The results presented in Figs. 2 and 3 confirm the ability of the BAF model to correctly represents the dynamic and longterm operations of a semi-industrial pilot-scale Biostyr type. With an initialization step (inflow rate, loading rate, sludge concentration, filtration-backwash runs, temperature, etc.) as accurate as possible, autotrophic biomass concentration and maximum nitrification rate into the filtering media as well as the predicted nitrogen concentrations correspond very closely to the experimental measurements. Then, the best fit was reached when applying the calibration protocol suggested by Vigne et al. (2007) with the set of calibrated parameters presented in Table 4. Additional experimental data were used to verify the calibrated model. Simulations were carried out at
Fig. 2 e Observed and predicted nitrogen concentration (period 129e142 days) in the effluent (at the top), biological variables into the filtering media (in the middle) and nitrogen bed profiles along the filter (at the bottom) during 3 h hydraulic peak load period for the first nitrogen loading rate (Run 1) with the default parameters (on the left) and with the calibrated parameters (on the right).
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 9 e4 4 1 0
4407
Fig. 3 e Observed and predicted nitrogen concentration (period 273e294 days) in the effluent (at the top), biological variables into the filtering media (in the middle) and nitrogen bed profiles along the filter (at the bottom) for the second nitrogen loading rate (Run 2) with the calibrated parameters for Run 1 (on the left) and with the calibrated parameters (on the right).
the first and the second nitrogen loading rate with different magnitudes and durations of hydraulic dynamic peak-loads applied to the system. These simulations have shown good agreement between observed and predicted values. As an example, for a simulation period of 60 days at the first loading rate during summer 2006, the difference between observed and predicted data is around 5% for NO3eN and NH4eN concentrations, around 10% for autotrophic biomass concentration and 5% for the RV, MAX.
3.3.2.
Dynamic modelling for a long-term period (12 months)
The comprehensive study has shown that, due to the hydraulic loading rate effect, some values of the parameters of the biofilm calibrated for Run 1 needed to be modified to fit the observed data of Run 2. For instance, the increase from 3 to 6 m3 (m2.h)1 requires modification of the attached liquid film thickness (Lf,lim) of the biofilm from 120 to 90 mm, whereas this parameter is assumed constant in the BAF model. The impact of mixing conditions and reactor scale should have a significant influence on kinetic parameters, contrary to microbial kinetic and stoichiometric parameters that are not influencing
parameters, as demonstrated by Brockmann et al. (2007) using AQUASIM (Reichert, 1998) as a simulation software. The modification of the ammonium bed profiles along the filter proves that the nitrification rate increased between Run 1 and Run 2. The increase in the hydraulic loading rate and the increased aeration intensity can be responsible for a higher mass transfer of soluble components inside the biofilm layers. Between Run 1 and Run 2, the modifications are performed with the modification of Lf,lim or fD coefficients whereas the kinetic and the stoichiometric parameters are maintained constant (same value for ka or mA,MAx). In that case, the correlation from Ohashi et al. (1981) links hydraulic systems and Lf,lim. It has showed a good fit between simulations and experimental values. The effect of aeration intensity was taken into account by varying the fD coefficient. The tests carried out during Run 3 have shown the increase of the biofilm dry density compared to Run 1 and 2, while the frequency of backwashing has been increased from once a day to three times a day. The increased backwash frequency could have induced a more important biofilm abrasion resulting in the biomass compression with the conservation of
4408
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 9 e4 4 1 0
Fig. 4 e Observed and predicted biological variables (period 420e447 days) into the filtering media (on the left) and nitrogen bed profiles along the filter (on the right) for the third nitrogen loading rate with the calibrated parameters for Run 1.
autotrophic biomass concentration set in the deeper layers of the biofilm in accordance with literature (Ohashi et al., 1995). For a nitrogen loading rate increase of 50% between Run 2 and 3, the increase of XB,A in the filtering media is about 33% as shown in Fig. 4, whereas it was about 6% between Run 1 and Run 2 for a nitrogen loading rate increase of 100%. Such observations are important to include in the BAF model and make it more accurate for long-term simulations, and thus extend its domain of valid operating conditions. The elaboration of a continuous predictive relation between the backwashing frequency and the biofilm density is also required to get a realistic model. In Run 3, the slope of the ammonium bed profile is more important than in Run 1 and Run 2 since the quantity of XB,A is more important despite backwash frequency.
4.
Conclusion
The application of a previously published calibration protocol allowed to determine more precisely some parameters integrated in the BAF model, which governed nitrification performances as shown by experimentation: quantity of autotrophic biomass accumulated inside the biofilter and extracted during backwashing, ammonium profiles along the biofilter as well as nature of nitrogen compounds in the effluent. The combination of experimental observations and simulation results has highlighted the capacity of the BAF model to predict actual nitrification even during dynamic conditions. Among the calibrated parameters, four parameters considerably reduced the difference between the model predictions and the observations: density of biofilm (r), attached liquid film thickness of biofilm (Lf,lim), ammonification rate (ka) and solids removal rate during backwash (kextr). Concerning kinetic parameters, they did not need to be modified once maximum autotrophic growth rate mA,MAX was measured by batch-tests and decay rate bA was assessed more precisely from recent studies carried out on activated sludge process. However, this study pinpoints the limitations concerning the estimation of the parameters of the BAF model, which must be applicable under a wide variety of conditions and within realistic boundaries. For this extensive study, the need to change a mass transfer parameter (attached liquid film thickness of biofilm Lf,lim) after an increase of the hydraulic loading rate by a factor of two was shown. An improvement would also be to introduce the influence of aeration on the Re
number and consequently on the liquid film thickness in the Ohashi et al. (1981) correlation. The same conclusion applies to the density of biofilm, which should be increased with the biofilm formation in time and with the increase of backwashing frequency, the main parameter that influences the repartition of biomass along the filter. In fact, it must be concluded that some of the model parameters should be changed into variables that have to be modelled themselves. To obtain a more reliable model, the estimation of the parameters from a combination of experiments carried out at different experimental conditions is necessary. Even if the BAF model considers the whole fundamental mechanisms found in the filtering media, it is the acquisition of several data sets that will allow to enhance the understanding of the process dynamics and will improve model predictions by the establishment of correlations between some operating variables conditions and BAF parameters. From this point on, with the set of calibrated parameters obtained for this study, the project will focus on the comparison between model predictions and experiments carried out on solids budgets (TSS) and total head loss development in the biofilter in order to validate the model robustness or pinpoint some weaknesses in the translation of some phenomena linked to head loss development.
Acknowledgements Experimental work has been supported by OTVeOmnium de Traitement et Valorisation and Anjou Recherche, France in collaboration with Cemagref. The authors also thank NSERC of Canada for financial support. The authors thank the Cemagref technical staff for their assistance in data collection and chemical analysis.
references
APHA, AWWA, WPCF, 1998. Standard Methods for the Examination of Water and Wastewater, 20th ed. Amer Public Health Assoc., Amer. Water Works Assoc, Water Poll. Control Fed., Washington, DC. Boller, M., Gujer, W., Tschui, M., 1994. Parameters affecting nitrifying biofilm reactors. Water Science and Technology 29 (10e11), 1e11. Boller, M., Kobler, D., Koch, G., 1997. Particle separation, solids budgets and head loss development in different biofilters. Water Science and Technology 36 (4), 239e247.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 9 e4 4 1 0
Brockmann, D., Rosenwinkel, K.H., Morgenroth, E., 2007. Estimation of kinetic parameters of a model for deammonification in biofilms and evaluation of the model. Water Science and Technology 55 (8e9), 291e299. Canler, J.P., Perret, J.M., Lengrand, F., Iwema, A., 2003. Nitrification in biofilters under variable load and low temperature. Water Science and Technology 47 (11), 129e136. Choubert, J.M., Racault, Y., Grasmick, A., Beck, C., Heduit, A., 2005. Nitrogen removal from urban wastewater by activated sludge process operated over the conventional carbon loading rate limit at low temperature. Water SA 31 (4), 503e510. Choubert, J.M., Marquot, A., Stricker, A.E., Gillot, S., Racault, Y., He´duit, A., 2008. Maximum growth and decay rates of autotrophic biomass to simulate nitrogen removal at 10 C with municipal activated sludge plants. Water SA 34 (1), 71e76. Choubert, J.M., Stricker, A.E., Marquot, A., Gillot, S., Racault, Y., He´duit, A., 2009. Updated ASM1 model parameter values for improved prediction of nitrogen removal in activated sludge processes: validation at 13 full-scale plants. Water Environment Research 81 (9), 858e865. Christiansen, P., Hollesen, L., Harremoes, P., 1995. Liquid film diffusion on reaction rate in submerged biofilters. Water Research 29 (3), 947e952. Copp, J.B., Murphy, K.L., 1995. Estimation of the active nitrifying biomass in activated sludge. Water Research 29 (8), 1855e1862. Deront, M., Samb, F., Adler, N., Pe´ringer, P., 1998. Biomass growth monitoring using pressure drop in a cocurrent biofiltre. Biotechnology and Bioengineering 60 (1), 97e104. Dold, P.L., 2002. Importance of Decay Rate in Assessing Nitrification Kinetics. WEFTEC, Chicago. Farabegoli, G., Chiavola, A., Rolle, E., Stracquadanio, S., 2004. Experimental study on nitrification in a submerged aerated biofilter. Water Science and Technology 49 (11e12), 107e113. Han, D.W., Yun, H.O., Kim, D.J., 2001. Autotrophic nitrification and denitrification characteristics of an upflow biological aerated filter. Journal of Chemical Technology and Biotechnology 76, 1112e1116. Henze, M., Grady Jr., C.P.L., Gujer, W., Marais, G.V.R., Matsuo, T., 1987. A general model for single-sludge wastewater treatment systems. Water Research 21 (5), 505e515. Hidaka, T., Tsuno, H., 2004. Development of a biological filtration model applied for advanced treatment of sewage. Water Research 38 (2), 335e346. Horn, H., Hempel, D.C., 1995. Mass transfer coefficients for an autotrophic and a heterotrophic biofilm system. Water Science and Technology 32 (8), 199e204. Horn, H., Hempel, D.C., 1997. Growth and decay in an auto-/ heterotrophic biofilm. Water Research 31 (9), 2243e2252. Horn, H., Alex, J., Beier, M., Brockmann, D., Morgenroth, E., Ochmann, C., So¨rensen, K., Wanner, O. and Wichern, M. (2008). Simulation of nitrification in a full-scale biofilter e comparison of different approaches. In Proceedings of Biofilm Technologies Conference, Singapore, 8e10 January 2008. pp 229e231. Horner, R.M.W., Jarvis, R.J., Mackie, R.I., 1986. Deep bed filtration: a new look at the basic equations. Water Research 20 (2), 215e220. Hydromantis Inc. (2003). GPS-X 4.1 Technical reference. pp. 130e138 Jeong, J., Hidaka, T., Tsuno, H., Oda, T., 2006. Development of biological filter as tertiary treatment for effective nitrogen removal: biological filter for tertiary treatment. Water Research 40 (6), 1127e1136. Lessard, P., Lamouroux, F., Vigne, E., Samie, G. and Rocher V. (2008). Simulation du comportement d’une station par biofiltration. In Proceedings of Canadian Society for Civil Engineers Conference, Que´bec, Qc, Canada, June 10e13 2008.
4409
Mamais, D., Jenkins, D., Pitt, P., 1993. A rapid physical chemical method for the determination of readily biodegradable soluble COD in municipal wastewater. Water Research 27 (1), 195e197. Mann, A.T., Stephenson, T., 1997. Modelling biological aerated filters for wastewater treatment. Water Research 31 (10), 2443e2448. Manser, R., Gujer, W., Siegrist, H., 2006. Decay processes of nitrifying bacteria in biological wastewater treatment systems. Water Research 40 (12), 2416e2426. Marquot, A., 2006. Modelling Nitrogen Removal by Activated Sludge on Full-scale Plants: Calibration and Evaluation of ASM1. Universite´ de Pau et des Pays de l’Adour, 250 pp. Marquot, A., Stricker, A.E., Racault, Y., 2005. ASM1 dynamic calibration and long term-validation for an intermittently aerated WWTP. Water Science and Technology 53 (12), 247e256. Monti, A., Hall, E.R., 2008. Comparison of nitrification rates in conventional and membrane-assisted biological nutrient removal processes. Water Environment Research 25 (6), 497e506. Nowak, O., Schweighofer, P., Svardal, K., 1994. Nitrification inhibitioneA method for the estimation of actual maximum autotrophic growth rates in activated sludge systems. Water Science and Technology 30 (6), 9e19. Ohashi, A., Harada, H., 1994. Adhesion strength of biofilm developped in an attached-growth reactor. Water Science and Technology 29 (10e11), 281e288. Ohashi, A., Viraj de Silva, D.G., Mobarry, B., Manem, J.A., Stahl, D. A., Rittmann, B.E., 1995. Influence of substrate C/N ratio on the structure of multi-species biofilms consisting of nitrifiers and heterotrophs. Water Science and Technology 32 (8), 75e84. Ohashi, H., Sugawara, T., Kikuchi, K., Konno, H., 1981. Correlation of liquid-side mass transfer coefficient for single particles and fixed beds. Journal Chemical Engineering Japan 14, 6. Pasztor, I., Thury, P., Pulai, J., 2009. Chemical oxygen demand fractions of municipal wasteawater for modeling of wastewater treatment. International Journal of Environmental Science and Technology 6 (1), 51e56. Payraudeau, M., Paffoni, C., Gousailles, M., 2000. Tertiary nitrification in an up flow biofilter on floating media: influence of temperature and COD load. Water Science and Technology 41 (4e5), 21e27. Petersen, B., Gernaey, K., Henze, M., Vanrolleghem, P., 2002. Evaluation of an ASM1 model calibration procedure on a municipal-industrial wastewater treatment plant. Journal of Hydroinformatics 4, 15e38. Plattes, M., Henry, E., Schosseler, P.M., Weidenhaupt, A., 2006. Modelling and dynamic simulation of a moving bed bioreactor for the treatment of municipal wastewater. Biochemical Engineering Journal 32, 61e68. Reichert, P., 1998. AQUASIM 2.0eUser Manual, Computer Program for the Identification and Simulation of Aquatic Systems. Swiss Federal Institute for Environmental Science and Technology (EAWAG), Du¨bendorf, CH. Rocher, V., Paffoni, C., Goncalves, A., Azimi, S., Winant, S., Legaigneur, V., Gousailles, M., 2007. La biofiltration des eaux use´es: comparatif technique et e´conomique de diffe´rentes configurations de traitement. La Houille Blanche 1, 95e102. Roeleveld, P.J., Van Loosdrecht, M.C.M., 2002. Experience with guidelines for wastewater characterisation in The Netherlands. Water Science and Technology 45 (6), 77e87. Sanz, J.P., Freund, M., Hother, S., 1996. Nitrification and denitrification in continuous upflow filters process modelling and optimization. Water Science and Technology 34 (1e2), 441e448. Spengel, D.B., Dzombak, D.A., 1992. Biokinetic modeling and scale-up considerations for rotating biological contactors. Water Environment Research 64 (3), 223e235.
4410
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 9 e4 4 1 0
Tschui, M., Boller, M., Gujer, W., Eugster, J., Mader, C., Stengel, C., 1994. Tertiary nitrification in aerated pilot biofilters. Water Science and Technology 29 (10e11), 53e60. Vigne, E. (2007). Etude et mode´lisation dynamique d’un proce´de´ par biofiltration. Ph.D. thesis, Universite´ Laval, Que´bec, Canada. Vigne, E., Choubert, J.M., Canler, J.P., Heduit, A., Lessard, P., 2007. Toward an operational dynamic model for tertiary nitrification by submerged biofiltration. Water Science and Technology 55 (8e9), 301e308. Viotti, P., Eramo, B., Boni, M.R., Carucci, A., Leccese, M., Sbaffoni, S., 2002. Development and calibration of
a mathematical model for the simulation of the biofiltration process. Advances in Environmental Research 7 (1), 11e33. Wanner, O., Eberl, H., Morgenroth, E., Noguera, D.R., Picioreanu, C., Rittmann, B.E. and van Loosdrecht, M. (2006). Mathematical Modeling of Biofilms. IWA, Scientific and Technical Report. Zhang, T.C., Bishop, P.L., 1994. Density, porosity, and pore structure of biofilms. Water Research 28 (11), 2267e2277. Zhu, S., Chen, S., 2001. Impacts of Reynolds number on nitrification biofilm kinetics. Aquacultural Engineering 24, 213e229.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
A theoretical study of a direct contact membrane distillation system coupled to a salt-gradient solar pond for terminal lakes reclamation Francisco Sua´rez a,*, Scott W. Tyler b, Amy E. Childress c a
Graduate Program of Hydrologic Sciences, University of Nevada, Reno, 1664 N. Virginia St. MS 175, Reno NV 89557, USA Department of Geological Sciences and Engineering, University of Nevada, Reno, 1664 N. Virginia St. MS 175, Reno NV 89557, USA c Department of Civil and Environmental Engineering, University of Nevada, Reno, 1664 N. Virginia St. MS 258, Reno NV 89557, USA b
article info
abstract
Article history:
Terminal lakes are water bodies that are located in closed watersheds with the only output
Received 23 February 2010
of water occurring through evaporation or infiltration. The majority of these lakes, which
Received in revised form
are commonly located in the desert and influenced by human activities, are increasing in
5 May 2010
salinity. Treatment options are limited, due to energy costs, and many of these lakes
Accepted 29 May 2010
provide an excellent opportunity to test solar-powered desalination systems. This paper
Available online 9 June 2010
theoretically investigates utilization of direct contact membrane distillation (DCMD) coupled to a salt-gradient solar pond (SGSP) for sustainable freshwater production at
Keywords:
terminal lakes. A model for heat and mass transport in the DCMD module and a thermal
Thermal desalination
model for an SGSP were developed and coupled to evaluate the feasibility of freshwater
Renewable energy
production. The construction of an SGSP outside and inside of a terminal lake was studied.
Lake remediation
As results showed that freshwater flows are on the same order of magnitude as evaporation, these systems will only be successful if the SGSP is constructed inside the terminal lake so that there is little or no net increase in surface area. For the study site of this investigation, water production on the order of 2.7 103 m3 d1 per m2 of SGSP is possible. The major advantages of this system are that renewable thermal energy is used so that little electrical energy is required, the coupled system requires low maintenance, and the terminal lake provides a source of salts to create the stratification in the SGSP. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Terminal lakes are water bodies that are located in closed watersheds and therefore, the only output of water occurs through evaporation and infiltration. Because evaporation and infiltration are the only outflows, terminal lakes are typically more sensitive to environmental pollutant inputs than lakes that have river outflows (Beutel et al., 2001; UNEP, 2001). Terminal lakes can exist in any climate, but they are
commonly found in desert locations. The Caspian Sea (Europe and Asia), Dead Sea (Asia), Lake Chad (Africa), Salton Sea (CA), Great Salt Lake (UT), and Walker Lake (NV) are some examples of terminal lakes that occur around the world. In natural conditions, the water level of terminal lakes is in a long-term equilibrium, where the inflow from precipitation, rivers and streams, and groundwater discharge is equal to the evaporation and infiltration. However, if inflows are reduced, e.g., due to agricultural diverting, the water level in terminal lakes can
* Corresponding author. Tel.: þ1 775 784 4986; fax: þ1 775 784 1953. . E-mail addresses:
[email protected] (F. Sua´rez),
[email protected] (S.W. Tyler),
[email protected] (A.E. Childress). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.05.050
4602
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
Fig. 2 e Configuration of a salt-gradient solar pond and heat fluxes used in the simplified thermal model (UCZ represents the upper convective zone, NCZ is the nonconvective zone, and LCZ is the lower convective zone).
Fig. 1 e Temperatures, concentrations, and heat and mass transfer resistances in direct contact membrane distillation.
drop significantly. This changes the storage in the lake and slowly increases the concentration of salts, which are continuously transported to the lake by surface water and groundwater (Lopes and Allander, 2009). The presence of excess salinity and solar irradiation in the majority of these lakes provides an excellent opportunity to test desalination systems driven by solar energy. One promising solution to decrease water salinity in terminal lakes is the utilization of low-temperature membrane distillation (MD) coupled to a renewable energy source. Membrane distillation is a temperature-driven membrane separation process that has the potential to become a viable tool for water desalination (Al-Obaidani et al., 2008). Direct contact membrane distillation (DCMD) is a configuration of MD where warmer feed solution is in contact with one side of a microporous hydrophobic membrane and cooler water (permeate) is in direct contact with the opposite side of the membrane (Fig. 1) (El-Bourawi et al., 2006). DCMD uses hydrophobic membranes where only volatile components are transported through the membrane pores. The DCMD process is characterized by simultaneous heat and mass transfer; the driving force is the vapor pressure gradient across the membrane, and the water vapor flux through the membrane results in highly pure permeate (Cath et al., 2004). DCMD is one of the simplest configurations of MD; it requires only a membrane module, low-grade heat source, and two low-pressure pumps to pass the liquids over the membrane. The simplicity of this configuration makes it highly suitable for implementation in remote locations where technical support is not readily available and the capital cost is low compared to other membrane systems driven by renewable energy (Hsu et al., 2002; Mathioulakis et al., 2007). An ideal method for providing a renewable source of heat for the DCMD system at a terminal lake is a salt-gradient solar pond (SGSP). An SGSP is an artificially stratified water body
that is heated by absorption of solar radiation and that can provide long-term thermal storage and recovery for the collected energy (Kurt et al., 2000). It consists of three thermally distinct layers (Fig. 2): the upper convective zone (UCZ), the non-convective zone (NCZ), and the lower convective zone (LCZ). The UCZ is a relatively thin layer of cooler and fresher water. The NCZ consists of a salt gradient that suppresses convection within the pond, and thus, the NCZ acts as insulation for the LCZ. The LCZ is the layer where the salt concentration and temperature are highest. The solar radiation that penetrates the pond’s upper layers reaches the LCZ and heats the highly concentrated brine. The LCZ can reach temperatures greater than 90 C and the useful heat can be used directly for low-temperature thermal applications (Rabl and Nielsen, 1975; Lu et al., 2004). SGSPs have been used previously to provide heat for desalination (Solis, 1999; Lu et al., 2001), and according to Mathioulakis et al. (2007), solar pond-powered desalination plants are amongst the most cost-effective alternative energy systems for desalinating water. The most notable work began in 1987 at a 3000-m2 solar pond in El Paso, Texas. At this site, a small multieffect, multi-stage flash distillation unit with a brine concentration and recovery system, and a 2.94-m2 air-gap MD (AGMD) unit were tested in conjunction with the solar pond to evaluate the long-term reliability of this technology. These two desalination units were chosen as they are usually operated with lowgrade heat and thus, are more suited to operate with the thermal energy generated in solar ponds (Lu et al., 2001). The multieffect, multi-stage unit provided an average water production of 3.3 L min1, which was equivalent to a water production of 1.6 103 m3 d1 per m2 of SGSP. This unit, however, required temperatures higher than 60 C to start the separation process, as well as large amounts of electricity to operate the system at approximately 30 kPa (Barron, 1992). Using AGMD, a maximum flux of 6.7 LMH (L m2 h1), i.e., a water production of 0.158 103 m3 d1 per m2 of SGSP, was achieved (Solis, 1999). This maximum water production was obtained with a temperature difference of 41 C across the membrane. The water production substantially decreased when the AGMD system was operated with lower temperature difference across the membrane. Other configurations of MD (e.g., DCMD) were not
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
tested at the El Paso Solar Pond. Because evaporation over the SGSP greatly exceeded water production, more water was lost than freshwater produced; however, this issue was not addressed in the El Paso studies. Evaporation is less of a problem when low-quality water is available to replenish the water evaporated. However, it is a problem for locations with limited water availability, such as in terminal lakes. Walker Lake is a desert terminal lake located in western Nevada, USA, which is fed by precipitation, the Walker River, and groundwater discharge (Beutel et al., 2001). The Walker River and groundwater carry dissolved solids to the Lake. As water evaporates, the dissolved solids concentrate inside the lake and increase the Lake’s total dissolved solids concentration. Human influences on the water budget of Walker River (water diverted for agriculture) combined with a dry climate have caused a decrease in the inflow to the lake, lowering the water level approximately 44 m in the last century. This drop resulted in an 80% decrease in the Lake’s volume from 36,422 to 7284 106 m3 and has contributed to a rise in total dissolved solids concentration from 3 to greater than 17 g L1 (Lopes and Allander, 2009). These salinity levels are dangerous to the aquatic habitat and have resulted in loss of biological diversity (Dickerson and Vinyard, 1999; Beutel et al., 2001). This work evaluates the feasibility of coupling a DCMD module with an SGSP for sustainable freshwater production in an environment such as that at Walker Lake. The overall objective is to develop a heat and mass transport model for the coupled DCMD/SGSP system in order to maximize water production. The specific objectives of this study are to determine the useful heat that can be collected from the SGSP and the energy that is required for distilling the water that permeates through the membrane, the freshwater production that can be achieved by the DCMD system, and the necessary membrane surface area for different operating conditions of the DCMD/SGSP coupled system. The approach presented in this paper is general and is useful to any terminal lake or even to other inland or coastal desalination applications. The major advantages of this system are that renewable thermal energy is used so that little electrical energy is required, the coupled system requires relatively low maintenance, and the terminal lake provides a source of salts to create the stratification in the SGSP.
2.
Theoretical approach
2.1.
Membrane module, membrane and feed solution
A membrane module that has symmetric channels on both sides of the membrane was used to theoretically evaluate the performance of the DCMD system. Each flow channel is 50 mm wide, 3 mm high, and 200 mm long. The small crosssectional area of the channels allows operation of the system at higher Reynolds numbers while maintaining low-pressure drop along the channels. The membrane module operates with countercurrent flow (see Fig. 1) to improve energy efficiency (Cath et al., 2004; Martinetti et al., 2009). A polytetrafluoroethylene (PTFE) membrane with pore size of 0.45 mm, porosity of 89%, thickness of 77 mm, and effective
4603
thermal conductivity of 0.027 W m1 C1 (Izquierdo-Gil et al., 1999) was used to theoretically evaluate the performance of the DCMD system. The combination of large pore size and porosity and low effective thermal conductivity results in a more efficient distribution of the heat used to produce the distillation across the membrane (Cath et al., 2004; Martinez and Rodriguez-Maroto, 2006). The feed solution consisted of water from the UCZ of the SGSP. The total dissolved solids concentration of this feed solution depends on diffusion from the LCZ and is typically between 50 and 75 g L1. To estimate the activity coefficients and the molar fraction of the solute in the feed solution of the DCMD module, NaCl was assumed to be the major constituent of the total dissolved solids in the lake and in the SGSP. Previous studies showed that approximately 57% of the dissolved solids in Walker Lake are sodium and chloride, and the remainder include calcium, magnesium, and sulfates (Cooper and Koch, 1984; Beutel et al., 2001).
2.2.
DCMD heat and mass transfer model
The simultaneous heat and mass transfer in a DCMD module is illustrated in Fig. 1. Assuming that the pores of the membrane surface are at liquidevapor equilibrium, the water flux, J (kg m2 s1), across the hydrophobic membrane can be expressed as (Schofield et al., 1990): J ¼ Cm pfm ppm (1) ¼ Cm p0 Tfm 1 c Sfm x Tfm ; Sfm p0 Tpm where Cm (kg m2 s1 Pa1) is the distillation coefficient of the membrane, p (Pa) is the vapor pressure, T ( C) is the temperature, p0(T ) (Pa) is the vapor pressure of the pure substance at a temperature T, cðSÞ () is the mole fraction of the solute at a concentration S (% w/w), and x(T,S ) () is the activity coefficient at a temperature T and at a solute concentration S, which can be estimated from empirical correlations (Curcio and Drioli, 2005). The subscripts fm and pm represent the feed and permeate sides of the membrane surface, respectively. The Antoine equation can be used to estimate p0(T ) (Yun et al., 2006). The inclusion of the activity coefficient and the mole fraction of the solute at the feed side of the membrane takes into account the reduction in vapor pressure due to the presence of a nonvolatile solute. In addition, the presence of solute in the feed side changes the fluid dynamics and heat transfer because it affects both the density and viscosity, as well as the thermal conductivity and heat capacity of the feed solution (Schofield et al., 1990). The vapor transport through the membrane pores typically occurs by combined Knudsen and molecular diffusion mechanisms, and the distillation coefficient, Cm, can be estimated by (Martinez and Rodriguez-Maroto, 2006): Cm ¼
1 f M 1 pa þ sd RT DK PDwa
(2)
where f (), s (), and d (m) are the porosity, tortuosity, and thickness of the membrane, respectively; M (kg mol1) is the molecular weight of water; R (J C1 mol1) is the gas constant; DK (m2 s1) is the Knudsen diffusion coefficient; Dwa (m2 s1) is the diffusion coefficient of water vapor in air; pa (Pa) is the
4604
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
partial pressure of air entrapped in the pores; and P (Pa) is the total pressure within the pores. The Knudsen diffusion coefficient, DK, is given by: 1=2 2 8 RT DK ¼ r 3 p M
PDwa ¼ 4:46 106 T2:334
(4)
where the average temperature inside the pores was used (i.e., T ¼ (Tfm þ Tpm)/2). In order to use Equation (1) to estimate the water flux through the membrane, the solute concentration, Sfm, and the temperatures at the surfaces of the membrane, Tfm and Tpm, must be determined. Using film theory and assuming 100% solute rejection by the membrane, Sfm can be estimated from a mass balance in the concentration boundary layer given by (Yun et al., 2006): ! J (5) Sfm ¼ Sf exp rf K where Sf (%) is the solute concentration in the bulk feed, rf (kg m3) is the density of the feed solution, and K (m s1) is the film mass transfer coefficient. To estimate Tfm and Tpm, a steady-state heat transfer analysis yields: qf ¼ qm ¼ qp
(6)
where qf and qp (W m2) are the convective heat transfer in the feed and permeate, respectively; and qm (W m2) is the heat transferred across the membrane. These heat fluxes are given by: qf ¼ hf Tf Tfm (7) qp ¼ hp Tpm Tp
(8)
qm ¼ qcm þ qvm ¼
km
dT km JHv ðTÞ DTm þ þ ½JHv ðTÞ ¼ d dx DTm
¼ ðhc þ hv ÞDTm
(9) 2
1
C ) are the heat transfer coefficients where hf and hp (W m in the feed and permeate sides, respectively; hc (W m2 C1) is the heat transfer coefficient for conduction across the membrane (a heat loss); and hv (W m2 C1) is the heat transfer coefficient for the vapor flow across the membrane; Tf and Tp ( C) are the bulk temperatures in the feed and permeate, respectively; qcm (W m2) is the conductive heat flux through the membrane; qvm (W m2) is the enthalpy heat flux through the membrane; km (W m1 C1) is the effective thermal conductivity of the membrane; x (m) is the distance in the mass flux direction; Hv ðTÞ (J kg1) is the vapor enthalpy (latent heat of vaporization) at temperature T ( C); and DTm ( C) is the difference between Tfm and Tpm. The heat transfer coefficients in the feed and permeate sides (i.e., hf and hp), as well as the mass transfer coefficient (i.e., K ) can be estimated using empirical correlations for different flow regimes. Larger water fluxes across the membrane are achieved when the flow regime in the feed and permeate channels is turbulent (Cath et al., 2004). Therefore, our work was focused in this flow regime, and the following correlations were used (Gnielinski, 1976; Phattaranawik et al., 2003):
# "
ðf =8Þ Ref;p 1000 Prf;p kf;p 6dh 1þ L 1 þ 12:7ðf =8Þ1=2 Pr2=3 1 dh f;p
(10)
(3)
where r (m) is the mean pore radius. The value of PDwa is given by:
hf;p
kf;p ¼ Nuf;p ¼ dh
K ¼ Shf
Scf ¼
Df ¼ dh
mf rf Df
"
1þ
Ref;p ¼
# ðf =8Þ Ref 1000 Scf 6dh D f L 1 þ 12:7ðf =8Þ1=2 Sc2=3 1 dh f
(11)
rf;p vf;p dh mf;p
(12)
Prf;p ¼
Cpf;p mf;p kf;p
where Nu, Sh, Sc, Re, and Pr () are the Nusselt, Sherwood, Schmidt, Reynolds, and Prandtl numbers, respectively; dh and L (m) are the hydraulic diameter and length of the channels in the membrane module, respectively; f () is the friction factor; Df (m2 s1) is the diffusion coefficient of the solute; k (W m1 C1), r (kg m3), m (Pa s), and Cp (J kg1 C1) are the thermal conductivity, density, dynamic viscosity, and specific heat capacity of the liquid streams, respectively; and v (m s1) is the velocity in the channels of the membrane module. The subscripts “f” and “p” represent the feed and permeate sides, respectively. The heat transfer correlations presented above are valid for Re between 3 103 and 106 (Mills, 1999). Using the resistance-electrical analogy (Fig. 1), Tfm and Tpm can be found using the following equations (Schofield et al., 1990): Tfm ¼ Tf
Tf Tp h1 f
1
1 h1 f þ hp þ ðhc þ hv Þ
Tpm ¼ Tp þ
Tf Tp h1 p 1
1 h1 f þ hp þ ðhc þ hv Þ
(13)
(14)
As Tfm and Tpm are both included in hv, an iterative solution is required. For the calculation procedure, Tfm and Tpm are assumed to be equal to Tf and Tp, respectively. With these temperatures, and assuming xðTfm ; Sfm Þ ¼ xðTf ; Sf Þ and cðSfm Þ ¼ cðSf Þ the water flux is estimated using equation (1). Then hf, hp, and K are evaluated using equations (10) and (11); a new value of Sfm is calculated using equation (5), and hc and hv are evaluated as shown in equation (9). The new values of Tfm and Tpm are calculated using equations (13) and (14). This iterative process is repeated until the rate of change is less than 0.001%. Using this heat and mass transfer model for the DCMD module, the operational parameters of the DCMD module (i.e., feed and permeate velocity, and partial pressure of air entrapped in the pores) were determined. Determination of the operational parameters was performed for an assumed permeate temperature of 20 C and feed temperatures of 40 and 60 C (Cath et al., 2004; Martinetti et al., 2009). The NaCl feed concentrations ranged from 0 to 25%. After determination of operational parameters, the model was used to evaluate the freshwater and heat flux across the membrane.
2.3.
SGSP thermal model
To evaluate the thermal performance of an SGSP, meteorological data from Walker Lake (Walker, NV) were used. The data from Walker Lake were collected and recorded by the U.S.G.S. (http://nevada.usgs.gov/walker/walkerlakeet.htm). Net radiation (W m2) and wind speed (m s1) were both measured at
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
2.5 m above the Lake surface; air temperature ( C) and relative humidity (%) were both measured at 1.5 and 2.5 m above the Lake surface; and Lake temperatures were measured at the surface and at 1 m, 2 m, and 3 m depths. As short-wave radiation heats up the entire water column within a water body and long-wave radiation is mainly reflected or absorbed in the first centimeters of water (Mobley, 1994), a differentiation between short- and long-wave radiation is necessary. Using the meteorological data (i.e., net radiation, air temperature, water temperature, relative humidity, and wind speed), incoming, outgoing, and net long-wave radiation were estimated using the description of the surface energy balance for lakes and reservoirs presented by Henderson-Sellers (1986). The net long-wave radiation was subtracted from the total net radiation to obtain the net short-wave radiation. A simplified one-dimensional steady-state thermal model was developed to estimate the energy collected by the SGSP. The assumptions of this model are: (1) stationary interfaces between the internal zones of the SGSP and a stable configuration (i.e., salt diffusion is negligible or is being controlled and the density gradient in the NCZ is always high enough to suppress convection); (2) complete mixing in the convective zones, i.e., uniform temperatures and concentrations in the UCZ and LCZ; (3) constant thermal conductivity, k (W m1 C1); and (4) insulated bottom and sides of the pond, hence all the solar energy that reaches the LCZ is absorbed in this zone. While a more sophisticated model of the dynamics is possible (Sua´rez et al., 2010), a simple model that assumes SGSP management was chosen in order to obtain a semianalytical solution. The NCZ was modeled using the energy equation for pure conduction and internal heat generation: d2 T 1 ¼ q000 ðzÞ dz2 k
(15)
where T ( C) is the temperature and z (m) is the depth within the SGSP. q000 ðzÞ (W m3) is the internal heat generation due to solar radiation absorption: dq00 ðzÞ q000 ðzÞ ¼ dz
(16)
where q00 ðzÞ (W m2) is the short-wave solar radiation heat flux at a depth z within the SGSP, which is represented using a simplified four-term series fit that parses the full spectrum into bandwidths with different extinction coefficients (Rabl and Nielsen, 1975): q00 ðzÞ ¼ q00 ð0Þ
4 X
Si expfli zg
(17)
i¼1
where q ð0Þ (W m2) is the short-wave radiation that penetrates the airewater interface, Si () is the fraction of energy contained in the ith bandwidth, and li (m1) is the composite attenuation coefficient of the ith bandwidth. The temperature distribution inside the NCZ can be found by integrating equation (15) and using the temperature of the UCZ and LCZ, TU and TL ( C), respectively, as boundary conditions: 00
zL z z zU ½FðzU Þ ðTL TU Þ þ FðzL Þ FðzÞ zL zU zL zU
TðzÞ ¼ TL þ
(18)
FðzÞ ¼
4 q00 ð0Þ X Si expfli zg k i¼1 mi
4605
(19)
where zU and zL (m) are the depths of the UCZeNCZ and LCZeNCZ interfaces, respectively (Fig.2). The conductive heat transfer within the NCZ, q00C ðzÞ (W m2), can be expressed as: dT k q00C ðzÞ ¼ k ¼ ½FðzU Þ FðzL Þ ðTL TU Þ q00 ðzÞ dz zL zU
(20)
for zU z zL. The temperature in the UCZ (i.e., at the water surface) can be estimated using an energy balance over the entire SGSP: 00 00 00 ! qS þ! q USE ¼ 0 q ð0Þ þ !
q00USE
(21)
2
(W m ) is the useful heat that can be extracted where from the SGSP, and q00S (W m2) is the heat flux across the surface of the SGSP, which includes evaporative heat flux, long-wave radiation, and sensible heat flux; therefore, it is a function of the air temperature, water surface temperature (i.e., TU), wind speed, relative humidity and, to a lesser extent, the cloud fraction. The evaporative or latent heat flux, q00E (W m2), across the water surface includes the wind-forced convection, q00F (W m2), and natural (or free) convection, q00N (W m2), under an unstable atmosphere (Adams et al., 1990): jq00E j ¼
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðq00F Þ2 þðq00N Þ2
(22)
The complete surface heat flux description (including longwave radiation and sensible heat) can be reviewed elsewhere (Henderson-Sellers, 1986). The temperature in the LCZ can be estimated using an energy balance in the LCZ: 00 00 00 ! q C ðzL Þ þ ! q ðzL Þ ¼ 0 q USE þ !
q00C ðzL Þ
(23)
2
(W m ) is the conductive heat loss to the NCZ, where and q00 ðzL Þ (W m2) is the solar energy absorbed in the LCZ. To estimate the temperature profile within the SGSP, the air temperature is used as the initial guess of TU. Equation (21) is solved iteratively using the secant method to obtain the correct value of TU. After this, TL is estimated using equation (23) and the temperature within the NCZ is evaluated using equation (18). The evaporation rate from the SGSP, ERSGSP (m s1), is given by: ERSGSP ¼
jq00E j Hv ðTU Þr
(24)
where Hv ðTÞ (J kg1) is the latent heat of vaporization (or vapor enthalpy) and r (kg m3) is the density of water. The simplified SGSP thermal model was used to determine the optimum positions of zU and zL for an SGSP at Walker Lake. These optimum positions are defined as the positions that, for a given operating TL, maximize the useful heat that can be extracted from the SGSP ðq00USE Þ. As the useful heat will be used in a DCMD module that is typically operated with temperature differences of 20 or 40 C between the feed and permeate sides (Cath et al., 2004; Martinetti et al., 2009), and the permeate temperature was selected as 13.9 C (explained later), temperatures of 33.9 and 53.9 C were used to define the optimum positions of zU and zL. Hereafter, these operating temperatures will be referred to as “typical operating
4606
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
conditions”. The model was also used to estimate the performance of the coupled system and to study the effect of heat extraction on evaporation rate in the SGSP.
2.4.
Water production for terminal lakes reclamation
After selection of the operational parameters for the DCMD module (e.g., feed and permeate velocity, partial pressure of air entrapped in the pores) and design of the SGSP (e.g., surface area, thickness of each zone), the performance of the coupled DCMD/SGSP system was evaluated. Specifically, useful heat from the SGSP, freshwater production, and energy required for distilling the water that permeates through the membrane were determined. The necessary membrane surface area to use all the energy collected in the SGSP was also determined. In addition, the meteorological data for the study site were used to estimate the water loss due to evaporation that occurs in the SGSP. When the heat extracted from the SGSP is used without losses, the required membrane area for the DCMD module, ADCMD (m2), can be found by equating the energy collected in the SGSP with the energy used in the DCMD module. Thus: ASGSP q00USE
¼ ADCMD qm
(25)
where ASGSP (m2) is the surface area of the SGSP. The water flow produced in the DCMD module, QW (m3 s1), is given by: QW ¼
J ADCMD r
(26)
As a first approach, it was assumed that the SGSP is constructed outside of the terminal lake (Fig. 3a). Constructing a pond outside a water body is simpler and less expensive than constructing it inside a water body. However, the surface area over which evaporation occurs increases because the water evaporates from both the terminal lake and the SGSP. This results in an additional water loss, QE (m3 s1), which is given by: QE ¼ ERSGSP ASGSP
(27) 1
where ERSGSP (m s ) is the evaporation rate that occurs in the SGSP. The total water production, when evaporation is considered, is defined in this work as “net water production”, QN (m3 s1), and is given by: 00 J qUSE ERSGSP ASGSP QN ¼ QW QE ¼ (28) r qm Fig. 3c shows the coupling of the DCMD module to the SGSP. The feed solution to the membrane module was water taken from the UCZ. The concentration of this feed solution depends on diffusion from the LCZ and typically is less than 7.5% w/w (w75 g L1). The heat extracted from the LCZ was transferred to the feed solution using an ideal heat exchanger (i.e., 100% effectiveness), and it was assumed there were no heat losses in the coupled system, thus, the average feed temperature was assumed to be equal to the temperature in the LCZ, i.e., Tf ¼ TL. To maintain a colder temperature in the permeate solution, a second heat exchanger was connected to the lake (not shown in Fig. 3a). In this way, the average permeate temperature was equal to the temperature at a depth of 3 m in the lake (i.e., 13.9 C).
The selection of this temperature is considered conservative since hypolimnetic temperatures in Walker Lake ranged from 6 to 8 C in the winter and from 10 to 12 C in the summer (Beutel et al., 2001). Water from the lake was used to replenish the water loss due to evaporation and the water cleaned in the DCMD module. The concentrate that is produced is then returned to the LCZ to account for salt diffusion, and the excess of salts can be disposed of or sold if it has commercial value. Thus, the SGSP was expected to have stationary interfaces between the internal zones and maintain a stable configuration (Sua´rez et al., 2010). In this configuration, the DCMD module treats a feed solution that is slightly more concentrated than the water from the lake. If water from the terminal lake is cleaned directly and the UCZ is not treated, this zone will be concentrated by diffusion from the LCZ and by evaporation to the atmosphere. Therefore, configurations that do not treat the water from the UCZ will not be sustainable over time. Considering that the UCZ is a completely mixed zone, and that the salts carried out by evaporation are negligible, a steady-state salt budget yields (Fig. 3c): SL r SU rU ASGSP ¼ QW SU rU ðQW þ QE ÞSLAKE rLAKE þ D L zL zU
(29)
where SLAKE, SU and SL (%) are the salinities of the terminal lake, UCZ and LCZ, respectively; rLAKE, rU and rL (kg m3) are the densities of the terminal lake, UCZ and LCZ, respectively; and D (m2 s1) is the salt diffusivity. The salinity of the UCZ, SU, can be found by solving equation (29). SU will be the feed concentration, Sf, to the DCMD module (i.e., Sf ¼ SU). As the DCMD module and the SGSP are connected by a heat exchanger, maintenance of each system is independent and thus, relatively simple. The performance of the DCMD/SGSP coupled system can be evaluated by an iterative process. First, the operating temperature and salinity of the LCZ, TL and SL (%), respectively, are selected. The SGSP thermal model is then used to estimate the temperature in the UCZ (TU), the useful heat that can be extracted from the SGSP ðq00USE Þ, and the evaporation rate from the SGSP (ERSGSP). As a first approximation, the feed concentration to the DCMD module, Sf, is considered equal to the lake concentration, i.e., SLAKE ¼ 1.7%. The water flux across the membrane is estimated using this feed concentration and assuming an ideal heat exchanger, i.e., Tf ¼ TL. The membrane area and the freshwater flow are estimated using equations (25) and (26), respectively. Then equation (29) is used to estimate SU. This salinity is compared against the feed concentration. If they differ, SU is used as the new feed concentration and the procedure is repeated until these values agree to less than 0.1%.
3.
Results and discussion
3.1.
DCMD heat and mass transfer model
The effect of the velocity in the feed and permeate channels, vf and vp, respectively, as well as the partial pressure of air entrapped in the pores of the membrane, pa, on the performance of the DCMD module is presented in Fig. 4. A permeate temperature of 20 C and feed temperatures of 40 and 60 C
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
4607
Fig. 3 e (a) SGSP constructed outside the terminal lake. (b) SGSP constructed inside the terminal lake. (c) Coupling of the DCMD module to the SGSP. Water from the UCZ is used as feed solution. A heat exchanger is used to pass the energy collected in the SGSP to the DCMD module.
were utilized. A feed concentration of 1.7% (w17 g L1), which is approximately the total dissolved solids concentration in Walker Lake (Lopes and Allander, 2009), was used. When the velocities in the feed and permeate channels were increased (Fig. 4a and b), the water and heat fluxes increased. The
increase in these fluxes is non linear and approaches an asymptotic value. This occurs because turbulence is enhanced, producing more mixing in the channels and reducing the thermal boundary layer thickness (Mills, 1999; Cath et al., 2004). Consequently, Tfm and Tpm approach the
4608
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
30 J
25
qm
20
qvm
15 10
Tf = 40 °C
qcm
5 0
120
pa = 100 kPa Sf = 1.7%
b 100
-2
pa = 100 kPa Sf = 1.7%
-1
a
35
Water Flux, J [kg m hr ] -2 Heat Flux, q [kW m ]
Water Flux, J [kg m-2hr-1 ] Heat Flux, q [kW m -2 ]
40
1.0
1.5
2.0
2.5
J 80
qm
60
qvm
40
Tf = 60 °C
20 0 1.0
3.0
1.5
Velocity in the channels [m s -1 ]
J 25 qm 20 qvm
15 10
Tf = 40 °C
qcm
5 50
60
70
80
100
J
90
80
qm
60
qvm
40
Tf = 60 °C
20 0 50
100
qcm
60
Partial pressure of air entrapped in pores, pa [kPa]
70
20
25
pa = 100 kPa
vf = vp = 2.0 m s
-1
e
35
90
100
NaCl concentration [%]
15
0
5
10
vf = vp = 2.0 m s
f
100
15
20
25
pa = 100 kPa
-1
-2
30
120
-1
10
Water Flux, J [kg m hr ] -2 Heat Flux, q [kW m ]
Water Flux, J [kg m-2hr-1 ] Heat Flux, q [kW m -2 ]
40
5
80
Partial pressure of air entrapped in pores, pa [kPa]
NaCl concentration [%] 0
3.0
vf = vp = 2.0 m s -1 Sf = 1.7%
d
-2
30
0
2.5
120
-1
vf = vp = 2.0 m s-1 Sf = 1.7%
c
35
2.0
Velocity in the channels [m s-1 ]
Water Flux, J [kg m hr ] -2 Heat Flux, q [kW m ]
Water Flux, J [kg m-2hr-1 ] Heat Flux, q [kW m -2 ]
40
qcm
25
J
20
qm
15
qvm
10
qcm
5
Tf = 40 °C
0 0
25
50
J qm
80 60
qvm 40 qcm
20
Tf = 60 °C
0 75
100
125
150
175
200
225
250
275
-1
NaCl concentration [g L ]
0
25
50
75
100
125
150
175
200
225
250
275
NaCl concentration [g L- 1]
Fig. 4 e Performance of the DCMD module. (a) Effect of the velocity in the feed and permeate channels (vf and vp) when the feed temperature (Tf) is 40 C. (b) Effect of vf and vp when Tf is 60 C. (c) Effect of the partial pressure of air entrapped in pores ( pa) when Tf is 40 C. (d) Effect of pa when Tf is 60 C. (e) Effect of the feed (NaCl) concentration (Sf) when Tf is 40 C. (f) Effect of Sf when Tf is 60 C.
bulk temperatures, Tf and Tp, respectively, maximizing the temperature and vapor pressure differences across the membrane. This increases the driving force as well as the conductive heat flux across the membrane. For channel velocities greater than 2.0 m s1 (Ref > 16,500 and Rep > 9600), the water flux only increases slightly, thus, 2.0 m s1 was selected as the operating velocity for both the feed and permeate channels in the DCMD module. Velocities of this magnitude have been used in previous experimental studies (Phattaranawik et al., 2003; Cath et al., 2004). When pa is decreased (Fig. 4c and d), the water and heat fluxes are increased because molecular diffusion is enhanced, resulting in a higher water flux, even when the temperature difference
remains constant. The increased water flux also increases convective heat flux because more water vapor crosses the membrane. Similar behavior has been reported in previous numerical and experimental investigations (Cath et al., 2004; Yun et al., 2006; Safavi and Mohammadi, 2009). When pa is decreased from 100 to 50 kPa, the water flux increases by approximately 50%. Even though a reduction in the partial pressure of air entrapped in the pores, pa, results in a considerable increase in water flux, it also increases the energy consumption to create the vacuum inside the membrane module. For this reason, a pa of 80 kPa was selected as the operating pressure as this vacuum is not expected to require a substantially higher energy investment (Cath et al., 2004).
4609
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
ZU = 0.01 m
ZU = 0.05 m
ZU = 0.10 m
0.25
0.25
0.50
0.50
0.75
0.75 1.00
Depth of the LCZ-NCZ interface, ZL [m]
1.00 1.25
1.50
1.75
1.75
2.00
2.00
2.50 20
a
2.25
Tf = 33.9 °C 30
40
50
60
70
80
90
0.25
0.50
0.50
0.75
0.75
1.00
1.00
1.25
1.25 1.75
2.00
2.00
2.50 20
c
2.25
Tf = 53.9 °C 30
40
50
b
Tf = 33.9 °C 2.5
3.0
3.5
4.0
4.5
5.0
increasing ZU
1.50
increasing ZU
1.75 2.25
increasing ZU
2.50 2.0
0.25
1.50
ZU = 0.20 m
1.25
increasing ZU
1.50
2.25
ZU = 0.15 m
60
70
80
-2
Useful heat, quse [W m ]
90
d
2.50 2.0
Tf = 53.9 °C 2.5
3.0
3.5
4.0
4.5
5.0
-1
Evaporation rate [mm d ]
Fig. 5 e Optimization of the positions of the UCZeNCZ and LCZeNCZ interfaces. (a) Useful heat when Tf [ 33.9 C. (b) Evaporation rate from the SGSP when Tf [ 33.9 C. (c) Useful heat when Tf [ 53.9 C. (d) Evaporation rate from the SGSP when Tf [ 53.9 C. When zU [ 0.15 m and Tf [ 33.9 C, maximization of the useful heat and minimization of the evaporation rate occur at zL [ 1.33 m. When zU [ 0.15 m and Tf [ 53.9 C, maximization of the useful heat and minimization of the evaporation rate occur at zL [ 2.14 m.
A pa of 80 kPa increases the permeate flux by approximately 15% when compared to a pressure of 100 kPa (atmospheric pressure). In addition, operating the membrane module at near atmospheric pressure reduces the risk of wetting the membrane pores (Solis, 1999). The effect of the feed (NaCl) concentration on the performance of the DCMD module is shown in Fig. 4e and f. As the feed concentration increases, flux decreases. This is due to decreased vapor pressure on the feed side because of the presence of a non-volatile solute. For NaCl concentrations less than 5% (w60 g L1) and for both temperatures, the water flux reduction (compared to the case when the feed solution is pure water) is less than 5%. For NaCl concentrations of 25% (w250 g L1) at 40 C, the water flux reduction is approximately 35%.
3.2.
SGSP thermal model
The optimal design of the SGSP, which maximizes the useful heat, provides a balance between the conductive heat losses across the NCZ and the energy collected in the LCZ. The effect of the positions of zU and zL on the useful heat that can be extracted and the evaporation rate over the SGSP is presented in Fig. 5. The zU and zL were varied from 0.01 to 0.20 m and from 0.25 to 2.5 m, respectively. For operating temperatures in the LCZ, TL, of 33.9 and 53.9 C, it was found that when zU
increases, the useful heat decreases (Fig.5a and c) and the evaporation rate increases (Fig. 5b and d). This occurs because an increase in the UCZ thickness e or zU typically results in a decrease of the NCZ thickness. This increases the conductive heat losses across the NCZ and increases the temperature in the UCZ. Therefore, the amount of heat available in the LCZ is reduced and evaporation is enhanced. On the other hand, when zL increases, the NCZ thickness increases. This results in a large insulating layer and thus, greater storage of heat in the LCZ and less conductive heat losses that would increase the temperature in the UCZ. However, the increased NCZ thickness will also enable less solar radiation to reach the LCZ and less energy to be collected and stored by the system. Consequently, as zL increases the useful heat will also increase, but only until an optimum depth is reached; below which, the useful heat decreases. The opposite behavior is observed for the evaporation rate: as zL increases, the evaporation rate decreases until the optimum depth is reached. After this, the evaporation rate slowly increases. When the temperature in the LCZ is set to 33.9 C the optimal depths for zU and zL were found to be 0.01 and 0.94 m, respectively. In this configuration, the useful heat and the evaporation rate are 72.1 W m2 and 2.24 mm d1, respectively (Fig.5a and b). However, in practice, it is not easy to maintain such a thin UCZ because of wind mixing or heat exchange that
4610
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
Table 1 e Water production and energy requirements of the DCMD/SGSP coupled system per m2 of SGSP when operated under typical conditions. Tp [ 13.9 C, pa [ 80 kPa, vf [ vp [ 2.0 m sL1, zU [ 0.15 m, ERLAKE [ 3.58 mm dL1. The performance of the system is presented for different salinities in the LCZ. q00USE (W m2)
ER SGSP 103 (m3 d1)
J (kg m2 h1)
ADCMD 103 (m2)
QW 103 (m3 d1)
QN 103 (m3 d1)
qvm (kW m2)
qcm (kW m2)
10.41
66.2
2.371
20.5
3.272
1.606
14.5
5.8
33.9
10.41
66.2
2.371
20.3
3.291
1.603
14.3
5.8
6.6
33.9
10.41
66.2
2.371
20.1
3.312
1.598
14.2
5.8
25
7.4
33.9
10.41
66.2
2.371
19.9
3.336
1.594
14.1
5.8
2.14
10
4.9
53.9
10.74
58.1
2.555
69.5
0.970
1.617
49.5
10.5
2.14
15
5.3
53.9
10.74
58.1
2.555
69.2
0.973
1.615
49.3
10.5
2.14
20
5.8
53.9
10.74
58.1
2.555
68.8
0.977
1.614
49.0
10.5
2.14
25
6.3
53.9
10.74
58.1
2.555
68.5
0.981
1.612
0.765 a 2.815b 0.768 a 2.812b 0.773a 2.807b 0.777a 2.803b 0.938a 2.642b 0.940a 2.640b 0.941a 2.639b 0.943a 2.637b
48.8
10.5
ZL (m)
SL (%)
SU (%)
Tf ¼ TL ( C)
1.33
10
5.0
33.9
1.33
15
5.8
1.33
20
1.33
TU ( C)
a Net water production when the SGSP is constructed outside of the terminal lake. b Net water production when the SGSP is constructed inside of the terminal lake.
occurs at the surface. For this reason, the zU was set to 0.15 m to be easier to maintain, yet still provide useful heat extraction and relatively low evaporation rates. In this case, the optimum depth for zL was found to be 1.33 m. This configuration can provide 66.2 W m2 of useful heat (34% efficiency), approximately 8% less than the optimal configuration. The evaporation rate for this configuration is 2.37 mm d1, which corresponds to an increase of 6% compared to the configuration where zU ¼ 0.01 m. When 53.9 C is used as the temperature in the LCZ and zU ¼ 0.15 m, the optimum depth for zL was found to be 2.14 m. This configuration provides 58.1 W m2 as useful heat (30% efficiency), and the evaporation rate under this configuration is 2.56 mm d1 (Fig. 5b and d). A higher efficiency of the SGSP is observed when the temperature in the LCZ is set to 33.9 C. This occurs because in a solar pond with lower temperatures, a large percentage of the collected heat can be extracted from the SGSP as useful heat, while in a solar pond with higher temperatures, more of the collected heat must be used to maintain the high temperature in the LCZ and only the remainder can be extracted as useful heat. As a comparison, a flat-plate thermal solar collector, with an overall heat loss coefficient of 2.0 W m2 C1, provides approximately 84.5 and 44.5 W m2 when used to warm water to 33.9 and 53.9 C, respectively (Duffie and Beckham, 2006). When the temperature in the LCZ is 53.9 C, the SGSP delivers more useful heat than this flat-plate thermal solar collector. On the other hand, for a temperature in the LCZ of 33.9 C, the flat-plate collector delivers more useful heat than the SGSP. However, a flat-plate collector cannot operate during the night and thus, an extra energy storage unit is required for 24-h operation. The total depth of the pond, for both of the previous situations, should be defined depending on the desired storage capabilities of the pond. Constructed ponds have depths that vary from 1.0 to 4.5 m (Hull et al., 1989). Total depths of 2.5 and 3.5 m, for temperatures in the LCZ of 33.9 and 53.9 C, respectively, should be sufficient to provide the necessary storage for
the desalination process (Lu et al., 2004). Because it was assumed that the SGSP is well insulated and that all the solar energy that reaches the LCZ is absorbed in this zone, the total depth does not affect the useful heat that can be extracted from the SGSP. However, this depth is important for construction costs and also to maintaining a desired temperature in the LCZ during the night (to allow continuous operation).
3.3.
Water production for terminal lakes reclamation
The performance of the coupled DCMD/SGSP system for the typical operating conditions, i.e., Tf ¼ 33.9 or 53.9 C, is summarized in Table 1. The performance is presented as a function of the salinity in the LCZ. At the higher feed temperature, the temperature difference across the DCMD membrane and the water flow (QW) are higher. As salinity in the LCZ increases, the vapor pressure on the feed side and the water flow decrease. Thus, as shown in Table 1, the highest water flow obtained in the DCMD module occurs when the feed side is at 53.9 C and the salinity in the LCZ is 10%. At these conditions, and when treating a feed solution with a salinity of 4.9%, the coupled system delivers 1.617 103 m3 d1 per m2 of SGSP with 0.970 103 m2 of membrane per m2 of SGSP. If lower salinities in the LCZ are used, the coupled system produces larger water flows. However, lower salinities in the LCZ would increase the risk of instabilities inside the SGSP. When the system is operated at Tf ¼ 53.9 C, a dynamic stability analysis in the SGSP reveals that the NCZ is stable when SL 10% (Sua´rez et al., 2010). Thus, this is the minimum operating salinity of the LCZ. As can be seen in Table 1, the flow produced by the coupled system is a relatively small flow rate that is on the same order of magnitude as evaporation occurring in desert conditions. Further, when the SGSP is constructed outside the terminal lake, and for typical operating conditions, the net water production (QN) becomes negative. This occurs because the
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
4611
Fig. 6 e (a) Maximization of the freshwater flow produced in the DCMD/SGSP coupled system. (b) Membrane area needed to maximize the freshwater flow. Tp [ 13.9 C (lake temperature), SL [ 10% (w100 g LL1), vf [ vp [ 2.0 m sL1, pa [ 80 kPa, and zU [ 0.15 m.
additional evaporation created by the SGSP is larger than the permeate flow from the DCMD module. Therefore, construction of the SGSP outside the terminal lake is not a logical solution to produce freshwater unless there is excess lowquality water that can be used to replenish the UCZ of the SGSP or if evaporation can be suppressed. As the water needed in the UCZ should have a relatively low salt concentration (e. g., typically between 0 and 10% (Kurt et al., 2000)) compared to the salt concentration in the LCZ (e.g., between 20 and 25%), brackish water or seawater (w3.5%) could be used for this purpose. However, methods to reduce evaporation should also be considered. The other alternative is to construct the SGSP inside the terminal lake (Fig. 3b). In this way, the surface area for evaporation is not increased and additional water loss is reduced. Additional water loss is given by: QE ¼ ðERSGSP ERLAKE ÞASGSP
(30)
where ERLAKE (m s1) is the evaporation rate from the lake. The net water production is given by: 00 J qUSE ðERSGSP ERLAKE Þ ASGSP QN ¼ QW QE ¼ (31) r qm For the study site of this investigation, due to heat extraction from the SGSP, the temperature in the UCZ is lower than the temperature at the surface of the terminal lake. This results in a smaller evaporation rate from the SGSP than from the terminal lake. Thus, extra water gain is obtained, and the water production exceeds the water loss due to evaporation. As shown in Table 1, when the SGSP is constructed inside the lake, the highest net water production occurs when the feed side is at 33.9 C. This is due to a smaller evaporation rate from the SGSP for this operating condition. At this temperature, and at a salinity of 10% in the LCZ, the net water production is 2.815 103 m3 d1 per m2 of SGSP. Although operating at 33.9 C results in a 7% higher net water production than that obtained at 53.9 C, the necessary membrane area is three times higher. Required membrane area is an important
parameter to determine because one of the current issues that MD faces for water desalination is how to maintain the desired temperature difference between the feed and permeate sides along the entire membrane (Criscuoli et al., 2008). If membrane areas are the limiting factor, then operating at higher temperatures may be more desirable since it requires less membrane area. For example, data from Table 1 shows that when the coupled system operates at Tf ¼ 53.9 C, slightly higher freshwater flow is produced than at Tf ¼ 33.9 C and only 30% of the membrane area is required. This strongly points to the need to improve membrane modules to reduce this loss of driving force across the membrane. The net water production values are an obvious improvement considering that when the SGSP is constructed outside the lake the net water production is negative. This also strongly points to the need to develop methods to suppress evaporation from solar ponds to increase the freshwater production of this coupled system. For instance, in the extreme case when evaporation from the SGSP is suppressed completely, the useful heat collected in the SGSP increases (e. g., from 66.2 to 68.7 W m2); thus, more energy is available for the distillation process, the freshwater production (QW) increases, and the net water production increases from 2.815 103 to 5.258 103 m3 d1 per m2 of SGSP, which corresponds to an increase of 87% compared to the net production obtained in the typical conditions.
3.3.1.
Maximization of water production
The parameters that affect the quantity of water produced from the DCMD module are the energy collected in the SGSP, which is dependent on the depth of the LCZeNCZ interface, zL, the operating feed temperature, Tf ¼ TL, and to a lesser extent, the feed concentration, Sf ¼ SU, which is dependent on the salinity of the LCZ, SL (as shown in Table 1). Fig. 6a shows the freshwater production, QW, as a function of TL and zL for SL ¼ 10%. Fig. 6b presents the corresponding membrane area needed to use all the heat collected in the SGSP. In this case, a maximum freshwater flow of 1.619 103 m3 d1 per
4612
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
m2 of SGSP occurs at TL ¼ 47.9 C, when the zL is at 1.9 m depth and the energy collected in the SGSP is 60.1 W m2. A membrane area of 1.328 103 m2 per m2 of SGSP is needed to obtain this flow rate. This maximum freshwater production is slightly larger than that produced when operating at TL ¼ 53.9 C (i.e., 1.617 103 m3 d1 per m2 of SGSP). However, a larger membrane area is needed for the lower temperature.
3.3.2. Sensitivity analysis, energetic inefficiencies, and additional energy requirements Operational parameters affect the performance of the coupled DCMD/SGSP system. As shown in Fig. 4, the velocity in the channels of the membrane module and the pressure of air entrapped in the pores of the membrane are important parameters that not only affect the water flux across the membrane, but also the energetic requirements of the coupled system. The effect of these parameters over the water flow and membrane area is presented in Fig. 7a and b. As the velocity in the channels increases, the water flux across the membrane also increases, and because the energy collected by the SGSP is limited, the membrane area needed to use all the collected energy decreases. This combination of increased water flux and reduced membrane area, results in an almost constant water flow for a wide range of channel velocities. When the channel velocity is increased from 1 to 3 m s1 the water flow does not change significantly and the membrane area decreases by w22% (Fig. 7a). Thus, higher pumping cost must be weighted against higher membrane cost. When the partial pressure of air entrapped in the pores of the membrane is reduced (Fig. 7b), the water flow is increased and the membrane area is reduced. When this pressure is reduced from 100 to 4 kPa (the lowest operating pressure reported for an MD module (Safavi and Mohammadi, 2009)), the water flow production increases by w24% and the membrane area decreases by w67%. However, the energy consumption to produce this vacuum is large and operating at this pressure provides greater risk of wetting the membrane pores (Solis, 1999). The results presented above were obtained for an ideal system operating without heat losses. The heat losses throughout the DCMD/SGSP coupled system could be divided into two main components: the effectiveness of the heat exchanger used to warm the feed solution and the conductive heat losses that occur across the system. If the effectiveness, 3 (%), is less than 100%, the temperature of the feed solution will be lower than the temperature in the LCZ. Using the definition of the effectiveness of a counterflow heat exchanger (Mills, 1999), the temperature at the exit of the cold side of the heat exchanger is (see heat exchanger detail in Fig. 3c): CpL ðTL TU Þ Tf ¼ TU þ 3 CpU
(32)
where TU and Tf ( C) are the temperatures at the inlet and exit of the cold side of the heat exchanger, respectively. These temperatures correspond to the UCZ and feed temperatures, respectively; TL ( C) is the temperature at the inlet of the hot side of the heat exchanger, which corresponds to the LCZ temperature; and CpL and CpU (J kg1 C1) are the heat
capacities of the hot and cold sides of the heat exchanger, respectively, and correspond to the heat capacities of the LCZ and UCZ. Thus, for the heat capacities observed at the UCZ and LCZ of solar ponds, the real effectiveness most likely will be approximately 90% (Mills, 1999). Fig. 7c shows that as the heat exchanger effectiveness decreases the water flow also decreases and the required membrane area increases. A heat exchanger with an effectiveness of 60% produces 90% of the water flow produced by an ideal heat exchanger and requires 135% more membrane area. Therefore, design of the heat exchanger to couple the DCMD module to the SGSP is very important. When 3 ¼ 90%, freshwater production of 1.581 103 m3 d1 per m2 of SGSP is achieved with a membrane area of 1.604 103 m2 per m2 of SGSP. This results in a net water production of 2.653 103 m3 d1 per m2 of SGSP. To account for conductive or other types of heat losses in the entire system, equation (25) can be modified using an arbitrary overall thermal efficiency factor, h (%):
ASGSP q00USE h ¼ ADCMD qm
(33)
This thermal efficiency factor reduces the energy that is transferred to the DCMD feed water, and thus allows the performance estimation of a non-ideal system. Fig. 7d shows the effect of the thermal efficiency factor on freshwater production when 3 ¼ 90%. As the thermal efficiency decreases, both water flow and membrane area decreases because less energy is available to drive the thermal desalination process. Therefore, a proper thermal insulation is required to minimize these heat losses. Additional energy requirements for the coupled DCMD/ SGSP system include the energy required to pump the solutions through the system and the energy required to reduce the partial pressure of air entrapped in the pores of the membrane ( pa). The energy required to circulate the solutions inside the system was estimated by assuming that frictional losses are the main resistance to flow. Frictional losses were estimated in the piping system as well as inside the membrane module. In addition, it was assumed that singular losses are approximately 20% of the frictional losses. Approximately 0.3 W is needed to circulate the solutions through the desalination system (i.e., 0.5% of the energy that can be extracted from the SGSP). The energy required to create a vacuum of less than 5 kPa in an MD module is on the order of 1 kWh per m3 of freshwater produced (Cabassud and Wirth, 2003). This corresponds to 0.1% of the energy that can be extracted from the SGSP. Thus, the energy required to reduce pa from atmospheric pressure to 80 kPa and the energy required to circulate the solutions are negligible.
3.3.3. Comparison of the coupled DCMD/SGSP system with other thermal desalination systems The performance of the coupled DCMD/SGSP system, for the conditions that maximize water production, was compared to a coupled AGMD/SGSP system, a DCMD module coupled to a flat-plate thermal solar collector, and a single-basin-type solar still system (Table 2). These thermal desalination systems were chosen as they all are solar-powered distillation systems. A theoretical model for an AGMD presented
4613
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
3.5 3.0 2.5
Water flow
2.0 1.4 1.5
ADCMD 1.2
1.0
a 1.0 1.0 1.2
1.4 1.6 1.8
2.0 2.2 2.4
2.6 2.8
0.5 3.0
2.5 1.6 2.0 1.4 1.2
1.5
ADCMD 1.2
1.0
c 80
85
90
95
0.5 100
Heat exchanger effectiveness, ε [%]
Water flow x 103 [m3 d-1]
Water flow x 103 [m3 d-1]
2.0
1.4
75
40
60
0.5 100
80
vf = vp = 2.0 m s -1 pa = 80 kPa
ε = 90%
1.8
3.5 3.0 2.5
1.6 2.0 1.4
Water flow
1.5
ADCMD 1.2 1.0 75
ADCMD x 103 [m 2 ]
Water flow
70
20
2.0
3.0 2.5
65
1.0
b 0
3.5
1.6
1.0 60
1.5
1.0
ADCMD x 103 [m 2 ]
1.8
ADCMD
Partial pressure of air entrapped in pores [kPa]
vf = vp = 2.0 m s -1 pa = 80 kPa
η = 100%
3.5 3.0
1.8
Velocity in the channels [m s-1] 2.0
vf = vp = 2.0 m s -1
Water flow
ADCMD x 103 [m2]
1.8 1.6
2.0
Water flow x 103 [m3 d-1]
pa = 80 kPa
ADCMD x 103 [m 2 ]
Water flow x 103 [m3 d-1]
2.0
1.0
d 80
85
90
0.5 100
95
Thermal Efficiency of the system, η [%]
Fig. 7 e Effect of operational parameters and energetic efficiencies over the water production and membrane area required for the DCMD/SGSP coupled system. (a) Velocity in the channels. (b) Partial pressure of air entrapped in pores. (c) Heat exchanger effectiveness. (d) Thermal efficiency of the system. Tp [ 13.9 C, SL [ 10% (w100 g LL1), zU [ 0.15 m, zL [ 1.9 m, TL [ 47.9
by Izquierdo-Gil et al. (1999) was used to evaluate the performance of a coupled AGMD/SGSP system. The operating conditions, membrane dimensions, and membrane properties of the coupled DCMD/SGSP system were utilized in the
coupled AGMD/SGSP system. The air gap of the AGMD module was assumed to be 0.9 mm (Izquierdo-Gil et al., 1999). As shown in Table 2, the freshwater flow produced in the coupled DCMD/SGSP system is four times larger than that
Table 2 e Comparison of the DCMD/SGSP coupled system with other desalination systems. Water production and energy requirements are presented per m2 of solar collector. Tp [ 13.9 C, pa [ 80 kPa, vp [ 2.0 m sL1, zU [ 0.15 m, zL [ 1.9 m, TL [ 47.9 C, SL [ 10%, 3 [ h [ 100%. Solar collector Membrane module q00USE (W m2) Tf( C) vf(m s1) Sf(%) qvm (kW m2) qcm (kW m2) qvm (kW m2) J (kg m2 h1) A membrane 103 (m2) QW 103 (m3 d1)
SGSP DCMD 60.1 47.9 2.0 4.9 36.1 9.2 45.3 50.8 1.328 1.619
a A solar still does not use a membrane to distill water. b Energy collected in the flat-plate thermal solar collector. c Average temperature of the water inside the solar still.
AGMD 60.1 47.9 2.0 4.9 9.2 0.8 10.0 12.9 1.328 0.412
Thermal solar collector
Solar stilla
DCMD
e
b
56.5 47.9 3.0 103 1.7 7.5 3.7 11.2 10.7 1.328 0.340
58.3 29.1c e e e e e e e 2.060
4614
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
produced in the coupled AGMD/SGSP system. Even though the AGMD system reduces considerably the conductive heat losses across the membrane ðqcm Þ, there is an additional mass transfer resistance produced by the air gap, which results in substantially lower water and enthalpy fluxes through the membrane when compared to the DCMD system. The flatplate thermal solar collector (described before) was used to determine the water flow that can be warmed from the Lake temperature to 47.9 C. For this condition, the flat-plate solar collector delivers less useful heat than the SGSP. This occurs because the heat losses from the flat-plate solar collector to the surroundings are approximately five times larger than the conductive heat losses in the NCZ of the SGSP. Using the water flow warmed in the flat-plate solar collector and the dimensions of the DCMD module, the velocity in the feed channel was estimated. Assuming that the average feed temperature in the DCMD module was 47.9 C, the DCMD heat and mass transfer model was utilized to evaluate the freshwater production. A freshwater production five times smaller than that produced in the coupled DCMD/SGSP was achieved. This lower water production is a result of the low velocity of the feed solution inside the membrane module. Note that, for the previous assessment, correlations for laminar flow were used to estimate the heat and mass transfer coefficients (Mills, 1999; Phattaranawik et al., 2003). A solar still that absorbs all the energy reaching the water surface inside the collector, loses 58.3 W m2 due to evaporation, which results in a water production that is 27% larger than that of the coupled DCMD/SGSP system. Note that solar stills do not use membranes to distill the water; instead they use their cover to condense the evaporated water. However, solar stills are typically more expensive to construct than SGSPs and they require more maintenance and more careful design to avoid vapor leakage (Duffie and Beckham, 2006; Cipollina et al., 2009).
4.
Conclusions
The coupled DCMD/SGSP system is capable of providing freshwater for terminal lakes reclamation and is a sustainable solution for the global need for inexpensive clean water and low-cost energy. In this paper, a heat and mass transport model for the coupled DCMD/SGSP system was used to analyze the effect of operating conditions on the performance of the coupled system. It was found that the coupled system produces water flows on the order of 1.6 103 m3 d1 per m2 of SGSP with membrane areas ranging from 1.0 to 1.3 103 m2 per m2 of SGSP. For the study site of this investigation, the coupled system will only be feasible when the SGSP is constructed inside the terminal lake. In this manner, the net water production is on the order of 2.6e2.8 103 m3 d1 per m2 of SGSP. It is important to note that the cost of maintaining this system is small, since it uses renewable thermal energy to drive the desalination process. More research on suppression of evaporation from solar ponds and maintaining the temperature difference along the entire membrane in the MD module is needed to make the coupled system more efficient.
Acknowledgments The authors wish to thank the Department of Energy for funding Grant No. DE-FG02-05ER64143. The authors greatly appreciate the helpful comments and constructive suggestions of the two anonymous reviewers.
references
Adams, E., Cosler, D., Helfrich, K., 1990. Evaporation from heated water bodies: predicting combined forced plus freeconvection. Water Resour. Res. 26, 425e435. Al-Obaidani, S., Curcio, E., Macedonio, F., Di Profio, G., Ai-Hinai, H., Drioli, E., 2008. Potential of membrane distillation in seawater desalination: thermal efficiency, sensitivity study and cost estimation. J. Membr. Sci. 323, 85e98. Barron, J.M., 1992. Installation and operation of multi-effect multi-stage separator coupled with a solar pond, Ms. thesis. University of Texas at El Paso. Beutel, M.W., Horne, A.J., Roth, J.C., Barratt, N.J., 2001. Limnological effects of anthropogenic desiccation of a large, saline lake, Walker Lake, Nevada. Hydrobiologia 466, 91e105. Cabassud, C., Wirth, D., 2003. Membrane distillation for water desalination: how to chose an appropriate membrane? Desalination 157, 307e314. Cath, T., Adams, V., Childress, A., 2004. Experimental study of desalination using direct contact membrane distillation: a new approach to flux enhancement. J. Membr. Sci. 228, 5e16. Cipollina, A., Micale, G., Rizzuti, L., 2009. Seawater Desalination, first ed.. Springer-Verlag, Berlin, Heidelberg. Cooper, J.J., Koch, D.L., 1984. Limnology of a desertic terminal Lake, Walker Lake, Nevada, USA. Hydrobiologia 118, 275e292. Criscuoli, A., Carnevale, M.C., Drioli, E., 2008. Evaluation of energy requirements in membrane distillation. Chem. Eng. Process 47, 1098e1105. Curcio, E., Drioli, E., 2005. Membrane distillation and related operations e a review. Sep. Purif. Rev. 34, 35e86. Dickerson, B.R., Vinyard, G.L., 1999. Effects of high levels of total dissolved solids in Walker Lake, Nevada, on survival and growth of Lahontan cutthroat trout. Trans. Am. Fish. Soc. 128, 507e515. Duffie, J.A., Beckham, W.A., 2006. Solar Engineering of Thermal Processes, third ed. John Wiley & Sons, Inc, Hoboken, New Jersey. El-Bourawi, M.S., Ding, Z., Ma, R., Khayet, M., 2006. A framework for better understanding membrane distillation separation process. J. Membr. Sci. 285, 4e29. Gnielinski, V., 1976. New equations for heat and mass transfer in turbulent pipe and channel flow. Int. Chem. Eng. 16, 359e368. Henderson-Sellers, B., 1986. Calculating the surface energy balance for lake and reservoir modeling: a review. Rev. Geophys. 24, 625e649. Hsu, S.T., Cheng, K.T., Chiou, J.S., 2002. Seawater desalination by direct contact membrane distillation. Desalination 143, 279e287. Hull, J.R., Nielsen, C.E., Golding, P., 1989. Salinity-gradient Solar Ponds. CRC Press, Boca Raton, Florida. Izquierdo-Gil, M.A., Garcia-Payo, M.C., Fernandez-Pineda, C., 1999. Air gap membrane distillation of sucrose aqueous solutions. J. Membr. Sci. 155, 291e307. Kurt, H., Halici, F., Binark, A., 2000. Solar pond conception e experimental and theoretical studies. Energy Convers. Manage. 41, 939e951. Lopes, T.J., Allander, K.K., 2009. Water budgets of the Walker River basin and Walker Lake, California and evada: U.S. Geological Survey scientific investigations. Report 2009-5157.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 0 1 e4 6 1 5
Lu, H., Walton, J., Swift, A., 2001. Desalination coupled with salinity-gradient solar ponds. Desalination 136, 13e23. Lu, H.M., Swift, A.H.P., Hein, H.D., Walton, J.C., 2004. Advancements in salinity gradient solar-pond technology based on sixteen years of operational experience. J. Sol. Energy Trans. ASME 126, 759e767. Martinetti, C.R., Childress, A.E., Cath, T.Y., 2009. High recovery of concentrated RO brines using forward osmosis and membrane distillation. J. Membr. Sci. 331, 31e39. Martinez, L., Rodriguez-Maroto, J.M., 2006. Characterization of membrane distillation modules and analysis of mass flux enhancement by channel spacers. J. Membr. Sci. 274, 123e137. Mathioulakis, E., Belessiotis, V., Delyannis, E., 2007. Desalination by using alternative energy: review and state-of-the-art. Desalination 203, 346e365. Mills, A.F., 1999. Heat Transfer, second ed.. Prentice Hall, Inc, New Jersey. Mobley, C.D., 1994. Light and water: radiative transfer in natural waters. Academic Press, San Diego.
4615
Phattaranawik, J., Jiraratananon, R., Fane, A.G., 2003. Heat transport and membrane distillation coefficients in direct contact membrane distillation. J. Membr. Sci. 212, 177e193. Rabl, A., Nielsen, C., 1975. Solar ponds for space heating. Sol. Energy 17, 1e12. Safavi, M., Mohammadi, T., 2009. High-salinity water desalination using VMD. Chem. Eng. J. 149, 191e195. Schofield, R., Fane, A., Fell, C., Macoun, R., 1990. Factors affecting flux in membrane distillation. Desalination 77, 279e294. Solis, S., 1999. Water desalination by membrane distillation coupled with a solar pond, Ms. thesis. University of Texas at El Paso. Sua´rez, F., Tyler, S.W., Childress, A.E., 2010. A fully coupled, transient double-diffusive convective model for salt-gradient solar ponds. Int. J. Heat Mass Trans. 53, 1718e1730. UNEP, 2001. The watershed: water from the mountains into the sea. Available online at: http://www.unep.or.jp/Ietc/ Publications/Short_Series/LakeReservoirs-2/index.asp. Yun, Y.B., Ma, R.Y., Zhang, W.Z., Fane, A.G., Li, J.D., 2006. Direct contact membrane distillation mechanism for high concentration NaCl solutions. Desalination 188, 251e262.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 3 7 e4 5 4 9
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Anaerobic acidogenic digestion of olive mill wastewaters in biofilm reactors packed with ceramic filters or granular activated carbon Lorenzo Bertin a,*, Silvia Lampis b, Daniela Todaro c, Alberto Scoma a, Giovanni Vallini b, Leonardo Marchetti a,c, Mauro Majone d, Fabio Fava a a
Department of Applied Chemistry and Material Science (DICASM), Faculty of Engineering, University of Bologna, via Terracini 28, 40131 Bologna, Italy b Department of Biotechnology, University of Verona, Strada Le Grazie 15 e Ca’ Vignal, I-37134 Verona, Italy c INCA e Interuniversitary Consortium “Chemistry for the Environment”, via delle Industrie 21/8, I-30175 Marghera (VE), Italy d Department of Chemistry, Sapienza University of Rome, Piazzale Aldo Moro 5, I-00185 Rome, Italy
article info
abstract
Article history:
Four identically configured anaerobic packed bed biofilm reactors were developed and
Received 15 April 2010
employed in the continuous acidogenic digestion of olive mill wastewaters to produce
Received in revised form
volatile fatty acids (VFAs), which can be exploited in the biotechnological production of
4 June 2010
polyhydroxyalkanoates. Ceramic porous cubes or granular activated carbon were used as
Accepted 11 June 2010
biofilm supports. Aside packing material, the role of temperature and organic loading rate
Available online 18 June 2010
(OLR) on VFA production yield and mixture composition were also studied. The process was monitored through a chemical, microbiological and molecular biology integrated
Keywords:
procedure. The highest wastewater acidification yield was achieved with the ceramic-
Olive mill wastewaters
based technology at 25 C, with an inlet COD and an OLR of about 17 g/L and 13 g/L/day,
Acidogenesis
respectively. Under these conditions, about the 66% of the influent COD (not including its
Packed bed biofilm reactors
VFA content) was converted into VFAs, whose final amount represented more than 82% of
Vukopor S10 ceramic cubes
the influent COD. In particular, acetic, propionic and butyric acids were the main VFAs by
Polyhydroxyalkanoates
composing the 55.7, 21.5 and 14.4%, respectively, of the whole VFA mixture. Importantly,
Microbial speciation
the relative concentrations of acetate and propionate were affected by the OLR parameter. The nature of the packing material remarkable influenced the process performances, by greatly affecting the biofilm bacterial community structure. In particular, ceramic cubes favoured the immobilization of Firmicutes of the genera Bacillus, Paenibacillus and Clostridium, which were probably involved in the VFA producing process. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Polyhydroxyalkanoates (PHAs) are promising microbial biopolymers, mainly because: a) they show physical properties
similar to those of some petroleum-derived plastics [e.g. the copolymer poly(3-hydroxybutyrate/3-hydroxyvalerate) [P3-(HB/HV)] can replace polypropylene in a wide range of applications (Lee, 1996)]; b) they can be produced by means of
* Corresponding author. Tel.: þ39 051 2090317; fax: þ39 051 2090322. E-mail addresses:
[email protected] (L. Bertin),
[email protected] (S. Lampis),
[email protected] (D. Todaro), alberto.scoma2@ unibo.it (A. Scoma),
[email protected] (G. Vallini),
[email protected] (M. Marchetti),
[email protected] (M. Majone),
[email protected] (F. Fava). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.025
4538
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 3 7 e4 5 4 9
renewable resources (Serafim et al., 2008); c) they are completely biodegradable (Braunegg et al., 1998) and biocompatible (Mochizuki, 2002). To date, industrial PHA productions have taken advantage only of pure bacterial cultures and welldefined synthetic media (Reddy et al., 2003; Khanna and Srivastava, 2005; Philip et al., 2007). PHA production processes carried out under such well-defined conditions can provide excellent results, but are not economically competitive with those of petroleum-derived plastics (Noda et al., 2005; Philip et al., 2007). Thus, the exploitation of alternative lowcost feeding stocks (e.g. organic wastes) and mixed bacterial consortia (i.e. highly biodiverse microbial populations selectively enriched for the ability of producing PHA) is of great relevance to get an economically feasible PHA production process. In this respect, many papers have dealt with PHAs production by activated sludge microorganisms from different waste (Reis et al., 2003; Chua et al., 2003; Rhu et al., 2003; Dionisi et al., 2005a; Albuquerque et al., 2007; Salmiati et al., 2007; Bengtsson et al., 2008a,b). In particular, Dionisi et al. (2005a) proposed the exploitation of olive mill wastewaters (OMWs) in a three-stage integrated anaerobic-aerobic PHA producing process: in the first anaerobic stage, the organic waste is fermented under acidogenic conditions to obtain an effluent rich in volatile fatty acids (VFAs), which are suitable substrates for the PHA biological synthesis occurring in the following second and third aerobic steps. Although this process has been already assessed for the performances of the aerobic phases (Dionisi et al., 2004, 2005b, 2006), a little has been done so far to optimize the acidogenic step. In this respect, the employment of reactors capable of supporting high flow rate feedings, such as Packed Bed Biofilm Reactors (PBBRs) (Bertin et al., 2004), can allow to minimize the VFAconsuming methanogenic activity, which is normally mediated by bacteria with very low specific growth rates. Thus, we recently developed a biofilm reactor packed with Ceramic Cubes (CCs) with the aim of fermenting OMWs to generate a VFA-rich effluent employed in the production of PHAs (Beccari et al., 2009). However, no efforts to optimize the process in terms of concentration and relative amounts of produced VFAs were made in the previous study. Given the key role played by the relative content of carboxylic acids containing even or odd number of carbon atoms on the properties of resulting PHAs (Bengtsson et al., 2008b), this research was undertaken to evaluate the influence of some key process parameters, such as temperature and Organic Loading Rate (OLR), on concentration and relative amounts of produced VFAs by the recently developed OMW acidogenic process (Beccari et al., 2009). Moreover, the role of the packing material was also evaluated by developing parallel reactors packed with Granular Activated Carbon (GAC), a biomass carrier which was already employed in the biomethanization of OMWs (Bertin et al., 2004), and by determining the microbial speciation of the biofilms generated onto the surface of CC and GAC samples collected at the end of the study. To the very best of our knowledge, this is the first work in which an immobilized cell-based acidogenic anaerobic digestion of OMWs was studied in terms of the influence of its main process parameters and assessed through an integrated chemical, microbiological and molecular biology methodology.
2.
Materials and methods
2.1.
Olive mill wastewaters
Two OMWs, named OMW1 and OMW2, were kindly purchased by the Sant’Agata d’Oneglia (Imperia, Italy) and Grassanese (Matera, Italy) three phase olive mills, respectively, and employed in the research. Their COD was about 25 and 35 g/L, respectively, partially due to VFAs (about 10 and 12 gCOD/L, respectively) and phenols (about 2 g/L in both OMWs). Total and volatile suspended solids were 20 and 12 g/L, respectively, in OMW1; 32 and 13 g/L, respectively, in OMW2. Their pH values were 4.3 and 4.6, respectively.
2.2.
Packed bed biofilm reactors
Four identically configured up-flow PBBRs were developed following the approach reported elsewhere (Beccari et al., 2009) and employed as described in Section 2.3. Each PBBR consisted of a 2.5 L-hermetically closed glass column (diameter: 80 mm; height: 450 mm) wrapped with a silicon tubing serpentine continuously recycling thermostated water and equipped with a recycle line. The recycling ratio, expressed as the ratio between the recycled broth flow and the whole flow entering the column, was about 0.97. The liquid and gas effluents were collected in a tank, connected to a “Mariotte” bottle through which the produced biogas volume was determined. A pH probe (81-04 model, ATI Orion, Boston, MA) was placed at the top of the bioreactor. Two of the reactors were packed with Ceramic Cubes (CCs, Vukopor S10 product, Lanik, Boskovice, CZ) whose dimensions, porosity and density were 25 25 18 mm, 10 ppi and 2.38 g/mL, respectively (CCPBBRs), while the other two with Granular Activated Carbon (GAC, CP4-60 product, Chemviron Carbon, Feluy, Belgium), consisting of cylinders of about 3 mm diameter and 10 mm length, whose density was 1.32 g/mL (GAC-PBBRs). As a result of support addition, the reaction volumes of CC-PBBRs became 2.25 and 2.28 L (CC1 and CC2, respectively), while the ones of GAC-PBBRs became 1.50 and 1.80 L (GAC1 and GAC2, respectively).
2.3.
Olive mill wastewater digestion experiments
Where not differently described, the reactor influent flows consisted of OMWs diluted with an equal volume of tap water and amended with urea (0.45 g/L) and with a 10 N NaOH solution (added to correct their pH to 5.5). One PBBR per group, i.e. CC1 and GAC1, where thermostated at 55 C, whereas the other two, i.e. CC2 and GAC2, at 35 C. They were all filled anaerobically with amended OMW1, which was previously inoculated at 10% v/v with the same deoxygenated, high-density suspension of microbial biomass employed to develop the process described in Beccari et al. (2009), which did not harbor any detectable taxa belonging to archaeal domain. The main detected bacteria occurring in the inoculum were a strain belonging to the Flexibacter-Cytophaga-Bacteroides group and a Syntrophus sp.. The reactor medium was replaced with deoxygenated fresh amended OMW1 for three successive two-weeks batch cycles. The
4539
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 3 7 e4 5 4 9
reactors were then forced to operate under continuous mode. Five experiments of about one-month long were carried out with the aim of studying the effect of temperature and loading conditions on process performances. Steady state conditions were considered to be attained when VFAs, COD and phenols concentrations along with produced biogas were appreciably constant (with standard deviations generally not exceeding 15%) for at least a week (i.e. at least 4 times each process HRT). Steady state was typically achieved within two weeks after a new loading condition was set. The main process working parameters are reported in Table 1. The first experiment, through which information on the influence of temperature on process performances was attained, was carried out by feeding both thermophilic (experiment No. 1a) and mesophilic (experiment No. 1b) reactors with the amended OMW1. The four successive experiments, through which the former information was integrated and the influence of OLR was studied, were performed with CC2 and GAC2 reactors, which were fed with the amended OMW1 (experiments No. 2 and 3) or OMW2 (experiments No. 4 and 5). Experiments No. 3, 4 and 5 were carried out by thermostating the reactors at 25 C. PBBRs were fed at an OLR of about 9 (experiments No. 1 and 3) and 13 (experiments No. 2 and 4) g/L/day. During the last experiment, OMW2 was not diluted with tap water, so that the influent COD was doubled respect to the one of experiment No. 4 (Table 1). Moreover, during the last experiment the pH of GAC2 inlet flow was not modified by means of NaOH amendment. Liquid samples (collected on daily bases) were filtered on 0.22 mm cellulose-nitrate filters and then analyzed for COD, VFA and total phenols (parameters related to influent flows are reported with the subscript suffix “IN”, e.g. CODIN, while the ones related to the effluents are reported with the subscript suffix “OUT”, e.g. CODOUT). VFA amounts were expressed as g of equivalent COD (gCOD) by means of stoichiometric calculations. The net COD conversion into VFAs (COD/VFA) percentage, representing the ratio between the net VFA production and the influent COD excluding its VFA fraction, was employed to define the process efficiency and calculated as follows: COD/ VFA ¼
VFAOUT VFAIN 100 CODIN VFAIN
(1)
The whole process yield was calculated as the percentage of the ratio between the effluent VFA total amount and the influent COD:
Process yield ¼
VFAOUT 100 CODIN
(2)
The concentration of the total immobilized biomass occurring at the top, the middle and the bottom of CC2 and GAC2 was quantified at the end of the study by collecting samples of about 1 g of support from the bottom, the middle and the top of the fixed-beds (50, 200 and 380 mm from the column bottom, respectively) and by subjecting them to protein analysis. The biofilms immobilized on the surface of the same samples together with the last experiment influent and effluent flows were microbiologically characterized by means of DNA extraction as described below.
2.4. Total DNA extraction, PCR amplification and denaturing gradient gel electrophoresis Total DNA extraction from CC and GAC samples was carried as reported elsewhere (Beccari et al., 2009). The 16S rRNA-genes were amplified by PCR using Taq DNA polymerase with primers targeting conserved domains. Bacterial 16S rRNA genes were selectively amplified using F8/R11 primers (Beccari et al., 2009) with the following thermocycling program: initial denaturation at 94 C for 2 min; 30 cycles of denaturation at 94 C for 45 s, annealing at 50 C for 30 s, and extension at 72 C for 2.5 min; final extension at 72 C for 5 min. Afterwards a nested PCR was performed as described in Beccari’s work (2009). Conditions were as above, except for number of cycles, 35, the annealing temperature, 57 C, and extension time, 35 s. For Archaea, primers A109-f (Grosskopf et al., 1998) and 1510-r (Lane, 1991) were used for nearly complete 16S rRNA gene amplification. Afterwards a nested PCR was performed on the hypervariable V2-V3 region using primers A109(T)-f and 515-GC-r (Roest et al., 2005), with a GC-clamp. The first archaeal PCR reaction was performed with the following thermocycle program: initial denaturation at 94 C for 5 min; 30 cycles of denaturation at 94 C for 45 s, annealing at 52 C for 30 s, extension at 72 C for 1 min; and final extension at 72 C for 5 min. The nested PCR was as above but with 35 cycles. All primers were purchased from Sigma-Genosys (Milan, Italy). The PCR products were quantified using Low DNA MassTM Ladder (Celbio, Italy) in a 2.0% agarose gel. DGGE analyses were performed on amplicons obtained both for bacterial V3 and archaeal V2-V3 regions following the procedures reported in Beccari’s et al.’s work (2009). Representative DGGE bands were excised and incubated for 4 h in
Table 1 e List of performed experiments and related main working parameters. Experiment No.1 was carried out with both reactor couples (CC1 and GAC1, experiment No. 1a; CC2 and GAC2, experiment No. 1b). Exp.
OMW
T ( C)
CODIN (g/L)
VFAIN (gCOD/L)
PhenolIN (g/L)
1a 1b 2 3 4 5
OMW1 OMW1 OMW1 OMW1 OMW2 OMW2
55 35 35 25 25 25
13.52 0.43 13.52 0.43 11.71 0.26 11.55 0.32 16.71 0.52 36.64 0.74
6.183 0.29 6.183 0.29 6.465 0.25 5.812 0.14 7.924 0.29 12. 52 0.46
1.034 0.08 1.034 0.08 1.009 0.04 1.399 0.06 0.978 0.03 2.393 0.04
a CC2. b GAC2.
OLR (g/L/day) 9.66 9.62 12.5 8.17 13.3 22.3a
44.4b
4540
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 3 7 e4 5 4 9
50 mL of sterile water. A total of 12 and 21 bands from amplicons obtained respectively for Archaea and Bacteria regions were sequenced.
2.5.
3.
Results
Significant amounts of VFAs were found to be generated in the bioreactors only when the anaerobic treatment was performed in CC2. The main results of the five successive experiments performed with the CC- and GAC-PBBRs under continuous mode of operation are summarized in Table 2. The highest VFAs production was observed when CC2 was thermostated at 25 C and fed with an OLR of about 13 g/L/day (experiment No. 4): under such conditions, the total VFAs concentration in the effluent was 13.73 gCOD/L and it was mainly due to acetic, propionic and butyric acids (representing the 55.7, 21.5 and 14.4% of the whole detected VFAs, respectively, Fig. 1a). The net COD conversion into VFAs was about the 66%. VFAs total amount represented about the 89% of the overall effluent COD (Table 3), while the process yield was about 82% (Table 3). The pH of such a VFA-rich effluent was 5.13 0.04.
Cloning, sequencing, and phylogenetic analysis
DGGE bands containing DNA to be sequenced were re-amplified. PCR amplification was carried out as described before, except for the use of non-GC-clamped primers. PCR products were transformed in Escherichia coli Xl1blue using the pGEM-T vector system according to the manufacturer’s instructions (Promega, Italy), sequenced on both strands, and finally searched for homology using the BLASTN database (Altschul et al., 1997). The sequences were initially aligned using the multiple alignment program CLUSTAL_X 1.83 (Thompson et al., 1997). A phylogenetic tree was constructed using the neighbourjoining method with the MEGA version 4.0 software package (Kumar et al., 2008). Bootstrap analysis was performed from 1000 bootstrap replications.
3.1. 2.6.
Effect of packing material on PBBRs performances
Analytical methods The effect of packing material on process performances was studied throughout the first four experiments performed with the parallel CC- and GAC-PBBR under the same temperature and loading conditions (CC1 and GAC1, employed in the sole experiment No. 1; CC2 and GAC2, Table 1). Higher VFA concentrations were always attained in the CC-PBBRs than in the GAC-PBBRs, where a large part of the influent VFA load was conversely depleted (Table 2). Compatible pH values for biological acidogenic processes were always measured for CCPBBRs (5.23 averagely), while in GAC-PBBRs such a parameter was in the range 6.43e6.97. High methane productions occurred in the latter reactors, where in particular CH4 production yields close to the maximum theoretical ones (i.e. 0.35 L of CH4 produced per g of COD removed) were obtained in the first two experiments (Table 2). At the same time, GACPBBRs provided very high COD removals which ranged between 58 and 86% of the initial COD. Conversely, COD removals were very low (between 3 and 31%) in CC-PBBRs, where a negligible methanogenic activity occurred during the experiments carried out at 25 C. Different mixtures of VFAs accumulated in the two packed bed reactor types: acetic, propionic and butyric acids were the main VFAs generally occurring in CC-PBBR effluents, whereas
VFA concentration was monitored through a HP GC-5890 (Agilent, Milano, Italy) equipped with a FID detector and a Supelcowax-10 column (SigmaeAldrich, Milano, Italy) under the following conditions: initial temperature 60 C; isothermal for 1 min; temperature rate 25 C/min; final temperature 150 C; isothermal for 6 min; temperature rate 4 C/min; final temperature 180 C; temperature rate 25 C/min; final temperature 240 C; injector and FID temperature 280 C; carrier gas flow rate (nitrogen) 17.6 mL/min. Before the analyses, the samples were diluted with an equal amount of a 60 mM oxalic acid solution. VFAs concentrations are expressed as g of COD equivalents/L (gCOD/L). COD and total phenol concentrations were determined spectrophotometrically according to the following methods: Hach Mn(III) (Miller et al., 2001) and Folin-Ciocalteu (Folin and Ciocalteu, 1927), respectively. Total phenols were determined by employing 4-hydroxybenzoate as the calibration standard. Biogas amount and composition were daily determined as reported elsewhere (Bertin et al., 2004). The concentration of the total biomass occurring at the bottom, the middle and the top of the reactor packed beds was quantified according to the Lowry method applied in previous studies (Bertin et al., 2004).
Table 2 e COD, VFA and phenol concentrations in PBBR effluents along with CH4 production yields related to all the performed experiments. Exp.
CC-PBBRs CODOUT (g/L)
1a 1b 2 3 4 5
10.93 9.350 10.33 11.20 15.47 28.52
a Depleted COD.
0.37 0.26 0.30 0.16 0.43 0.89
VFAOUT (gCOD/L) 6.138 6.207 6.232 6.121 13.73 13.67
0.11 0.52 0.30 0.32 0.49 0.42
GAC-PBBRs
CH4/CODDEPa (L/g) 0.197 0.334 0.380 0.010 0.015 0.024
0.02 0.05 0.06 0.00 0.00 0.01
PhenolOUT (g/L) 0.774 0.797 1.007 1.120 0.830 2.256
0.10 0.07 0.03 0.06 0.05 0.06
CODOUT (g/L) 2.827 2.133 1.677 1.787 5.396 15.27
0.16 0.06 0.02 0.14 0.35 1.05
VFAOUT (gCOD/L) 0.837 0.501 0.189 0.553 1.497 4.808
0.18 0.04 0.03 0.06 0.10 0.63
CH4/CODDEPa (L/g) 0.391 0.339 0.324 0.261 0.116 0.135
0.04 0.06 0.02 0.01 0.01 0.02
PhenolOUT (g/L) 0.071 0.076 0.287 0.389 0.207 0.954
0.04 0.02 0.03 0.03 0.06 0.16
4541
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 3 7 e4 5 4 9
A 100 80 60 40 20 0 Ace
tic nic ric pio uty yric ic Pro i-B But aler eric ers i-V Val Oth
5
4
3
2
1b
1a
55 C; CC2 and GAC2 at 35 or 25 C, Table 1). Concerning CCPBBRs, an accumulation of VFAs was observed in CC2 at 25 C (experiment No. 3), while no appreciable differences among VFA concentrations occurring in influent and effluent flows were observed at both 55 C (experiment No. 1a) and 35 C (experiment No. 1b) (Tables 1 and 2). Accordingly, methanogenesis was almost absent at 25 C (Table 2). In addition, no significant COD depletions were measured for CC2 under the latter experimental conditions. At the opposite, high methane productions along with high COD depletions were always attained within GAC-PBBRs (Table 2). No effects related to temperature on VFA distribution were observed in both CC- and GAC-systems.
B 3.3. Effect of organic loading rate on PBBRs performances
100 80 60 40 20 0 tic c Ace pioni ric uty yric ic Pro i-B But aler eric ers i-V Val Oth
5
4
3
2
1b
1a
Fig. 1 e Single VFA relative amounts with respect to the total VFA amounts detected in CC-PBBR (A) and GAC-PBBR (B) effluents in all the performed experiments (expressed by their identificative number).
only the former two were prominent in GAC-PBBRs (Fig. 1), where propionic acid concentration was averagely higher than in CC-PBBRs (28 and 23.5%, respectively). The two reactor systems also exhibited different ability to remove the OMW phenols: such compounds tent to persist in all CC-effluents while they were depleted up to more than 90% within the GAC-PBBRs (Table 2).
3.2.
Effect of temperature on PBBRs performances
The effect of temperature on VFAs and biogas production was studied throughout experiments No. 1 (a and b) and 3, by feeding each of the reactors with very similar OLRs and by differently thermostating each reactor couple (CC1 and GAC1 at
The effect of OLR on process performances was studied throughout two couples of experiments in which CC2 and GAC2 were fed with increasing OLRs (about 9 and 13 g/L/day). In particular, experiments No. 1b and 2 were carried out at 35 C, while experiments No. 3 and 4 at 25 C. When exposed to the higher OLR, both packing materials gave rise to higher VFA productions at 25 C but lower at 35 C, where high methanogenic activities were observed (Table 2). Concerning the effect on VFA mixture composition, the increase of OLR generally caused the decrease of the relative abundance of acetic acid and, at 25 C, the increase of propionic acid in both CC- and GAC-PBBRs (Fig. 1). On the basis of the evidences observed within the first four experiments, a fifth final experiment was carried out by differently loading CC2 and GAC2. In particular, CC2 was fed with a higher OLR (from 13.3 to 22.3 gCOD/L/day) by doubling the incoming COD and lowering the flow rate: the aim of this approach was to maintain the high net conversion yields previously obtained by processing higher concentrated wastewaters with a lower flow, this resulting in anaerobic effluents with higher VFA concentrations. As concerns GAC2, the last run was aimed at limiting methanogenesis and this by strongly increasing the OLR (more than three times higher, from 13.3 to 44.4 gCOD/L/day), by both increasing the flow rate and the incoming COD concentration. Furthermore, to verify if it was possible to operate with a lower pH with respect to the one adopted in all the former experiments, GAC2 was fed with a pH which was not increased by NaOH amendment. A marked decrease of VFAs production was observed in CC2,
Table 3 e Net COD conversions into VFAs, amount of VFAs with respect to the COD in the influent and effluent flows along with process yields related to all the performed experiments. All data are expressed as percentages. Exp.
1a 1b 2 3 4 5
CC-PBBRs
GAC-PBBRs
COD/VFA
(VFA/COD)IN
(VFA/COD)OUT
Process yield
COD/VFA
VFAIN/CODIN
VFAOUT/CODOUT
Process yield
0.61 0.33 4.44 5.38 66.1 4.77
45.7 45.7 55.2 50.3 47.4 34.2
56.2 66.4 60.3 54.6 88.7 47.9
45.4 45.9 53.2 53.0 82.2 37.3
72.9 77.4 120 91.6 73.1 32.0
45.7 45.7 55.2 50.3 47.4 34.2
29.6 23.5 11.3 30.9 27.7 31.5
6.19 3.70 1.61 4.79 8.96 13.1
4542
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 3 7 e4 5 4 9
Fig. 2 e Top: DGGE fingerprints of the eubacterial communities. Samples collected from top, middle and bottom sections of the reactors are marked with T, M and B, respectively. OMWIN refers to the inlet flow, CC2OUT and GAC2OUT to the effluents discharged under steady state conditions. Roman numerals (I and II) correspond to different replicates. Letters indicate bands that have been excised, cloned and sequenced. Bottom: Similarity dendrograms indicate relationships among the different DGGE profiles. which probably suffered from the high-applied OLR. However, in agreement with the above described trend, acetic and propionic acid relative amounts further decreased and increased, respectively. Concerning GAC2, methane production yield was still comparable to the one observed in experiment No. 4 although the influent flow pH was not increased to 5.5. However, the highest tested loading condition led to the lowest COD consumption ever detected in such a reactor and the VFAs concentration was correspondently the highest obtained within GAC-PBBRs. Nevertheless, VFAs were yet significantly consumed as a result of the methanogenesis (Table 2).
3.4.
Microbiological features of the biofilms
The total amount of immobilized biomass in CC2 and GAC2, along with its microbial composition, was analyzed at the end of the study, when the two reactors were sacrificed and
samples of both packing materials were collected at three different heights of the reactor packed beds. The total immobilized biomass (quantified as mg of dried biomass/g of dried support) found in the reactor packed beds collected from the bottom, the middle and the top of the reactors was 12.6, 15.2 and 18.1, respectively, for CC2, and 1.8, 3.5 and 2.6, respectively, for the GAC2. Considering the dry weight of the support initially introduced in the GAC- and CC-PBBR (739.8 and 1228.1 g, respectively), it was estimated that the total immobilized biomass available in the CC- and GAC-PBBR was 11.31 and 3.215 g, respectively. The composition of both Bacteria and Archae cenoses enriched at the end of the study in CC2 and GAC2 reactors and in their inlet and outlet flows was monitored through PCRDGGE analysis. All the evidences gathered for Bacteria in relation to either packing matrices or influent and effluent wastewaters indicate the presence of heterogeneous communities, quite rich in bacterial biodiversity (Fig. 2,
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 3 7 e4 5 4 9
4543
Table 4 e Taxonomic characterization of the major bands in the DGGE profiles obtained with primers for Eubacteria related to the biofilm samples collected from CC2 and GAC2 and to the last experiment influent flow (OMWIN) and effluents discharged under steady state conditions from CC2 (CC2OUT) and GAG2 (GAC2OUT). Different sampling positions along the profile of the reactor are marked with T (top), M (middle), and B (bottom). Sample CC2 CC2T-a CC2T-b CC2M-c CC2M-d CC2B-e CC2M-f GAC2 GAC2T-h GAC2T-i GAC2T-l GAC2T-m GAC2T-n GAC2T-o GAC2M-p GAC2M-q GAC2M-r Flows OMWIN-s OMWIN-t OMWIN-u OMWIN-v OMWIN-z CC2OUT-j CC2OUT-y GAC2OUT-x
Phylogenetic group Firmicutes Firmicutes Firmicutes Firmicutes Firmicutes Firmicutes
Firmicutes Actinobacteria b-Proteobacteria b-Proteobacteria g-Proteobacteria g-Proteobacteria a-Proteobacteria Firmicutes Firmicutes a-Proteobacteria a-Proteobacteria Firmicutes Firmicutes b-Proteobacteria
Panel B). The dendrogram obtained through UPGMA method shows low similarity values (<0.2) among the profiles corresponding to samples collected within the differently packed reactors. In fact, it is particularly worth noting a taxonomical grouping in two distinct clusters according to samples drawn from the two reactors. As far as the profile of the influent OMW is concerned, a bacterial community composition closer to that revealed in CC2 compared to the GAC-digester speciated one was found, even if in the presence again of a low similarity (<0.2). The effluents collected from both CC2 and GAC2 presented a bacterial composition similar to those observed within the reactors they came from. Nevertheless, in the case of CC loaded reactor, the microbial cenosis structure of the effluent shows a higher similarity (>0.8) to that acclimated in the corresponding digester with respect to what observed between GAC loaded reactor and its effluent (similarity value equal to 0.2). Major bands in DGGE gels were excised, cloned and sequenced. Results from sequencing (Table 4, Fig. 3) demonstrated the presence of Lactobacillus sp. and Acetobacter sp. in the OMW fed to the reactors. Meanwhile, the bacterial community in CC2 has resulted mainly composed by Firmicutes belonging to the genera Bacillus, Paenibacillus, and Clostridium. On the other hand, GAC2 has revealed a prevailing presence of Proteobacteria, distributed among alpha, beta and gamma sub-classes. In particular, the prominent genera found were Acinetobacter, Comamonas and Massilia. A selective speciation towards bacterial strains belonging to the phyla of
TAXON
Identity (%)
Bacillus sp. Paenibacillus sp. Uncultured Clostridium sp. Clostridiaceae bacterium Clostridium sp. Pasteuriaceae bacterium
100 100 100 99 100 100
Uncultured bacterium clone Uncultured Clostridiales bacterium Uncultured Chloroflexi Eggerthella sinensis Comamonas sp. 2009I4 Uncultured Massilia sp. Acinetobacter sp. Acinetobacter baumannii Uncultured Alphaproteobacteria
100 98 98 99 98 98 100 98 99
Lactobacillus suebicus Lactobacillus camelliae Acetobacter pasteurianus Acetobacter sp. Uncultured Lactobacillus sp. Uncultured bacterium isolate Uncultured Clostridium sp. Aquabacterium sp.
100 100 99 100 98 97 100 98
Firmicutes and Proteobacteria was finally recorded in CC and GAC effluents, respectively. As far as the community structure of Archaea is concerned, DGGE profiles neatly demonstrated distinct speciations inside the two differently packed reactors as well as in the corresponding effluents, with dominant, restricted microbial consortia represented by major bands migrating in the gels (Fig. 4, Panel A). Even in the case of Archaea, a marked diversity in the composition of the communities within reactors packed with CC or GAC filling materials was observed. The formation of two different clusters corresponding to samples from the two reactors was actually observed. Interestingly, the DGGE profiles obtained for the inlet OMW have shown an archaeal composition strongly close (similarity value equal to 0.7) to that recorded for the CC filled reactor, opposite the situation found in GAC-packed digester. No significant differences have been detected between the archaeal populations in effluents from both CC and GAC packed digesters and those acclimated inside the respective reactor. Once again, the similarity was much higher in the case of CC2. Sequencing of major bands from DGGE gel (Table 5, Fig. 5) has evidenced a dominance of the Methanobacterium genus (Methanobacteria family) in the OMW fed to the reactors. The presence of strains belonging to the genera Methanobacterium and Methanobrevibacter was revealed in both CC2 and the corresponding outlet wastewater. On the other hand, GAC filled reactor resulted to be populated mostly by Methanomicrobia such as Methanosarcina sp. and Methanocella sp..
4544
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 3 7 e4 5 4 9
Fig. 3 e Neighbour-joining tree based on the sequence of the hypervariable V3 region of the 16S rRNA gene, showing the phylogenetic relationship of different microbial components (DGGE bands marked with bold letters) and related species within the domain Eubacteria. Bootstrap values are shown for nodes that had >50% support in a bootstrap analysis of 1000 replicates. The scale bar indicates 0.02 substitutions per nucleotide position.
4.
Discussion
The utilization of pure cultures and synthetic media in PHA production strongly contributes to the final high costs of the biopolymer (Noda et al., 2005; Philip et al., 2007). The use of cheap organic wastes as feedstock, along with the
employment of enriched PHA producing mixed consortia in tailored biotechnological processes, can markedly lower such costs. In this respect, the acidogenic anaerobic digestion of a COD-rich waste (such as Olive Mill Wastewater, OMW) could be employed to generate VFAs, which can be used to feed PHA producing bacteria (Dionisi et al., 2004).
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 3 7 e4 5 4 9
4545
Fig. 4 e Top: DGGE fingerprints of the archaebacterial communities. Samples collected from top, middle and bottom sections of the reactors are marked with T, M and B, respectively. OMWIN refers to the inlet flow, CC2OUT and GAC2OUT to the effluents discharged under steady state conditions. Roman numerals (I and II) correspond to different replicates. Letters indicate bands that have been excised, cloned and sequenced. Bottom: Similarity dendrograms indicate relationships among the different DGGE profiles.
Freely suspended cell bioreactors were already employed for producing VFAs from several organic waste and effluents, including OMWs (Ntaikou et al., 2009). The use of Packed Bed Biofilm Reactor (PBBR) filled with Ceramic Cubes (CCs) was also recently proposed by us for the same purpose (Beccari et al., 2009); in the present study, the same process was flunked by a GAC-packed technology with identical configuration and both were assessed by evaluating the influence of temperature and Organic Loading Rate (OLR) on concentration and relative amounts of produced VFAs. Packing material played a crucial role in the biofilm biological activities, as also evidenced by other authors (Ince et al., 1995; Tay and Show, 1999; Picanco et al., 2001; Yang et al., 2004). In particular, the biofilm generated on CCs always gave rise to higher VFA concentrations than those obtained in the parallel GAC reactors where, on the contrary,
a remarkable methanogenic activity was observed. Moreover, GAC-PBBRs gave rise to much higher COD depletions, with potential additional adverse effects on PAH production due to the influent COD subtraction. Thus, the most favourable conditions for a subsequent PHA production were achieved with the CC-based process, where methanogenesis was negligible or absent at all at 25 C and no significant COD removals occurred (Table 2). In particular, the latter evidence allowed process yields, which link VFA amounts to the influent COD (equation (2)), closed to the ratios between VFA and COD concentrations in the effluents (Table 3). Packing material played also a paramount role in the removal of polyphenols, which were significantly depleted only in GACPBBRs (Table 2). Since they are known to inhibit mostly methanogenic populations (Dionisi et al., 2005a), their persistence in CC-systems may have contributed to their
4546
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 3 7 e4 5 4 9
Table 5 e Taxonomic characterization of the major bands in the DGGE profiles obtained with primers for Archaea related to the biofilm samples collected from CC2 and GAC2 and to the last experiment influent flow (OMWIN) and effluents discharged under steady state conditions from CC2 (CC2OUT) and GAG2 (GAC2OUT). Different sampling positions along the profile of the reactor are marked with T (top), M (middle), and B (bottom). Phylogenetic group
Taxon
Identity (%)
CC2 CC2B-a CC2B-b
Methanobacteria Methanobacteria
Methanobrevibacter acididurans Methanobrevibacter sp.
99 99
GAC2 GAC2T-c GAC2T-d GAC2T-e
Methanobacteria Methanomicrobia Methanomicrobia
Methanobacterium bryantii Uncultured Methanosarcinales archaeon Methanocella paludicola
99 99 97
Flows OMWIN-f OMWIN-g OMWIN-h GAC2OUT-i GAC2OUT-l
Actinobacteria Methanobacteria Methanobacteria Methanobacteria Actinobacteria
Atopobium rimae Methanobacterium sp. Methanobacterium bryantii Methanobacterium beijingenses Uncultured Coriobacteriaceae bacterium
96 99 99 99 97
higher efficiency in OMW acidification. On the contrary, temperature and OLR did not affect the removal of such compounds. In accordance with the lower methanogenic activities, acidogenic fermentation in CC-PBBRs raised proportionally by decreasing temperature (experiments No. 1a, 1b and 3) by achieving the best performance at 25 C (5.38% of net COD conversion into VFA, Table 3). Aside temperature, another parameter typically affecting methanogenesis is OLR (Ince
et al., 1995; Grover et al., 2001; Kennedy et al., 2006). Methanogenesis generally decreased when higher OLRs were applied (Table 2). However, temperature played a key role in determining to which extent this occurred, being at 25 C obtained the best results in term of methanogenesis reduction: once again, enhancements achieved with CCs support were much greater than with GAC: the net COD conversion into VFAs was one order of magnitude higher than what previously obtained (up to about 66%), leading to a whole
Fig. 5 e Neighbour-joining tree based on the sequence of the hypervariable V2-V3 region of the 16S rRNA gene, showing the phylogenetic relationship of different microbial components (DGGE bands marked with bold letters) and related species within the domain Archaebacteria. Bootstrap values are shown for nodes that had >50% support in a bootstrap analysis of 1000 replicates. The scale bar indicates 0.05 substitutions per nucleotide position. It is worth noting that the specific primers here used for Archaea efficaciously functioned even in amplification of 16S rRNA gene of Actinobacteria.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 3 7 e4 5 4 9
process yield of about 82% (experiments No. 4, Table 3). Such performances, which are the best among those obtained in this study, were significantly better than those achieved during the previous study performed with the same CC-based technology and at the same temperature (25 C) (Beccari et al., 2009), where the net COD conversion into VFAs and process yield were about 13 and 29%, respectively. On the other hand, when the OLR was further increased (experiment No. 5), acidogenic fermentation was inhibited and the net COD conversion into VFAs dropped back (Table 3). The high concentration of toxic polyphenols present in the wastewater could have contributed to this evidence (Kennedy et al., 2006) together with the inability of CCs, differently from GAC, to adsorb/desorb aromatic compounds, thus “buffering” such toxic effects (Bertin et al., 2004). However, the results related to the fifth experiment were closed to the ones achieved in the previous study, which was carried out under similar loading conditions (Beccari et al., 2009). Thus, even the previously developed process was probably negatively influenced by a high OLR. The employment of the CC-PBBR technology allowed also to obtain a higher OMW acidification with respect to that of a conventional CSTR fed with different OLRs (Ntaikou et al., 2009). In particular, the latter system higher process yield was about 30%, due to an influent COD of 19.5 g/L fed at about 14 g/L/day and to a total effluent VFA amount of about 6 gCOD/L. Furthermore, the CC-based system performances were higher than those of the study of Dionisi et al. (2005a), where OMW was pre-treated by means of bentonite amendment (even followed by centrifugation) and fermented under batch conditions: in that case, comparable total VFA amounts were achieved but with a high initial COD, this resulting in lower process yields (higher process yield: 43.6%, corresponding to an initial COD of 28.5 g/L and to a total final VFA amount of 12.4 g/L). The relative composition of the VFA mixture was mainly affected by the parameter OLR: in particular, CC-biofilms induced a decrease in the amount of acetate both at 35 (experiments No. 1b and 2) and 25 C (experiments No. 3, 4 and 5) (Fig. 1a); within the experiments performed with CC2 at 25 C, it was also reported a relative increase in the propionate content while butyrate amount was maintained constant. Thus, the OMW acidogenic digestion carried out in a CC-PBBR operating at 25 C (i.e. the best performing conditions) seems to allow to control the acetateepropionate relative amounts by regulating the applied OLR; however, the process optimization should take into consideration that the higher total VFA production would not correspond to the higher propionate production. On the contrary, lower hydraulic retention times often led to a reduction of propionic acid, whose highest production corresponded to the highest OMW acidification (Ntaikou et al., 2009; Bengtsson et al., 2008b); however, the opposite effect was also reported (Dinopoulou et al., 1988). The microbiological investigations, regarding the microbial structure of both the biofilms within the two differently packed bioreactors and the wastewaters, clearly indicated a different speciation of either Bacteria and Archaea depending on the two different support materials. This seems to be in accordance with results concerning the performances of the two bioreactors. In fact, the bacterial community that
4547
has been selected within CC2 is exclusively composed by Firmicutes such as Clostridium, Bacillus, and strains belonging to the Pasteuriaceae family. Actually, several bacterial species belonging to these genera are known to show acidogenic activity (Akao et al., 2007). In particular, some of the major bands in the DGGE profiles (Fig. 2, bands c, d, e) revealed the presence of strains closely related to C. tyrobutiricum as well as Clostridium aminovalericum (Fig. 3). C. tyrobutiricum has been shown to produce butyric and acetic acid as its major fermentation products from glucose and xylose (Liu et al., 2006). On the other hand, C. aminovalericum is capable to anaerobically degrade 5-aminovalerate to valerate, acetate, propionate, and ammonia (Barker et al., 1987). Conversely, the bacterial community selected within the GAC2 was mainly composed by Proteobacteria. Among the different major bands in the DGGE profile, Acinetobacter is well represented (Fig. 2, bands p, q). Nevertheless bacteria strains belonging to this genus are common members of microbial consortia involved in the biodegradation of biogenic and xenobiotic compounds, and its high potential for the treatment of phenol-containing wastewaters has been recently elucidated (Liu et al., 2009). As far as the Archaea speciation is concerned, the occurrence of species belonging mainly to the genus Methanobrevibacter in the bioreactor packed with ceramic support is consistent with the environmental conditions settled in such digester. Striking VFA concentrations were in fact observed in CC-PBBRs where acidic pHs (w5.0) might have favoured methanogens to become well established, among those more acidophilic (Savant et al., 2002; Rea et al., 2007). Moreover, as Methanobrevibacter grows on a H2/CO2 gas mixture or, in addition, can utilize formate, strict acidogenic conditions such as those in CC-PBBRs are coherent with the activity of possible consortia of incoming hydrogen-producing acetogens and this hydrogenotrophic methanogen (Fang, 2000). Furthermore, the archaeal composition of the CC2 biofilm was strongly closed to the one of the influent wastewater, this suggesting that the immobilization of a large spectrum of methanogenic consortia was not favoured in such a reactor. The opposite situation found in GAC2, whereas a high methanogenic activity was observed, seems to support the latter hypothesis: the presence of mostly Methanomicrobia is in agreement either with the pH conditions fluctuating nearly around neutrality (w6.8), better conducive to the acclimation of this kind of methanogens (Kendall and Boone, 2006), or with the particular tendency of GAC to be colonized especially by Methanosarcinales (Schmidt and Ahring, 1999). Interestingly, the sole member of Methanobacteria found in GAC2, namely Methanobacterium bryantii, is a species which shows an optimum pH range for growth between 6.9 and 7.2 (Boone, 1987). This picture fits well with recent findings on bacterial speciation in GAC-PBBRs treating OMWs (Rizzi et al., 2006).
5.
Conclusions
An effective OMW acidification process was developed through the employment of a PBBR technology which employed porous ceramic cubes for supporting the biofilm
4548
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 3 7 e4 5 4 9
generation. The highest COD conversion into VFAs was observed when the reactor was thermostated at 25 C and loaded with an OLR of about 13 g/L/day, conditions under which a total VFA concentration of about 14 g/L was achieved, this corresponding to about the 88 and the 82% of the influent and effluent COD, respectively. The relative amounts of the main VFAs can be controlled by regulating the applied OLR, and this is of particular interest in the perspective of feeding the acidogenic effluent to a biotechnological PHA producing process. The packing material appeared as a crucial parameter able to influence the process performances much more than temperature and OLR, and this by greatly affecting the biofilm bacterial community structure.
Acknowledgements The authors thank the Olive mills Sant’Agata d’Oneglia (Imperia, Italy) and Grassanese (Matera, Italy) for having provided the OMWs. This research was financially supported by the Italian Ministry of University and Research (PRIN 2005) and partially by the Fondazione Del Monte di Bologna e Ravenna (Bologna, Italy).
references
Albuquerque, M.G.E., Eiroa, M., Torres, C., Nunes, B.R., Reis, M.A.M., 2007. Strategies for the development of a side stream process for polyhydroxyalkanoate (PHA) production from sugar cane molasses. J. Biotechnol. 130, 411e421. Akao, S., Tsuno, H., Horie, T., Mori, S., 2007. Effects of pH and temperature on products and bacterial community in L-lactate batch fermentation of garbage under unsterile condition. Water Res. 41 (12), 2636e2642. Altschul, S.F., Madden, T.L., Schaffer, A.A., Zhang, J., Zhang, Z., Miller, W., Lipman, D.J., 1997. Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25, 3389e3402. Barker, H.A., D’Ari, L., Kahn, J., 1987. Enzymatic reactions in the degradation of 5-aminovalerate by Clostridium aminovalericum. J. Biol. Chem. 262 (19), 8994e9003. Beccari, M., Bertin, L., Dionisi, D., Fava, F., Lampis, S., Majone, M., Valentino, F., Vallini, G., Villano, M., 2009. Exploiting olive oil mill effluents as a renewable resource for production of biodegradable polymers through a combined anaerobiceaerobic process. J. Chem. Technol. Biotechnol. 84, 901e908. Bengtsson, S., Werker, A., Christensson, M., Welander, T., 2008a. Production of polyhydroxyalkanoates by activated sludge treating paper mill wastewater. Bioresour. Technol. 99, 509e516. Bengtsson, S., Hallquist, J., Werker, A., Welander, T., 2008b. Acidogenic fermentation of industrial wastewaters: effect of chemostat retention time and pH on volatile fatty acids production. Biochem. Eng. J. 40, 492e499. Bertin, L., Berselli, S., Fava, F., Petrangeli-Papini, M., Marchetti, L., 2004. Anaerobic digestion of olive mill wastewaters in biofilm reactors packed with granular activated carbon and “Manville” silica beds. Water Res. 38, 3167e3178. Boone, D.R., 1987. Replacement of the type strain of Methanobacterium formicicum and reinstatement of Methanobacterium bryantii sp. nov. nom. rev. (ex Balch and
Wolfe, 1981) with M.o.H. (DSM 863) as the type strain. Int. J. Syst. Bacteriol. 37, 172e173. Braunegg, G., Lefebvre, G., Genser, K.F., 1998. Polyhydroxyalkanoates, biopolyesters from renewable resources: physiological and engineering aspects. J. Biotechnol. 65, 127e161. Chua, A.S.M., Takabatake, H., Satoh, H., Mino, T., 2003. Production of polyhydroxyalkanoates (PHA) by activated sludge treating municipal wastewater: effect of pH, sludge retention time (SRT), and acetate concentration in influent. Water Res. 37 (15), 3602e3611. Dionisi, D., Carucci, G., Petrangeli-Papini, M., Riccardi, C., Majone, M., Carrasco, F., 2005a. Olive oil mill effluents as a feedstock for production of biodegradable polymers. Water Res. 39, 2076e2084. Dionisi, D., Beccari, M., Di Gregorio, S., Majone, M., PetrangeliPapini, M., Vallini, G., 2005b. Storage of biodegradable polymers by an enriched microbial community in a sequencing batch reactor operated at high organic load rate. J. Chem. Technol. Biotechnol. 8, 1306e1318. Dionisi, D., Majone, M., Papa, V., Beccari, M., 2004. Biodegradable polymers from organic acids by using activated sludge enriched by aerobic periodic feeding. Biotechnol. Bioeng. 85 (6), 569e579. Dionisi, D., Majone, M., Vallini, G., Di Gregorio, S., Beccari, M., 2006. Effect of the applied organic load rate on biodegradable polymer production by mixed microbial cultures in a sequencing batch reactor. Biotechnol. Bioeng. 93, 76e88. Dinopoulou, G., Rudd, T., Lester, N.J., 1988. Anaerobic acidogenesis of a complexwastewater: I. The influence of operational parameters on reactor performance. Biotechnol. Bioeng 31, 958e968. Fang, H.H.P., 2000. Microbial distribution in UASB granules and its resulting effects. Water Sci. Technol. 42 (12), 201e208. Folin, O., Ciocalteu, V., 1927. On tyrosine and tryptofan determinations in protein. J. Biol. Chem. 73, 627e650. Grosskopf, R., Janssen, P.A., Liesack, W., 1998. Diversity and structure of the methanogenic community in anoxic rice paddy soil microcosms as examined by cultivation and direct 16S rRNA gene sequence retrieval. Appl. Environ. Microbiol. 64, 960e969. Grover, R., Marwaha, S.S., Kennedy, J.F., 2001. Methanogenesis of black liquor in a two-stage biphasic reactor system using an immobilized cell system. J. Chem. Technol. Biotechnol. 76, 251e256. Ince, O., Anderson, G.K., Kasapgil, B., 1995. Control of organic loading rate using the specific methanogenic activity test during start-up of an anaerobic digestion system. Water Res. 29 (1), 349e355. Kendall, M.L., Boone, D.R., 2006. The order Methanosarcinales. In: Dworkin, M., Falkow, S., Rosenberg, E., Schleifer, K.-H., Stackebrandt, E. (Eds.), The ProkaryotesArchaea. Bacteria: Firmicutes, Actinomycetes, third ed., vol. 3. Springer, New York, pp. 244e256. Kumar, S., Nei, M., Dudley, J., Tamura, K., 2008. MEGA: a biologistcentric software for evolutionary analysis of DNA and protein sequences. Brief. Bioinformatics 9, 299e306. Kennedy, J.F., Panesar, P.F., Grover, R., Marwaha, S.S., 2006. Continuous methanogenesis of black liquor of pulp and paper mills in an anaerobic baffled reactor using an immobilized cell system. J. Chem. Technol. Biotechnol. 81, 1277e1281. Khanna, S., Srivastava, A.K., 2005. Recent advances in microbial polyhydroxyalkanoates. Process Biochem. 40, 607e619. Lane, D.J., 1991. 16S/23S rRNA sequencing. In: Stackebrandt, E., Goodfellow, M. (Eds.), Nucleic Acid Techniques in Bacterial Systematics. Wiley, Chichester, United Kingdom, pp. 115e175. Lee, S.Y., 1996. Bacterial polyhydroxyalkanoates. Biotechnol. Bioeng. 49, 1e14.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 3 7 e4 5 4 9
Liu, Y.J., Nikolausz, M., Wang, X.C., 2009. Biodegradation and detoxication of phenol by using free and immobilized cells of Acinetobacter sp. XA05 and Sphingomonas sp. FG03. Environ. Sci. Health A Tox. Hazard. Subst. Environ. Eng. 44 (2), 130e136. Liu, X., Zhu, Y., Yang, S.T., 2006. Butyric acid and hydrogen production by Clostridium tyrobutyricum ATCC 25755 and mutants. Enzyme Microb. Technol. 38 (3e4), 521e528. Miller, D.G., Brayton, S.V., Boyles, W.T., 2001. Chemical oxygen demand analysis of wastewater using trivalent manganese oxidant with chloride removal by sodium bismuthate pretreatment. Water Environ. Res. 73 (1), 63e71. Mochizuki, M., 2002. Properties and application of aliphatic polyester products. In: Steinbu¨chel, A., Doi, Y. (Eds.), Biopolymers, vol. 4. Wiley-vch, Weinheim. Noda, I., Green, P.R., Satkowski, M.M., Schectman, L.A., 2005. Preparation and properties of a novel class of polyhydroxyalkanoate copolymers. Biomacromolecules 6, 580e586. Ntaikou, I., Kourmentza, C., Koutrouli, E.C., 2009. Exploitation of olive oil mill wastewater for combined biohydrogen and biopolymers production. Biores. Technol. 100 (15), 3724e3730. Philip, S., Keshavarz, T., Roy, I., 2007. Polyhydroxyalkanoates: biodegradable polymers with a range of applications. J. Chem. Technol. Biotechnol. 82, 233e247. Picanco, A.P., Vallero, M.V.G., Gianotti, E.P., Zaiat, M., Blundi, C. E., 2001. Influence of porosity and composition of supports on the methanogenic biofilm characteristics developed in a fixed bed anaerobic reactor. Water Sci. Technol. 44, 197e204. Rea, S., Bowman, J.P., Popovski, S., Pimm, C., Wright, A.D.G., 2007. Methanobrevibacter millerae sp. nov. and Methanobrevibacter olleyae sp. nov., methanogens from the ovine and bovine rumen that can utilize formate for growth. Int. J. Syst. Evol. Microbiol. 57, 450e456. Reddy, C.S.K., Ghai, Rashmi, R., Kalia, V.C., 2003. Polyhydroxyalkanoates: an overview. Bioresour. Technol. 87, 137e146. Reis, M.A.M., Serafim, L.S., Lemos, P.C., Ramos, A.M., Aguiar, F.R., Van Loosdrecht, M.C.M., 2003. Production of
4549
polyhydroxyalkanoates by mixed microbial cultures. Bioprocess. Biosyst. Eng. 25 (6), 377e385. Rhu, D.H., Lee, W.H., Kim, J.Y., Choi, E., 2003. Polyhydroxyalkanoate (PHA) production from waste. Water Sci. Technol. 48 (8), 221e228. Rizzi, A., Zucchi, M., Borin, S., Marzorati, M., Sorlini, C., Daffonchio, D., 2006. Response of methanogen populations to organic load increase during anaerobic digestion of olive mill wastewater. J. Chem. Technol. Biotechnol. 81, 1556e1562. Roest, K., Heilig, H.G.H.J., Smidt, H., de Vos, W.M., Stams, A.J.M., Akkermans, A.D.L., 2005. Community analysis of a full-scale anaerobic bioreactor treating paper mill wastewater. Syst. Appl. Microbiol. 28, 175e185. Salmiati, Z., Ujang, M.R., Salim, M.F., Din, M.D., Ahmad, M.A., 2007. Intracellular biopolymer productions using mixed microbial cultures from fermented POME. Water Sci. Technol. 56 (8), 179e185. Savant, D.V., Shouche, Y.S., Prakash, S., Ranade, D.R., 2002. Methanobrevibacter acididurans sp. nov., a novel methanogen from a sour anaerobic digester. Int. J. Syst. Evol. Microbiol. 52, 1081e1087. Schmidt, J.E., Ahring, B.K., 1999. Immobilization patterns and dynamics of acetate-utilizing methanogens immobilized in sterile granular sludge in upflow anaerobic sludge blanket reactors. Appl. Environ. Microbiol. 65 (3), 1050e1054. Serafim, L.S., Lemos, P.C., Albuquerque, M.G.E., Reis, M.A.M., 2008. Strategies for PHA production by mixed cultures and renewable waste materials. Appl. Microbiol. Biotechnol. 81, 615e628. Tay, J., Show, K., 1999. Influence of support media on biomass growth and retention in anaerobic filters. Water Res. 33, 1471e1481. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F., Higgins, D.G., 1997. The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25, 4876e4882. Yang, Y., Tada, C., Shohidullah Miah, M., Tsukahara, K., Yagishita, T., Sawayama, S., 2004. Influence of bed materials on methanogenic characteristics and immobilized microbes in anaerobic digester. Mater. Sci. Eng. 24, 413e419.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Attenuation of total organic carbon and unregulated trace organic chemicals in U.S. riverbank filtration systems Christiane Hoppe-Jones, Gretchen Oldham, Jo¨rg E. Drewes* Advanced Water Technology Center (AQWATEC), Environmental Science and Engineering Division, Colorado School of Mines, Golden, CO 80401-1887, United States
article info
abstract
Article history:
There is increasing concern regarding the presence of unregulated trace organic chemicals
Received 6 March 2010
in drinking water supplies that receive discharge from municipal wastewater treatment
Received in revised form
plants. In comparison to conventional and advanced drinking water treatment, riverbank
3 June 2010
filtration represents a low-cost and low-energy alternative that can attenuate total organic
Accepted 7 June 2010
carbon (TOC) as well as trace organic chemicals (TOrC). This study examined the role of
Available online 15 June 2010
predominant redox conditions, retention time, biodegradable organic carbon, and temperature to achieve attenuation of TOC and TOrC through monitoring efforts at three
Keywords:
full-scale RBF facilities in different geographic areas of the United States. The RBF systems
Emerging trace organic chemicals
investigated in this study were able to act as a reliable barrier for TOC, nitrogen, and certain
Redox conditions
TOrC. Temperature (seasonal) variation played an important role for the make-up of the
Riverbank filtration
river water quality and performance of the RBF systems. Temperatures of less than 10 C
Total organic carbon
did not affect TOC removal but resulted in diminished attenuation of nitrate and select TOrC. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Most conventional freshwater resources in arid and semi-arid regions of the United States are insufficient in quantity to provide additional drinking water supplies for future demand caused by an increase in population. In these regions, current supplies might be further depleted by climate change impacts. This has created a need to develop other practices to augment freshwater supplies with unconventional resources of diminished quality, such as reclamation of municipal wastewater effluents and desalination of brackish groundwater and seawater. Drinking water augmentation to replenish groundwater resources has been practiced in the U.S. for more than 45 years via soil-aquifer treatment (SAT) and riverbank filtration (RBF) (Drewes and Khan, 2010). SAT relies on surface spreading basins that allow percolation of water through
a vadose zone before it recharges the underlying groundwater. RBF systems are characterized by a series of abstraction wells in the vicinity of a stream or lake resulting in groundwater depletion that forces river or lake water to infiltrate into the subsurface towards abstraction wells. The advantages of these managed aquifer recharge (MAR) systems are multifold including storage and recharge of local groundwater resources, buffering water quality changes in the source water, equilibration of temperature, and reliable attenuation of turbidity, bulk organic matter, pathogens, certain pesticides, and other water constituents of concern (Sontheimer, 1991; Ku¨hn and Mu¨ller, 2000; Drewes et al., 2003). In comparison to conventional and advanced drinking water treatment, MAR systems have a significantly lower energy demand, do not require input of chemicals or generate residuals, and thus exhibit a rather low carbon footprint.
* Corresponding author. E-mail address:
[email protected] (J.E. Drewes). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.022
4644
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
Advances in environmental analytical chemistry during the last 15 years have enabled occurrence studies of currently unregulated trace organic chemicals (TOrC), such as pharmaceutical residues, personal care products, household chemicals, and natural hormones, in drinking water sources. While TOrC occur in environmental samples at the partsper-trillion (ppt) level representing a minute contribution to the total organic carbon (TOC), findings of recent studies have spurred concern about their human health relevance since TOrC were first identified in U.S. watersheds in the mid 1960s (Stumm-Zollinger and Fair, 1965; Tabak and Bunch, 1970). Reports on surface and groundwater contamination in watersheds throughout the United States published during the last decade revealed that the occurrence of TOrC in drinking water sources is ubiquitous (Kolpin et al., 2002; Focazio et al., 2008; Benotti et al., 2009). While the occurrence patterns of these compounds might vary country by country as a function of different prescription practices and usage patterns, multiple studies have demonstrated that conventional wastewater treatment is not an efficient means of retaining these potentially harmful compounds completely (Ternes, 1998; Clara et al., 2005; Glassmeyer et al., 2005). Current data suggests that no acute adverse effects regarding human health are expected at concentrations in the ppt-range (Schwab et al., 2005; Snyder, 2008), however, the presence of these compounds is undesirable in drinking water because they indicate impact from wastewater discharge in the source water. MAR systems, such as RBF and SAT, may prove to be viable and sustainable options for attenuating TOrC. While these systems have been studied for decades regarding their efficacy to remove turbidity, dissolved organic carbon (DOC), and pathogens, only a few studies have demonstrated that riverbank filtration systems can reliably attenuate TOrC (Ku¨hn and Mu¨ller, 2000; Gru¨nheid et al., 2005). Subsurface conditions can vary among RBF sites and a basic understanding regarding the effect of predominant redox conditions, retention time, biodegradable organic carbon (BDOC), and temperature on the attenuation of TOrC is lacking. This study was designed to reveal the role of these factors for the removal of TOrC during
RBF through extensive monitoring efforts at three full-scale facilities in different geographic areas of the United States.
2.
Description of the field sites
In this study, performances of three full-scale RBF sites in the U.S. were investigated regarding their efficacy to remove TOC and TOrC. These sites are located on the South Platte River (Colorado), the Cedar River (Iowa), and the Ohio River (Kentucky). Site-specific information of the selected sites is summarized in Table 1.
2.1.
Brighton, Colorado
The RBF site at the South Platte River in Brighton, Colorado, was established as a demonstration-scale RBF site by the City of Aurora in 2005 and operated until 2007. The flow in the river is significantly impacted by snow-melt that can elevate the average flow of 5 m3/s to more than 120 m3/s during spring run-off. The test site consisted of one production well (PTW1) and 15 monitoring wells that were arranged in three transects (Fig. 1). The production well was located approximately 120 m off the river. The casing of the production well had an inner diameter of 0.46 m and a screening interval of 3 m. The monitoring wells were 0.05e0.1 m in diameter, drilled to the bedrock, and screened along the full length of the saturated aquifer. The aquifer in this area was approximately 9 m thick, with a sandy upper layer and a sand and gravel lower layer, underlain by bedrock. Screening intervals and depth of the production and monitoring wells are summarized in Table 2. The production well was separated from the influence of landside groundwater by a slurry wall located west of the production well and oriented parallel to the river. During the study, the production well was operated continuously at a pumping rate of 28.4 L/s. When the site was established in 2005, two soil samples, one from the riverbed (20 cm below surface) and one from the core of a monitoring well (5 m below surface), were collected and analyzed for their grain size
Table 1 e Site specific information of the three RBF study sites. River Annual mean daily flow Wastewater impact Average flow Low flow Production well(s) Design Depth Screening interval Operation Travel time Distance from the river Impact of landside GW Subsurface Soil characteristics Organic carbon content foc
Brighton, CO South Platte
Cedar Rapids, IA Cedar River
Louisville, KY Ohio River
14 m3/s
106 m3/s
3300 m3/s
Approx. 60%
Negligible
One Vertical 10.7 m Bottom 3 m Continuously at 1700 L/min 10e20 days 120 m Approx. 10e20%
Four Vertical 25e30 m n/a n/a 7 to 10 days 20 to 100 m Negligible
Approx. 1.4% Approx. 24% One Vertical 33 m Bottom 10 m Continuously at 2650 L/min n/a 70 m Approx. 30%
Mostly sand (with some gravel and silt) Approx. 0.05%
Sandy aquifer 0.01%
Glacial sand and gravel aquifer n/a
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
4645
Fig. 1 e Demonstration-scale RBF site, Brighton, Colorado with monitoring well transects and production well (PTW-1).
distribution and organic carbon content. The soil textures in both cases were characterized as alluvial sand. Samples for general water chemical analyses were collected from the river, the production well, and select monitoring wells twice a month throughout the course of the study. Samples for TOrC analysis were collected monthly over a period of three years.
2.2.
Louisville, Kentucky
The Louisville Water Company, Kentucky, operates a main horizontal collector well in addition to vertical wells located on the bank of the Ohio River. The vertical production well (Z-1) and three adjacent monitoring wells (S5, S2, and S4) were selected for this study. The well locations are indicated
in Fig. 2. Wells S5, S2, the production well, and S4 were located approximately 3 m, 35 m, 70 m, and 90 m off the river, respectively. The production well (Z-1) had an inner diameter of 0.75 m and was approximately 33 m deep and 27 m below river level. The well was screened at the bottom 10 m of the aquifer. During the course of this study, the production well was operated continuously at 44 L/s for a period of more than 12 months from Fall 2006 till the end of 2007. The wells were completed in a glacial sand and gravel aquifer. The aquifer was about 20 m thick and characterized as a finite-width strip, which paralleled the Ohio River. A clay unit confined the aquifer and shale and limestone bedrock underlay the aquifer. During the study period, the mean discharge of the river at the site was about 3300 m3/s (Drewes et al., 2009).
4646
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
Table 2 e Comparison of wells at RBF site in Brighton. Depths are reported relative to ground surface. Well PTW1 PTMW-2 PTMW-3 PTMW-4 PTMW-6 PTMW-7 PTMW-8 PTMW-9 PTMW-10 PTMW-11 PTMW-12 PTMW-13 PTMW-14 PTMW-15 PTMW-16 PTMW-17
Depth of well in m
Depth to bedrock in m
Screened interval in m
Diameter in cm
Estimated distance to river in m
10.7 9.4 7.3 8.8 10.1 10.1 10.1 9.8 9.8 8.8 8.5 9.1 10.1 8.2 8.8 5.8
10.7 9.8 9.1 9.1 10.1 10.2 10.1 10.1 9.8 8.8 8.5 9.1 10.1 8.8 9.1 6.7
7.6e10.7 3.7e9.4 1.5e7.3 3.0e8.8 4.0e10.1 4.0e10.1 4.0e10.1 3.7e9.8 3.7e9.8 2.7e8.8 2.4e8.8 2.4e9.1 4.0e10.1 2.1e8.2 2.7e8.8 2.7e5.8
46 5 5 5 5 5 5 5 5 5 5 5 5 10 10 10
114 35 79 110 82 107 114 128 84 46 12 105 73 52 38 15
Samples for general water quality parameters and TOrC were collected from the Ohio River, the monitoring wells S5, S2, and S4, and the production well in April and October of 2007.
2.3.
Cedar Rapids, Iowa
The RBF site in Cedar Rapids consists of both horizontal and vertical production wells generating drinking water for the Cedar Rapids Public Works Department. Four vertical production wells of the Eastern Well Field with different setbacks from the river were chosen for this study. These wells were #12, #5, #11, and #4 and their locations are illustrated in Fig. 3. While wells #11 and #12 were located approximately 20 m off the river, the setback to wells #4 and #5 was approximately 100 m. These wells were operated continuously for the duration of the study (Fall 2006 to Fall 2007). The aquifer at this site was characterized by relatively uniform sandy material, with an approximate depth of 25e30 m. The annual mean daily flow for the river was 106 m3/s (Drewes et al., 2009). Samples for general water quality analysis were collected monthly from the river and the four production wells during winter flow (January to March 2007) and once during summer flow conditions (September 2007). TOrC samples from these locations were collected twice during the course of this study, representing winter and summer flow conditions.
2.4.
Contribution of wastewater to the rivers
The three rivers exhibit different degrees of wastewater discharge upstream of the study sites. The average contribution of wastewater to the South Platte River at the study site has been estimated to be up to 60 percent annually (Krasner et al., 2008). The contribution of wastewater in the Ohio River was evaluated by the U.S. EPA in 1980 (Swayne et al., 1980). This study revealed that 1.4 percent of the yearly average flow in the Ohio River originated from wastewater. Under low flow conditions the contribution can increase to approximately 24 percent. The Cedar River is primarily influenced by agricultural
discharges with rather minor contributions from municipal wastewater effluents (Drewes et al., 2009).
2.5. Contribution of landside groundwater to the composition of the production wells A groundwater flow model to evaluate the dilution of riverbank-filtered water by landside groundwater was only available for the Brighton field site. This model suggests that dilution of the riverbank-filtered water abstracted at the pumping well PTW1 with landside groundwater is less than 20 percent (Drewes et al., 2009). In order to evaluate the dilution of the riverbank-filtered water at the other field sites, conservative and semi-conservative tracer ion concentrations were measured in the river and well samples. A summary of the cation and anion concentrations in river and well water for the Louisville and Cedar Rapids sites is provided in Table 3. At the Louisville site, the concentration of potassium, magnesium, calcium, and sulfate in the Ohio River, the production well (Z-1), and the landside groundwater (S4) were considered. Based on this assessment, the contribution of landside groundwater in the production well was estimated to vary between 20 and 40 percent. At the Cedar Rapids site, no water quality data of a landside monitoring well was available to assess the influence of native groundwater on the production wells. Therefore, the average ion signatures from three sampling campaigns of the river and the production wells were compared (Table 3). This assessment revealed a very similar ion composition suggesting that the majority of the well water originated from the river with negligible dilution from landside groundwater.
3.
Methods
3.1.
Sample collection and handling
Monitoring wells were purged for at least 30 min prior to sampling. Samples were collected in amber glass bottles and
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
Fig. 2 e Full-scale RBF site, Louisville, Kentucky with monitoring well transect and production well (Z-Well).
4647
4648
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
Fig. 3 e Full-scale RBF site, Cedar Rapids, Iowa with four production wells (#4, 5, 11, and 12) selected for this study.
shipped overnight on ice to the CSM laboratory. Bulk parameters were analyzed and TOrC samples extracted within 72 h after receiving.
3.2.
Bulk parameter analysis
Conductivity, pH, temperature, and dissolved oxygen concentrations were measured in situ with handheld instruments. In the laboratory, samples were analyzed for total organic carbon (TOC) and filtered through 0.45 mm to determine the DOC content. Both, TOC and DOC were measured according to Standard Method 5310B using
a Sievers 900 TOC analyzer. Samples for UV absorbance at 254 nm (UVA) analysis were also filtered and the measurement was conducted according to Standard Method 5910B using a Beckman DU800 UV/Vis Scanning Spectrophotometer (Fullerton, CA). The aromaticity of the organic matter in the samples was expressed as specific UV absorbance (SUVA). The SUVA was calculated as ratio between UV-A in 1/m and DOC in mg/L. Inorganic cation concentrations were determined using an Optima 3000 Inductive Coupled Plasma (ICP) Spectrometer (Perkin Elmer, Norwalk, CT) according to Standard Method 3120B. Inorganic anion concentrations were determined using a Dionex DS600 Ion Chromatograph
Table 3 e Conservative ion concentrations for Louisville and Cedar Rapids RBF sites. Louisville Ohio river Production well (Z1) Well (S4) Estimated contribution from the river
Cedar Rapids Cedar river East well 11 East well 4 Dilution
Naþ (mg/L)
Kþ (mg/L)
Ca2þ (mg/L)
Mg2þ (mg/L)
SO42 (mg/L)
3.8 2.2 0.9 45%
43.2 110 134 26%
13.9 26.1 35.3 43%
96 64 57 18%
Kþ (mg/L)
Ca2þ (mg/L)
11.2 2.8 13.1 3.2 12.3 2.9 Similar ion composition at all locations,
Mg2þ (mg/L)
76.4 71.6 70.2 marginal dilution.
22.1 27.2 23
Cl (mg/L)
SO2 4 (mg/L)
24 28 26
31 21 29
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
(IC) (Dionex, Sunnyvale, CA) equipped with an AS14 column and sodium bicarbonate eluent according to Standard Method 4110C.
3.3.
Results and discussion
4.1. Predominant redox conditions and bulk organic carbon removal in the subsurface
Selection of trace organic chemicals (TOrC)
To evaluate the attenuation processes of emerging TOrC in RBF systems, we selected mainly well water soluble, polar TOrC based on their prevalent occurrence in surface waters and their potential to be considered as performance indicator compounds in RBF systems. Most of these indicator compounds are biotransformed in subsurface systems under different redox conditions. A summary of the classes, physicochemical properties, and occurrence of the selected compounds observed in streams in Europe and the U.S. is presented in Table 4.
3.4.
4.
4649
Trace organic chemical analysis
All standards and derivatization agents were purchased from Sigma-Aldrich in the highest grade available with the exception of the chlorinated flame retardant, TCPP which was acquired from Pfaltz and Bauer. The isotope labeled surrogate standard ibuprofen-13C3 was purchased from Cambridge Isotope Lab, Andover, MA. Diclofenac-d4 and bisphenol A-d16 were acquired from C/D/N, Quebec, Canada. The TOrC analysis in this study was performed by adopting and slightly modifying two previously published methods by Reddersen and Heberer (2003). For the analysis of the target compounds, 0.25 to 0.5 L of each sample was collected and acidified to pH 2 using residue free hydrochloric acid. For increased accuracy, six surrogate standards, 100 ng of each of the following compounds were added to each sample: 2-(m-chlorophenoxy) propionic acid, ibuprofen-13C3, diclofenac-d4, dihydrocarbamazepine, carbamazepine-d10, and bisphenol A-d16. SPE was carried out by using 1 g of RP-C-18 material (Bakerbond Polar Plus, MallinckrodtBaker, Phillipsburg, NJ) filled in a 6 mL polyethylene cartridge. The cartridges conditioning and extraction of the samples was executed as described by Reddersen and Heberer (2003). For method A, the analytes were eluted from the cartridges one time with 3 mL of acetone, dried, and redissolved in 100 mL of a pentafluorobenzyl bromide (PFBBr) solution (2% in toluene). A volume of 4 mL of triethylamine was added as a catalyst into the sample vial, which was then capped and placed in a drying cabinet for 1 h at 100 C. The residue was dissolved again in toluene (100 mL) and analyzed by a HP 6890 gas chromatograph equipped with a HP 5973 single quadrupole mass spectrometer from Agilent Technologies (Palo Alto, CA). For method B, the analytes were eluted from the cartridges twice with 3 mL each of methanol, dried, and redissolved in 50 mL acetonitrile and 50 mL of N-methyl-N-[tertbutyldimethylsilyl] trifluoroacetamide (MTBSTFA). The vials were capped and placed in a drying cabinet for 1 h at 80 C. The samples were analyzed by a HP 6890 gas chromatograph equipped with a HP 5973 single quadrupole mass spectrometer from Agilent Technologies (Palo Alto, CA).
In order to investigate the change of redox conditions during RBF, samples from three field sites were collected. Dissolved oxygen (DO) measurements were only conducted at the Brighton, CO site. The South Platte River exhibited long-term average DO concentrations of 9.5 2.5 mg/L (Fig. 4). The average nitrate and total manganese concentrations in the river were 4.2 1.5 mg-N/L and 0.2 0.2 mg/L, respectively, representing a well-oxidized source water. Within the initial phase of infiltration and travel times of 2e5 days in the subsurface, the DO concentrations dropped to less than 1 mg/L and the average nitrate concentration decreased to less than 2.5 mg-N/L revealing the presence of reduced redox conditions in the groundwater. During additional travel in the subsurface, the DO and nitrate concentrations remained low and the manganese concentration increased to 0.4 mg/L by the time the water reached well PTMW4, representing the monitoring well closest to the production well. Because the production well PTW1 represents a blend of waters with different flow paths and is also partially impacted by landside groundwater, the DO concentration in this well increased to 1.6 mg/L. Both, nitrate and manganese, exhibited a slight decrease in concentrations in the production well, still indicating the presence of predominantly reduced conditions in the groundwater. The TOC in the South Platte River exhibited average concentrations of 7.6 mg/L (Fig. 4). During the initial phase of infiltration, the TOC concentration in the groundwater decreased to 4.3 mg/L in monitoring well PTMW2. Additional travel in the subsurface further decreased the TOC concentration to 3.7 mg/L at monitoring well PTMW4 representing travel times of 25 days. In addition to this reduction in organic matter during subsequent travel, the SUVA values increased slightly from 1.8 L/(mg m) in the river to 2.4 L/(mg m) in well PTMW4 indicating a preferred transformation of nonaromatic organic matter during RBF. A similar trend in shifting redox conditions from oxic in the river to anoxic during RBF at the Brighton site was also observed for both, the Cedar Rapids and the Louisville sites. Average concentrations of DO, nitrate, and manganese over a period of one year in the source and RBF treated water are presented for all three sites in Fig. 5. The DO concentrations in the Cedar River and the Ohio River were not measured. According to the Ohio River Valley Water Sanitation Commission (ORSANCO), the average DO concentration for the Ohio River between May and October 2007 at Cannelton downstream of the RBF site was 6.23 mg/L (http://www. orsanco.org/). The average nitrate concentration of 9.5 mgN/L in the Cedar River was the highest among the three rivers. RBF at the Cedar River resulted in nitrate concentrations of less than 1 mg-N/L. Manganese concentrations at this site increased from less than 0.05 mg/L to 1.2 mg/L revealing the presence of reducing conditions. It is noteworthy, that the amount of DOC removed in the subsurface cannot entirely account for the removal of more than 8 mg-N/L (assuming
4650
Table 4 e Trace organic chemicals selected for this study and their occurrence in European and U.S. Rivers. Compound
Compound class
Log P (Log DpH¼7)
Occurrence in European Rivers (ng/L) Elbe: <1e66e Taff: 11e68h Ely: <6e22h Rhine: 23e300a,b Elbe: <1e70e Lippe: <10e1700d Taff: 19e327h Ely: 9e647h Thongue River: 23.6j Rhine: <10e20b,c Elbe: <10e20e Main: <10e30c Lippe: <10e290d Taff: < 0.3e3h Ely: < 0.3e5h
Bisphenol A
Plasticizer
3.43 (3.43)
50 to 90%
Carbamazepine
Anti-epileptic drug
2.67 (2.67)
<25%
Clofibric acid
Metabolite of the blood lipid regulating agents clofibrate
2.72 (0.82)
<25%
Dichlorprop Diclofenac
Pesticide Non-steroidal anti-inflammatory drug
2.93 (1.07) 4.06 (1.06)
50 to 90% 50 to 90%
Fenofibrate
Blood lipid regulating agent Blood lipid regulating agent
4.80 (4.80)
50 to 90%
4.39 (2.19)
50 to 90%
Non-steroidal anti-inflammatory drug
3.72 (1.12)
50 to 90%
Gemfibrozil
Ibuprofen
Ketoprofen
Non-steroidal anti-inflammatory drug
2.81 (0.01)
50 to 90%
Mecoprop
Pesticide
2.84 (0.38)
50 to 90%
Rhine: <10e650a,b,c Elbe: <1e40e Main: 70e140c Lippe: <10e1800d Taff: 9e40h Ely: <0.5e119h Thongue River: 10.5j Rhine: <10c Main: <10c Rhine: <5e90a,b Elbe: <2e8e Main: <10e30c Thongue River: 1.0j Rhine: <5e70a,b,c Elbe: <2e70e Main: <10e20c Taff: 12e62h Ely: <0.3e74 h Thongue River: 3.5j Rhine: <10e25b Main: <20c Taff: < 0.5e7h Ely: < 0.5e12h Thongue River: 2.8j
Occurrence in US Rivers (ng/L) Detroit: 1e18g
Detroit: 0.1e2g IA: 170 MN: 90
Detroit: 2e12g
Detroit: 0.2e4g
Detroit: 2e42g West Prong Little Pigeon River: <10e113i
Detroit: 0.4e0.8g
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
Biodegradability based on previous subsurface studies (10 days of travel time)
Naproxen
Non-steroidal anti-inflammatory drug
Primidone TCEP
Anti-epileptic drug Flame Retardant
TCPP TDCPP
Flame Retardant Flame Retardant
3.00 (0.00)
50 to 90%
0.84 (0.84) 0.48 (0.48)
<25% 25 to 50%
1.53 (1.53) 1.79 (1.79)
<25% <25%
Rhine: <10e50b,c Elbe: <1e6e Taff: 17e146 h Ely: <0.3e113h Thongue River: 7.2j Rhine: <5e20b Rhine: 50e1000f Lippe: <10e240d Rhine: 30e150f N/A
Detroit: 1e14g
n/a IA: 220 MN: 75 IA: 200 MN: 60
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
Note: Average discharge of European rivers (wastewater impact): Rhine River 2300 m3/s (5e10%); Main River 160 m3/s; Elbe River 720 m3/s; Lippe River 46 m3/s; Taff River: 15.3 m3/s (80%); Ely River: 6.1 m3/s (69%); Thongue River: 1 m3/s (2e20%). Average discharge (wastewater impact) of U.S. rivers at the sampling location; West Prong Little Pigeon River: 8 m3/s (<10%); Detroit River: <5%; IA: 0.14 m3/s (93%); MN: 4.6 m3/s (13%). a Weil and Knepper (2006). b Walraven and Laane (2008). c Ternes (1998). d Dsikowitzky et al. (2004). e Wiegel et al. (2004). f Knepper et al. (1999). g Tabe et al. (2009). h Kasprzyk-Hordern et al. (2008). i Yu and Chu (2009). j Rabiet et al. (2006).
4651
4652
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
Fig. 4 e Change of redox parameters, TOC, and SUVA during RBF at the Brighton site (River: S. Platte; Monitoring wells: PTMW2, 3, and 4; Production well: PTW1).
a one mol/mol ratio between carbon and nitrogen needed for denitrification). It is likely that particulate organic carbon (POC) in the source water and sediments may have served as an additional carbon source for the microbial community engaged in denitrification. The nitrate concentration in the Ohio River was 1 mg-N/L and was reduced to 0.2 mg-N/L in the RBF treated water. Manganese concentrations in the RBF treated water also increased from less than 0.05 mg/L in the Ohio River to 1.2 mg/ L. It is noteworthy that the contribution of landside groundwater in this pumping well was considerably higher (see Table 3) than at the other sites. The average concentrations of TOC and SUVA for the source water and RBF treated water samples are compared among the three sites in Fig. 5. The South Platte River exhibited an average TOC concentration of 7.5 mg/L with
a standard deviation of 3.4 mg/L and an average SUVA value of 1.82 L/(mg m). The Cedar River exhibited also highly variable TOC concentration with an average of 3.8 mg/L with a standard deviation of 2.5 mg/L. The average SUVA value was determined to be 2.28 L/(mg m) with a standard deviation of 0.38 L/(mg m). During the two sampling campaigns conducted at the Louisville site, the Ohio River water quality parameters remained unchanged and the TOC concentration averaged 3.0 mg/L and the SUVA exhibited 3.05 L/(mg m). The South Platte River, therefore, exhibited not only the highest TOC concentration, but also the lowest SUVA value indicated the likely presence of the highest non-aromatic and potentially biodegradable fraction of organic carbon of the three rivers. During the subsurface travel, this organic carbon fraction can be potentially removed more easily than refractory organic carbon. Indeed, the biodegradable DOC (BDOC) at the Brighton
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
4653
Fig. 5 e Redox indicators, TOC concentrations, and SUVA in river and RBF treated water for the three field sites. Note: DO was not measured at Cedar Rapids and Louisville; error bars present standard deviations.
site was determined to be 4 mg/L as compared to approximately 2 mg/L at the Cedar Rapids site with increasing SUVA values after RBF indicating the preferred removal of nonaromatic carbon. While the TOC at the Brighton site was
removed by 52 percent, the organic matter at the Cedar Rapids site, which was more humic in character, was also reduced by 47 percent (Fig. 5). It is noteworthy that the Louisville site exhibited a different behavior with a TOC that was
4654
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
characterized by the highest SUVA value in the river water among the three sites. RBF at this site resulted in a TOC attenuation of 67 percent with a slight decrease in SUVA suggesting that landside (humic-rich) groundwater exhibiting lower TOC concentrations likely diluted the river water in the RBF well. The observed shift in redox conditions from oxic to anoxic during RBF at all sites and the rather rapid removal of TOC during the initial phase of infiltration is consistent with RBF studies conducted in Europe (Ziegler, 2001; Gru¨nheid et al., 2005; Massmann et al., 2008) and SAT studies performed in the U.S. (Fox et al., 2000; Rauch-Williams and Drewes, 2006).
4.2. Attenuation of bulk organic carbon during different seasons and flow conditions During the study period, the TOC concentrations in the South Platte River and the RBF treated water were monitored and depended highly both on season and river flow (Fig. 6). The TOC concentration in the river varied between 5 and 11 mg/L as a function of different flow conditions in the river, with higher TOC concentrations during low flow conditions. In order to evaluate the seasonal variability in performance and robustness of the RBF system at the Brighton site, groundwater wells in two transects were monitored bimonthly for water quality changes regarding bulk parameters. For one of these transects, seasonal temperature measurements of the South Platte River and the RBF waters are presented in Fig. 7. During the winter sampling campaigns the temperature in the river was 9.3 C on average with a standard deviation of 3.2 C. In the monitoring wells, the temperature averaged approximately 8 C 1 C. An increase in temperature to 11.3 C 1.9 C was observed in the production well. During summer sampling campaigns the temperature in the river, monitoring wells, and production well increased to 19.7 C, 19.0 C, and 15.9 C on average,
respectively. Overall, a greater temperature gradient was observed in the river for different seasons than in the wells. During winter low flow conditions the average TOC in the South Platte River exhibited an average concentration of 10.5 mg/L (Fig. 7). In the spring and summer, the TOC concentration was 8 mg/L on average. These results were confirmed by SUVA values, which increased between the winter and summer months from 1.53 to 2.22 L/(mg m), respectively. The DO concentrations in the river varied between 8 and 10 mg/L during summer and winter months, respectively. The majority of TOC attenuation during RBF occurred very rapidly in the initial phase of infiltration after about five days of travel time (to monitoring well PTMW2). During the summer high flow conditions, TOC removal in the initial phase of infiltration was slightly better as compared to winter low flow conditions, which is consistent with observations reported by Gru¨nheid et al. (2008), who observed only a small temperature influence on biotransformation of organic matter in a similar temperature range. During the initial phase of infiltration, the DO in the infiltrating river water was almost completely consumed by microorganisms. Similar results have been reported by Massmann et al. (2008), where DO during RBF was completely depleted after 0.2 m of infiltration. The change of redox conditions at the Brighton site is further supported by the depletion of nitrate in well PTMW2 by up to 22 percent during the winter months and between 54 and 67 percent during the spring and summer months. A further depletion of nitrate as well as the slight increase in manganese concentration in the subsequent wells and the production well support the observed shift from oxic to anoxic conditions in the subsurface. The TOC in the monitoring and production wells exhibited concentrations of 3.1e4 mg/L and never exceeded 4.5 mg/L independent of season and location. Similar to findings reported by Ku¨hn and Mu¨ller (2000), the subsurface acted as a buffer and compensated the highly variable concentrations of TOC and nitrate as well as
Fig. 6 e Flow variations and TOC concentrations of river and RBF treated water for the Brighton RBF site.
4655
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
Fig. 7 e Seasonal variations, redox conditions, and organic carbon concentration for a transect at the RBF-site in Brighton, CO (left to right: winter low flow conditions, summer high flow conditions).
temperature in the South Platte River providing a consistent water quality after RBF.
4.3.
TOrC occurrence in surface waters
TOrC were monitored at all three study sites in the river and groundwater well samples. Average TOrC concentrations and
frequencies for the rivers and the production wells are summarized in Table 5. The results of selected TOrC in the river for the Brighton site for two different seasons are also presented in Fig. 8. None of the target TOrC was detected in the Ohio or Cedar River during any sampling campaign. As such, none of the TOrC was detected in the wells. Unlike the Cedar and Ohio Rivers, 11 out of 16 target compounds were
Table 5 e Occurrence ranges and detection frequency of selected TOrC in river and RBF treated water samples. Concentration in ng/L
Bisphenol A Carbamazepine Clofibric acid Diclofenac Dichlorprop Fenofibrate Gemfibrozil Ibuprofen Ketoprofen Mecoprop Naproxen Primidone TCEP TCPP TDCPP
Brighton
Louisville
Cedar Rapids River
River
RBF
River
RBF
River
RBF
70e210 (6/6) 245e900 (8/8) <10 (0/11) <10e54 (7/11) <25e60 (2/11) <25 (0/11) 45e1105 (18/18) 70e1215 (18/18) <25 (0/11) <25e370 (10/11) 40e790 (18/18) 75e225 (8/8) 100e640 (18/18) 350e1450 (18/18) 70e405 (18/18)
<25e100 (3/4) <50e885 (8/9) <10 (0/13) <10 (8/13) <25 (1/13) <25 (0/13) <25e395 (15/16) <25e105 (11/16) <25 (0/13) <25e60 (9/13) <25e100 (16/16) <25e110 (6/6) 50e835 (16/16) 125e1490 (16/16) 85e335 (15/15)
<25 (0/2) <25 (0/2) <10 (0/2) <10 (0/2) <10 (0/2) <25 (0/2) <25 (0/2) <10 (0/2) <25 (0/2) <10 (0/2) <10 (0/2) <10 (0/2) <25 (0/2) <25 (0/2) <25 (0/2)
<25 (0/2) <25 (0/2) <10 (0/2) <10 (0/2) <10 (0/2) <25 (0/2) <25 (0/2) <10 (0/2) <25 (0/2) <10 (0/2) <10 (0/2) <10 (0/2) <25 (0/2) <25 (0/2) <25 (0/2)
<25 (0/2) <25 (0/2) <10 (0/2) <10 (0/2) <10 (0/2) <25 (0/2) <25 (0/2) <10 (0/2) <25 (0/2) <10 (0/2) <10 (0/2) <10 (0/2) <25 (0/2) <50 (0/2) <25 (0/2)
<25 (0/2) <25 (0/2) <10 (0/2) <10 (0/2) <10 (0/2) <25 (0/2) <25 (0/2) <10 (0/2) <25 (0/2) <10 (0/2) <10 (0/2) <10 (0/2) <25 (0/2) <50 (0/2) <25 (0/2)
4656
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
Fig. 8 e Occurrence of TOrC in the South Platte River (a), PTMW2 (b), and PTW1 (c) during winter (left) and summer (right). The flow during both seasons ranged between 4.7 and 16.7 m3/s.
detected in the South Platte River, which is characterized by the highest degree of wastewater impact. The absence of TOrC in the Ohio and Cedar Rivers was surprising. Dilution is most likely the explanation for their absence, considering the different flow regimes and the degree of wastewater impacts in the three rivers (Table 1). The RBF site at the Cedar River studied here is located in an agricultural region with rather low impact from municipal wastewater discharges. The Ohio River has with 1.4% (yearly average) the lowest wastewater impact of the three rivers studied (Swayne et al., 1980). Even though these numbers have been established in the late 1970s, an improvement in wastewater treatment might have resulted in an improved wastewater quality. While the Rhine River in Europe exhibits a flow rate of 2300 m3/s, which is of similar size like the Ohio River, the occurrence pattern of TOrC is different. TOrC occurrence data in the Rhine River and other European and U.S. rivers is summarized in Table 4. Most of the targeted TOrC in this study have been reported to occur in the Rhine River and other European rivers. The concentrations are comparable to those observed in the South Platte River, although its
degree of impact from wastewater discharge is up to 60% as compared to 5 to 10% impact for the Rhine River (according to oral communication with Prof. Heinz-Ju¨rgen Brauch, TZW Karlsruhe, Germany). One possible explanation for the relatively high concentration in the Rhine River is the difference in per capita water consumption between Europe and the U.S. The average water consumption in Europe is approximately half to a third of the U.S. (BDEW, 2009), which will also result in more concentrated wastewater in Europe as compared to the U.S. and, therefore, also in higher concentrations of TOrC in European rivers receiving a similar degree of wastewater discharge. Similar flow regimes, between 4.7 and 16.7 m3/s, but different seasons were chosen to compare the seasonal variability on TOrC concentrations in the South Platte River (Fig. 8a). The TOrC concentrations of individual compounds in the river ranged between 10 and 1450 ng/L during both seasons. The well degradable indicator compounds, such as gemfibrozil, ibuprofen, and naproxen, exhibited significantly lower concentrations during the summer as compared to the winter sampling campaigns. It is hypothesized that the higher
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
temperatures during the summer months enhance the removal efficiency of biodegradable compounds in wastewater treatment plants and the river. The occurrence of conservative TOrC, such as the antiepileptic drugs and chlorinated flame retardants, were not affected by different seasons (Fig. 8a). In general, the concentration levels of the target TOrC observed in the South Platte River were similar to those reported by other researcher in U.S. and European secondary/tertiary treated wastewaters or surface water heavily impacted by wastewater discharge (Kolpin et al., 2002; Drewes et al., 2002; Dickenson et al., in review). The blood lipid regulator compound gemfibrozil was detected in the South Platte River during all sampling campaigns with concentrations varying between 400 ng/L and 765 ng/L during summer and winter campaigns, respectively. Kolpin et al. (2002) reported a maximum concentration of 790 ng/L in impaired surface water. Blood lipid regulators and metabolites, such as clofibric acid and fenofibrate, were not detected in the South Platte River water. The concentrations of the anti-epileptic drugs carbamazepine and primidone averaged at 120 and 260 ng/L in the South Platte River. These concentrations are comparable to those reported to occur in secondary and tertiary treated effluent in the U.S. (Drewes et al., 2002). Carbamazepine concentrations in European rivers reached similar concentrations (Table 4). The non-steroidal anti-inflammatory drugs, ibuprofen and naproxen, exhibited average concentrations of 1215 and 790 ng/L during winter campaigns, respectively. Concentrations during summer campaigns decreased to 70 and 40 ng/L, respectively. Similar concentrations were measured in wastewater impacted streams in Europe and the U.S. (Table 4).
4.4.
TOrC attenuation during RBF
Average TOrC concentrations of the river and two wells for the Brighton site are presented in Fig. 8 for winter and summer
4657
sampling campaigns. In Fig. 9, TOrC measurements from single campaigns are presented for an entire transect for one winter and one summer campaign. The 11 targeted TOrC detected in the river were also detected in the production well, but concentrations were significantly lower for compounds that have been reported to be attenuated during subsurface treatment. During summer months, the concentrations of TOrC, such as naproxen, ibuprofen, and gemfibrozil, decreased to less than 50 ng/L after a travel time of 5 days (PTWM2) (Figs. 8 and 9). Additional travel in the subsurface did not result in additional attenuation and gemfibrozil, ibuprofen, and naproxen exhibited threshold concentrations between 10 and 25 ng/L, respectively. During winter months, attenuation of these biodegradable compounds also occurred within 5 days, but the removal was hindered and additional travel times up to 25 days were needed to further reduce the concentrations. Concentrations after extended travel time (25 days) were consistently higher during winter months, which suggests that the average temperature of less than 10 C in the winter resulted in a reduced attenuation of TOrC (Fig. 9). Gru¨nheid et al. (2008) conducted controlled laboratory studies and observed that the attenuation of TOrC in the subsurface was severely diminished at temperatures at 5 C as compared to 15 C. No reduction in concentration during RBF was observed for the anti-epileptic drugs primidone and carbamazepine. The slight changes in carbamazepine concentrations between river and RBF treated water are likely due to retardation in the subsurface. Previous studies have suggested varying degrees of adsorption of carbamazepine onto soil and wastewater solids and no retardation of primidone (Scheytt et al., 2005; Ternes, et al., 1998; Drillia et al., 2005; Drewes et al., 2009). In field and column studies, carbamazepine and primidone have been found to be very persistent during MAR with dissipation times of years (Drewes et al., 2003; Lo¨ffler et al., 2005). Therefore, primidone can serve as
Fig. 9 e TOrC concentrations in one transect at the South Platte River site at different seasons.
4658
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
a conservative tracer to quantify the degree of wastewater impact in subsurface systems. The chlorinated flame retardants TCEP, TCPP, and TDCPP did not change significantly in concentration between the river and the production well. Like primidone and carbamazepine, the chlorinated flame retardants have been found to be very persistent in subsurface systems (Drewes et al., 2003).
5.
Conclusions
Temperature (seasonal) and flow variation played an important role for the make-up of the river water with respect to TOC, nitrogen, and TOrC concentrations. However, the buffering capability of the subsurface was sufficient to produce a consistent and predictable riverbank-filtered water quality. TOC and SUVA values correlated well with the wastewater impact in the three streams with lower SUVA and higher TOC values in the highest impacted stream. During RBF, removal of TOC occurred within the initial phase of infiltration or 2 to 5 days of travel time preferably attenuating non-aromatic carbon. TOC concentrations decreased slightly during additional travel in the subsurface. Among the three river systems studied, TOrC were only detected in the South Platte River at concentration in the ppt to low ppb range. These concentrations were comparable to those observed in European Rivers, especially the River Rhine, although their degree of wastewater impact is significantly lower. Differences in water consumption per capita by factor 2e3 might explain the observed differences in occurrence resulting in more concentrated wastewater in Europe and therefore higher occurrence of TOrC in river water while exhibiting a lower degree of impact from wastewater discharge. None of the select TOrC was detected in the Ohio or Cedar Rivers due to their low wastewater impact. The performance of RBF during the winter with water temperatures of less than 10 C in the subsurface did not significantly affect TOC removal but resulted in a diminished denitrification rate and a reduced attenuation of TOrC. While significant attenuation was observed for select TOrC after 2 to 5 days of travel time during winter and summer months, travel times of more than 25 days were needed in the winter to reduce concentrations to less than 100 ng/L for these compounds. These findings confirm that RBF systems are able to act as a reliable barrier for TOC, nitrogen, and select TOrC regardless of seasonal and flow conditions in the river providing a sufficient retention time is maintained.
Acknowledgments The authors thank the Water Research Foundation (formerly AwwaRF) for its financial, technical, and administrative assistance in funding and managing the project through which this information was derived. The comments and views detailed herein may not necessarily reflect the views of the Water Research Foundation, its officers, directors, affiliates, or agents. The authors are grateful for the support provided by the City of Aurora, Louisville Water Company,
Cedar Rapids, and CH2 M Hill. The authors would also like to acknowledge assistance from Dr. Steve Hubbs, Louisville, KY, Ted Hartfelder, Aurora, CO, and Katie Chamberlain, Dr. Eric Dickenson, Dr. Dean Heil, and Professor Dr. John McCray at the Colorado School of Mines.
references
Benotti, M.J., Trenholm, R.A., Vanderford, B.J., Holady, J.C., Stanford, B.D., Snyder, S.A., 2009. Pharmaceuticals and endocrine disrupting compounds in U.S. drinking water. Environmental Science and Technology 43 (3), 597e603. Bundesverband der Energie und Wasserwirtschaft (BDEW), 2009. Overview on Water Facts Berlin (in German). Clara, M., Strenn, B., Gans, O., Martinez, E., Kreuzinger, N., Kroiss, C., 2005. Removal of selected pharmaceuticals, fragrances and endocrine disrupting compounds in a membrane bioreactor and conventional wastewater treatment plants. Water Research 39, 4797e4807. Dickenson, E.R.V., Drewes, J.E., Snyder, S., Sedlak, D. Indicator compounds for assessment of wastewater effluent contributions to flow and water quality. Water Research, in review. Drewes, J.E., Heberer, T., Reddersen, K., 2002. Fate of pharmaceuticals during indirect potable reuse. Water Science and Technology 46 (3), 73e80. Drewes, J.E., Heberer, T., Rauch, T., Reddersen, K., 2003. Fate of pharmaceuticals during ground water recharge. Ground Water Monitoring and Remediation 23 (3), 64e72. Drewes, J.E., Hoppe, C., Oldham, G., McCray, J., Thompson, K., 2009. Evaluation of Riverbank Filtration Systems to Optimize Removal of Bulk Organic Matter, Emerging Organic Micropollutants, and Nutrients. Final Report. Water Research Foundation, Denver, Colorado. Drewes, J.E., Khan, S., 2010. Water reuse for drinking water augmentation. In: Edzwald, J. (Ed.), Water Quality and Treatment, sixth ed. American Water Works Association, Denver, Colorado. Drillia, P., Stamatelatou, K., Lyberatos, G., 2005. Fate and mobility of pharmaceuticals in solid matrices. Chemosphere 60 (8), 1034e1044. Dsikowitzky, L., Schwarzbauer, J., Kronimus, A., Littke, R., 2004. The anthropogenic contribution to the organic load of the Lippe River (Germany). Part I: qualitative characterisation of low-molecular weight organic compounds. Chemosphere 57, 1275e1288. Focazio, M.J., Kolpin, D.W., Barnes, K.K., Furlong, E.T., Meyer, M.T., Zaugg, S.D., Barber, L.B., Thurman, M.E., 2008. A national reconnaissance for pharmaceuticals and other organic wastewater contaminants in the United States d II) Untreated drinking water sources. Science of the Total Environment 402 (2e3), 201e216. Fox, P., Narayanaswamy, K., Genz, A., Drewes, J.E., 2000. Water quality transformations during soil-aquifer treatment at the Mesa Northwest water reclamation plant, USA. Water Science and Technology 43 (10), 343e350. Glassmeyer, S.T., Furlong, E.T., Kolpin, D.W., Cahill, J.D., Zaugg, S. D., Werner, S.L., Meyer, M.T., Kryak, D.D., 2005. Transport of chemical and microbial compounds from known wastewater discharges: potential for use indicators of human fecal contamination. Environmental Science and Technology 39 (14), 5157e5169. Gru¨nheid, S., Amy, G., Jekel, M., 2005. Removal of bulk dissolved organic carbon (DOC) and trace organic compounds by bank filtration and artificial recharge. Water Research 39, 3219e3228. Gru¨nheid, S., Hu¨bner, U., Jekel, M., 2008. Impact of temperature on biodegradation of bulk and trace organics during soil
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 4 3 e4 6 5 9
passage in an indirect potable reuse system. Water Science and Technology 57 (7), 987e994. Kasprzyk-Hordern, B., Dinsdale, R.M., Guwy, A.J., 2008. The occurrence of pharmaceuticals, personal care products, endocrine disruptors and illicit drugs in surface water in South Wales, UK. Water Research 42, 3498e3518. Knepper, T.P., Sacher, F., Lange, F.T., Brauch, H.J., Karrenbrock, F., Roerden, O., Lindner, K., 1999. Detection of polar organic substances relevant for drinking water. Waste Management 19, 77e99. Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S. D., Barber, L.B., Buxton, H.T., 2002. Pharmaceuticals, hormones, and other organic wastewater contaminants in U. S. streams, 1999 & 2000: a national reconnaissance. Environmental Science and Technology 36 (6), 1202e1211. Krasner, S., Westerhoff, P., Chen, B., Amy, G., Nam, S., Chowdhury, Z., Sinha, S., Rittmann, B., 2008. Contribution of Wastewater to DBP Formation. Final Report. Water Research Foundation, Denver, Colorado. Ku¨hn, W., Mu¨ller, U., 2000. Riverbank filtration e an overview. Journal of the American Water Works Association 92 (12), 60e69. Lo¨ffler, D., Ro¨mbke, J., Meller, M., Ternes, T.A., 2005. Environmental fate of pharmaceuticals in water/sediment systems. Environmental Science and Technology 39 (14), 5209e5218. Massmann, G., Du¨nnbier, U., Heberer, T., Taute, T., 2008. Behaviour and redox sensitivity of pharmaceutical residues during bank filtration e investigations of residues of phenazone-type analgesics. Chemosphere 71, 1476e1485. Rabiet, M., Togola, A., Brissaud, F., Seidel, J.-L., Budzinski, H., Elbaz-Poulichet, F., 2006. Consequences of Treated Water Recycling as Regards Pharmaceuticals and Drugs in Surface and Ground Waters of a Medium-sized Mediterranean Catchment. Environmental Science & Technology 40 (17), 5282e5288. Rauch-Williams, T., Drewes, J.E., 2006. Using soil biomass as an indicator for the biological removal of effluent-derived organic carbon during soil infiltration. Water Research 40, 961e968. Reddersen, K., Heberer, T., 2003. Multi-compound methods for the detection of pharmaceutical residues in various waters applying solid phase extraction (SPE) and gas chromatography with mass spectrometric (GC-MS) detection. Journal of Separation Science 26 (15e16), 1443e1450. Schwab, B.W., Hayes, E.P., Fiori, J.M., Mastrocco, F.J., Roden, N.M., Cragin, D., Meyerhoff, R.D., D’Aco, V.J., Anderson, P.D., 2005. Human pharmaceuticals in US surface waters: a human
4659
health risk assessment. Regulatory Toxicology and Pharmacology 42 (2005), 296e312. Scheytt, T., Mersmann, P., Lindstadt, R., Heberer, T., 2005. Determination of sorption coefficients of pharmaceutically active substances carbamazepine, diclofenac, and ibuprofen, in sandy sediments. Chemosphere 60 (2), 245e253. Snyder, S.A., 2008. Occurrence, treatment, and toxicological relevance of EDCs and pharmaceuticals in water. Ozone: Science and Engineering 30, 65e69. Sontheimer, H., 1991. Trinkwasser aus dem Rhein? Academia Verlag, Sankt Augustin, Germany (in German). Stumm-Zollinger, E., Fair, G.M., 1965. Biodegradation of steroid hormones. Journal of the American Water Works Association 37, 1506e1510. Swayne, M.D., Boone, G.H., Bauer, D., Lee, J.S., 1980. Wastewater in Receiving Waters at Water Supply Abstraction Points. Tech. Rep. EPA-600-2-80-044. EPA. Tabak, H.H., Bunch, R.L., 1970. Steroid hormones as water pollutants. I. Metabolism of natural and synthetic ovulationinhibiting hormones by microorganisms of activated sludge and primary settled sewage. Developments in Industrial Microbiology 11, 367e376. Tabe, S., Jamal, T., Seth, R., Yue, C., Yang, P., Zhao, X., Schweitzer, L., 2009. PPCPs and EDCs e Occurrence in the Detroit River and Their Removal by Ozonation. Final Report. Water Research Foundation, Denver, Colorado. Ternes, T.A., 1998. Occurrence of drugs in German sewage treatment plants and rivers. Water Research 32 (11), 3245e3260. Walraven, W., Laane, R.W.P.M., 2008. Assessing the discharge of pharmaceuticals along the Dutch coast of the North Sea. In: Whitacre, D.M. (Ed.), Reviews of Environmental Contamination and Toxicology, vol. 199 (Chapter 1). Wiegel, S., Aulinger, A., Brockmeyer, R., Harms, H., Lo¨ffler, J., Reincke, H., Schmidt, R., Stachel, B., von Tu¨mpling, W., Wanke, A., 2004. Pharmaceuticals in the river Elbe and its tributaries. Chemosphere 57 (2), 107e126. Weil, H., Knepper, T.P., 2006. Pharmaceuticals in the River Rhine. Handbook Environmental Chemistry. Part L Vol. 5, 177e184. Yu, C.-P., Chu, K.-H., 2009. Occurrence of pharmaceuticals and personal care products along the west Prong Little Pigeon River in east Tennessee, USA. Chemosphere 75, 1281e1286. Ziegler, D., 2001. Untersuchung zur nachhaltigen Wirkung der Uferfiltration im Wasserkreislauf Berlins. PhD dissertation. Technical University of Berlin. http://opus.kobv.de/tuberlin/ volltexte/2001/321/pdf/ziegler_doerte.pdf Germany(in German).
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 2 9 e4 5 3 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Changes in polycyclic aromatic hydrocarbon availability in River Tyne sediment following bioremediation treatments or activated carbon amendment Sarah E. Hale, Paola Meynet, Russell J. Davenport, D. Martin Jones, David Werner* School of Civil Engineering and Geosciences, Newcastle University, Newcastle upon Tyne, NE1 7RU, United Kingdom
article info
abstract
Article history:
Bioremediation and activated carbon (AC) amendment were compared as remediation
Received 18 December 2009
strategies for sediment from the River Tyne containing 16.4 7.3 mg/g polycyclic aromatic
Received in revised form
hydrocarbons (PAHs) and approximately 5% coal particles by total dry sediment weight.
3 June 2010
Unamended, nutrient amended (biostimulated) and nutrient and Pseudomonas putida
Accepted 11 June 2010
amended (bioaugmented) sediment microcosms failed to show a significant decrease in
Available online 19 June 2010
total sediment PAH concentrations over a one month period. Polyethylene passive (PE) samplers were embedded for 21 days in these sediment microcosms in order to measure
Keywords:
the available portion of PAHs and accumulated 4.70 0.25, 12.43 1.78, and 23.49 2.73 mg
Sediment pollution
PAHs/g PE from the unamended, biostimulated, and bioaugmented microcosms, respec-
Bioremediation
tively. Higher PAH uptake by PE samplers in biostimulated and bioaugmented microcosms
Denaturing gradient gel electropho-
coincided with slower degradation of spiked phenanthrene in sediment-free filtrate from
resis
these microcosms compared to filtrate from the unamended microcosms. Microbial
Microbial community analysis
community analysis revealed changes in the bacterial community directly following the
Activated carbon amendment
addition of nutrients, but the added P. putida community failed to establish itself. Addition of 2% by dry sediment weight activated carbon reduced PAH uptake by PE samplers to 0.28 0.01 mg PAHs/g PE, a greater than 90% reduction compared to the unamended microcosms. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
The Tyne estuary in the Northeast of England was a major centre of the British economy from the seventeenth to early twentieth century. Industrial activities have resulted in legacy sediment and water pollution which includes polycyclic aromatic hydrocarbons (PAHs) (Woodhead et al., 1999). Studies on Platichthys flesus, Pomatoschistus minutus and Zoarces viviparus from the River Tyne have demonstrated ecotoxic effects caused by this pollution (Kirby et al., 1999, Stentiford et al., 2003). A primary goal of the Water Framework
Directive is to achieve good ecological and chemical status for water bodies in the European Union (European Parliament and Council of the European Union, 2000) and currently, the most prevalent remediation approach for polluted sites such as the River Tyne is to remove contaminated sediment to confined licensed disposal sites. Dredged sediments may be disposed of at sea without further testing only when the concentration of individual PAHs is below 0.1 mg/kg (all PAHs except dibenz[a, h]anthracene) and 0.01 mg/kg (dibenz[a,h]anthracene) (Department for Environment Food and Rural Affairs, 2002). Most river and estuary sediments in the Northeast of England
* Corresponding author. Tel.: þ44 (0)191 222 5099. E-mail address:
[email protected] (D. Werner). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.027
4530
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 2 9 e4 5 3 6
have PAH levels far above this (Woodhead et al., 1999), therefore treatment methods that allow sediment re-use are highly desirable. Bioremediation can potentially breakdown PAHs to harmless end products such as carbon dioxide. For bioremediation to be applied successfully, microorganisms capable of degrading native pollutants must be present under environmentally conducive conditions (Bamforth and Singleton, 2005). Improvements to the native microbiological conditions can be achieved by the addition of electron donors, acceptors and nutrients (biostimulation) (Ro¨ling et al., 2004) and additional microbial communities (bioaugmentation) (Abbondanzi et al., 2006). However, the accessibility of pollutants to microorganisms is related to the sediment’s physicochemical properties, especially the presence of strongly sorbing carbonaceous materials (Talley et al., 2002). Strong sorption by such materials causes pollutants to become inaccessible to microorganisms (Reid et al., 2000). Stabilisation of pollutants may reduce environmental risks without actually reducing the level of pollution. This technique is traditionally carried out via cement or lime based solidification of sediments and more recently, the addition of small amounts of activated carbon (AC) has been proposed to bind hydrophobic pollutants within contaminated sediment (Zimmerman et al., 2004). AC amendment has demonstrated dramatic reductions in aqueous PAH concentrations and the bioavailability of pollutants to diverse estuarine biota such as the clam Macoma balthica, the gastropod Hinia reticulate and the polychaete Nereis diversicolor (Cornelissen et al., 2006, McLeod et al., 2004). Potential obstacles to the successful application of this technique include AC fouling by dissolved organic matter in sediments (Pignatello et al., 2006) and a slow mass transfer of pollutants to AC under field conditions (Cho et al., 2007, Werner et al., 2006). Biodegradation and stabilisation may become competing processes as the addition of carbonaceous material lowers free aqueous concentrations and reduces the accessibility of pollutants to degrading microorganisms (Rhodes et al., 2008). If pollutants are biodegradable, the addition of sorbent materials may impede their breakdown; but if they are not readily biodegraded, sorbent amendment could be an effective alternative method to reduce environmental risk. The objective of this study was to compare natural attenuation and bioremediation in the form of biostimulation and bioaugmentation with AC amendment as potential treatments for PAH polluted sediment from the River Tyne. To this end, a number of complementary chemical, biological and modelling techniques were used as assessment tools; including the sediment PAH concentrations, the PAH availability and bacterial community analysis.
2.
Materials and methods
2.1.
Sediment sampling
Sediment was obtained from the Walker Riverside Country Park, Newcastle upon Tyne, UK, at a site where a tar works operated from 1920 to 1981. Seepage of hydrocarbons to the River Tyne occurs to the present day and deeper sediment is
severely contaminated. Sediment was sampled from the surface of the foreshore in the intertidal range, homogenised by stirring and stored at 4 C until use. The surface sediment contains lower levels of PAHs than deeper sediments, but concentrations are higher than those set by DEFRA for acceptable disposal of dredged sediment at sea (Department for Environment Food and Rural Affairs, 2002) and are similar to concentrations reported for other estuaries in the Northeast of England (Woodhead et al., 1999). PAH sources in the deeper sediment and the availability of atmospheric oxygen potentially provide ideal conditions for intrinsic aerobic PAH biodegradation in the intertidal surface sediment at this site.
2.2.
Biodegradation experiments: sediment microcosms
Three series of sacrificial sediment microcosms were used to determine changes in sediment PAH concentrations and the concentration of available PAHs after a period of biodegradation. The microcosms comprised: (1) an unamended system of 7 g (dry weight) sediment and 40 mL of River Tyne water and two amended systems with 7 g (dry weight) sediment and (2) 40 mL nutrient solution and (3) 40 mL nutrient solution inoculated with Pseudomonas putida DSM8368 (NCIMB 9816; (Yamamoto and Harayama, 1998)), a strain known to degrade naphthalene and phenanthrene (Barnsley, 1976, 1983), to obtain a final concentration of approximately 2 0.7$105 colony forming units/g wet sediment. Details of the source of the organism and their handling prior to use can be found in the supplementary data. The nutrient solution composition was based on the molar ratio of C:N:P ¼ 100:8:0.8, close to the recommended level of 100:10:1 (Leys et al., 2005), where C represents the organic carbon measured in the sediment. The nutrient solution consisted of 108.6 mg NH4Cl, 12.5 mg KH2PO4, 0.4 mg peptone, 0.2 mg CaCl2, 0.004 mg MnSO4, 0.004 mg ZnCl2, 0.004 mg CuSO4, 0.4 mg MgCl2 and 23.6 mg glucose in the 40 mL volume. Glucose was added as a cosubstrate at a level of 2.7% of the total organic carbon in order to stimulate microbial activity, as previously demonstrated (Keuth and Rehm, 1991) and because the addition of inorganic nutrients only had failed to increase cell numbers in preliminary work (see Table S1 in the supplementary data). Glucose is a simple carbon source, however P. putida are capable of preferentially utilising aromatic compounds in its presence (Basu et al., 2006). Biodegradation experiments were carried out in triplicate for durations of one week and one month at room temperature to allow for an eventual lag phase in the onset of biodegradation. Serum vials (50 mL) plugged with cotton wool were gently agitated at 35 rpm on a bench top shaker (Bibby Scientific Ltd, Staffordshire, UK), which allowed a 1 cm thick sediment layer to remain settled at the bottom of the microcosms. After the one week sampling event, a 0.15 0.01 g clean polyethylene (PE) passive sampler was added to the one month microcosms to measure PAH availability (Adams et al., 2007). Passive samplers are able to passively accumulate hydrophobic organic compounds from contaminated water until equilibrium is established. Compounds accumulated are considered available to be taken up by aquatic organisms (Reichenberg and Mayer, 2006). Details of the passive sampler
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 2 9 e4 5 3 6
can be found in the supplementary data. At both one week and one month sampling times, the microcosm content was stirred up and a 2 mL sample was taken and stored at 20 C prior to microbial analysis. Additionally, systems the same as microcosms (1) and (3) were also sampled at time zero. PE samplers in the one month microcosms were removed and extracted as outlined in the supplementary data. The sediment in all microcosms was collected on a grade 315 filter paper, retaining particles greater than 25 mm (VWR International Ltd, Leicestershire, UK), air dried overnight in the fume cupboard and solvent extracted for PAHs, together with the filter and the cotton wool plug using accelerated solvent extraction (ASE 200 Dionex, Surrey, UK) with 20 mL of 7% methanol in dichloromethane, three times. A blank consisting of the filter and a cotton wool ball but without sediment was run in parallel. PAHs concentration in the blank was near the analytical detection limit and less than 0.5% of those measured for the sediment samples.
of Similarity (ANOSIM) were performed using PRIMER-6 (Primer-E Ltd, UK). Bacterial diversity was measured as Shannon’s diversity (H0 ) and Pielou’s evenness (J ) indices and band richness (Magurran, 2004), which were calculated from DGGE band intensities as proxies for abundance and are shown in Table S2 in the supplementary data. For statistical analyses performed using MINITAB 15 Statistical Software (Minitab Inc), time-points for each treatment were grouped as replicates. Selected bands were excised from the DGGE gel (shown in Fig. S1 in the supplementary data), PCR amplified using primers 2/3, purified and sequenced with primer 2 or 3 (3.2 pmol/mes/mL) using the ABI prism Big Dye Terminator Cycle Sequencing Ready reaction Kit and an ABI Prism 377 DNA sequencer (Applied Biosystems, USA). Sequences were compared to the GenBank database using the BLAST algorithm (Benson et al., 2008) and are shown in Table S3 in supplementary data.
2.5. 2.3. Biodegradation experiments: aqueous phase microcosms At the end of the one month sediment microcosm experiment, filtrate (20 mL) from each of the sediment treatments was added to a 40 mL amber vial with a Teflon lined lid. The filtrate was spiked with 25 mL of phenanthrene from a 200 mg/mL standard in methanol (Sigma Aldrich, Dorset, UK) in order to test for the degradation of phenanthrene in the absence of sediment. Controls were prepared with deionized, autoclaved water and all vials were gently agitated and uncapped daily for 30 s in the fume cupboard to replenish oxygen. Experiments were carried out at room temperature (20 C) in amber glass vials. After one and seven days, duplicate samples were spiked with 5 mg of a d-10 phenanthrene surrogate standard and liquideliquid extracted three times with 10 mL hexane. The solvent extracts from this and the sediment microcosm studies were dried with sodium sulphate, cleaned-up and analysed as described in the supplementary data.
2.4. Microbial community composition and structure analyses The microbial community composition and structure was determined using the rapid community fingerprinting method, denaturing gradient gel electrophoresis (DGGE). Total bacterial DNA was extracted from 250 mL aliquots of the stored sample using the Fast DNA SPIN kit for soil as described in the manufacturer’s instructions (MP Bio-Medical, Cambridge, UK). Primers 2 and 3 were used to PCR amplify the V3 region of bacterial 16S rRNA gene fragments as previously described (Muyzer et al., 1993) and outlined in the supplementary data. The PCR products were analysed by DGGE as previously described (Ro¨ling et al., 2004) with minor modifications outlined in the supplementary data. DGGE gel banding patterns were used as a proxy of the microbial community structure using similarities in the presence and absence of bands (Dice coefficient) and were analysed using the image analysis software Bionumerics (Applied Maths NV, Belgium). Cluster analysis, non-metric multi-dimensional scaling and Analysis
4531
Activated carbon amendment
In all experiments a bitumen source AC (Chemviron Carbon Limited, Lancashire, UK) was used, which had a measured surface area of 1012 m2/g, a pore volume of 0.6 cm3/g and the following pore distribution: ultramicropores 44.6%, micro/ supermicropores 23% and meso/macropores 32.4%. The properties were measured by adsorption of nitrogen and carbon dioxide using an Intelligent Gravimetric Analyser (IGA) model 003 (Hiden Isochema Ltd., Warrington, U.K). To investigate PAH availability in sediment microcosms after AC amendment, triplicate batches containing River Tyne water (40 mL), sediment (7.9 g dry weight) and 2 wt % AC (particle size 63e125 mm) were mixed for 1 month as described for the sediment microcosm studies. After one week, a 0.15 0.01 g clean PE sampler was added to the batches to allow a direct comparison with the bioremediation microcosms, and after a further three weeks contact the PE samplers were removed and analysed as detailed in the supplementary data.
2.6.
Modelling
A previously described model (Ahn et al., 2008, Werner et al., 2006) was modified according to details in the supplementary data and used for the interpretation and discussion of PAH availability after biodegradation and AC amendment. The model simulates the mass transfer of hydrophobic pollutants from sediment to added sorbents such as PE samplers or AC particles. The model was calibrated with previously reported experimental data (Hale and Werner, 2010) and empirical relationships based on compound octanolewater partitioning coefficients were used to characterise the PE samplers (Booij et al., 2003).
3.
Results and discussion
3.1.
Sediment characterisation
The bulk sediment contained 16.4 7.3 mg/kg of the S16 USEPA PAHs. This concentration is moderate in comparison to those found in PAH pollution hotspots (Ghosh et al., 2003), but
4532
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 2 9 e4 5 3 6
individual PAH compound concentrations are up to 23 times higher than those set by DEFRA for acceptable disposal of dredged sediment at sea (Department for Environment Food and Rural Affairs, 2002). The total organic carbon content of the sediment was 5.06% and a petrographic analysis indicated that coal particles contributed approximately 88% to the total carbon. A Tenax bead desorption revealed that after 112 days just 8.7% of the total PAH mass desorbed from the sediment (Hale and Werner, 2010) and this low PAH availability could be explained by the abundance of carbonaceous particles. Readily available pollutant fractions (both considering chemical activity and accessibility (Reichenberg and Mayer, 2006)), determined by measuring the fraction desorbed within 6 h or pollutant uptake by passive sampling devices or SPME fibres correlate better with ecotoxicity than total PAH concentrations (Hawthorne et al., 2007, Kraaij et al., 2003, Landrum et al., 2007, Vinturella et al., 2004). Based on such findings, here we investigated the potential of inexpensive treatments in the form of bioremediation and sorbent amendment to further reduce the amount of available PAHs in River Tyne sediment and quantify this by employing PE passive samplers.
3.2.
Biodegradation experiments: sediment microcosms
For all bioremediation microcosms (unamended, biostimulated and bioaugmented), the total sediment PAH concentrations were not significantly different (t-test, p < 0.01), after one week and one month of treatment compared to concentrations at the start of the biodegradation experiments, as shown in Fig. 1. These results are not entirely unexpected since the desorption data indicates very limited availability of these PAHs. Even in cases where an active PAH degrading community is present, persistence of aged PAHs due to lack of desorption has been seen (Carmichael et al., 1997). Talley et al. (2002) observed limited PAH desorption from the coal based sediment fraction and concurrently no reduction in coal-associated PAH concentrations after a bioslurry treatment.
3.3. Biodegradation experiments: uptake by passive samplers PE passive samplers were employed to compare the availability of PAHs after one month in unamended, biostimulated and bioaugmented sediment microcosms. Compounds taken up by these devices are considered available as passive samplers measure free aqueous concentrations (Huckins et al., 1993) and it is these compounds that are potentially susceptible to biodegradation. A comparison of the compound pattern in Figs. 1 and 2 demonstrates that the lower molecular weight compounds which were more readily desorbed from the sediment (Hale and Werner, 2010), but initially present at lower concentrations than some of the higher molecular weight PAHs were transferred from the sediment to the samplers. Fig. 2 also shows that a much greater PAH uptake (t-test p < 0.05) was observed for the biostimulated and bioaugmented microcosms compared to the unamended, intrinsically bioremediated microcosm. This finding is contrary to the intended effect of the treatments and may be related to the production of surfactants, which can increase
Fig. 1 e Total bulk sediment PAH concentration (mg/kg) in all sediment microcosm treatments after (1) 1 week and (2) 1 month. Error bars represent one standard deviation from the mean of three replicates. Statistically significant differences (t-test p < 0.05) only occur for NAP and ACEN between the unamended and nutrient amended batches after 1 week treatment.
the availability of pollutants. Qualitatively, the presence of surfactants was observed in the liquideliquid extract of filtrate from the nutrient amended and especially the nutrient and P. putida amended treatments, which formed emulsions in the aqueous phase microcosms described in the next paragraph. Members of the genus Pseudomonas have been reported to produce biosurfactants (Kuiper et al., 2004, Tran et al., 2008, Tuleva et al., 2002) which can increase the mass transfer of compounds to the aqueous phase and therefore facilitate uptake by the passive sampler.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 2 9 e4 5 3 6
Fig. 2 e Uptake of PAHs by PE passive samplers added to sediment microcosms after one week of bioremediation or AC treatment and analysed after a further three weeks contact. Total uptake of PAHs was 4.70 ± 0.25 mg/g for the unamended batches, 12.43 ± 1.78 mg/g for the nutrient amended batches, 23.49 ± 2.73 mg/g for the nutrient and Pseudomonas putida amended batches and 0.28 ± 0.01 mg/g for the AC amended batches. Also included in the figure are model predictions for PAH uptake in the absence of biodegradation or AC amendment.
3.4. Biodegradation experiments: aqueous phase microcosms As shown in Fig. 3, phenanthrene biodegradation was observed in the sediment-free filtrate from the unamended microcosms and to a lesser extent in the filtrate from the biostimulated and bioaugmented microcosms. It appears that phenanthrenedegraders are present in the native microbial population of the sediment and are capable of degrading available, freely dissolved phenanthrene. Phenanthrene degradation was slower in the sediment-free filtrate from the biostimulated and bioaugmented treatments. This result agrees with the higher phenanthrene uptake by PE passive samplers in the biostimulated and bioaugmented microcosms compared to the unamended microcosm. The apparent absence or reduction of intrinsic phenanthrene degradation in the biostimulated and bioaugmented microcosm may have resulted from microorganisms other than phenanthrene-degraders adapting more successfully to the changed nutrient conditions.
3.5. Microbial community composition and structure analysis The bacterial community structures were significantly different (ANOSIM, p < 0.1) for the various treatments. Unamended microcosms showed little change in the community structure of the predominant bacterial members over the time of the biodegradation study, illustrated in Fig. 4. These samples exhibited the greatest diversity and evenness
4533
Fig. 3 e Phenanthrene mass in sediment-free filtrate from the unamended, biostimulated (nutrient amended) and bioaugmented (nutrient and P. putida amended) microcosms determined after one and seven days.
values (Table S2 in the supplementary data). By contrast, nutrient amendment (biostimulated) was followed by a change in the community structure of the predominant bacterial members after one week compared to the original community structure in the unamended microcosm at time zero, similar to previous reports (Miyasaka et al., 2006, Ro¨ling et al., 2004). However, the structure returned to being similar to the unamended microcosms after one month (Fig. 4), which has been reported previously for uncontaminated sediments (Miyasaka et al., 2006). This suggests that nutrient addition stimulated a previously minor proportion of the community, although this community was apparently not involved in PAH degradation. The coincidence of increased PAH uptake by passive samplers in the biostimulated microcosms compared to the unamended microcosms (Fig. 2) and slower degradation of spiked phenanthrene in the filtrate from the biostimulated microcosms compared to the unamended microcosms (Fig. 3), suggests the change in the community structure did not prove advantageous to PAH degraders. The bacterial community structure at time zero in the nutrient and P. putida amended microcosm (bioaugmented) was distinctly different from the other treatments, reflecting the addition of an abundant P. putida population. Together with the one week data for this treatment, the communities were significantly less diverse, rich and even than either nutrient amended or unamended microcosms ( p < 0.05), shown in Table S2 in the supplementary data. This is also reflected by the dominance of three strong bands in the DGGE gel shown in Fig. 4, two of which were identified as P. putida strain DSM8368 (NCIMB 9816; GenBank accession number D86000, (Yamamoto and Harayama, 1998)) and P. putida strain ATCC 1757/AJ (amongst others, GenBank accession number AY391278, (Danko et al., 2004)) as shown in Table S3 in the supplementary data. Members of the genus Pseudomonas possess on average 4.3 different rRNA operons (Lloyd-Jones et al., 2005) and have
4534
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 2 9 e4 5 3 6
Fig. 4 e A dendrogram showing cluster analysis of the similarities (Dice coefficient) between the different community composition profiles of sediment microcosm treatments over time examined by DGGE. The image of the DGGE has been normalised to markers run on the gel following analysis with Bionumerics.
up to 4 different ribotypes (Acinas et al., 2004), which explains the multiple bands detected by DGGE. After one month, none of the Pseudomonas-specific bands were detected and the resulting bacterial community fingerprint showed a very different structure to other treatments. These observations indicate that the spiked P. putida population did not degrade PAHs, but declined in abundance and was replaced with different taxa that may have grown on substrates other than PAHs such as the soluble breakdown products of the P. putida cells, biosurfactants or other dissolved organic carbon (DOC) compounds in the microcosm. Indeed, the comparative low level of free aqueous PAHs (0.001581 mg/L) compared to DOC (10.80 mg/L) in the sediment pore water (where the DOC determination method is provided in the supplementary data) may explain why biostimulation and bioaugmentation failed to accelerate PAH degradation in this sediment.
3.6.
Activated carbon amendment
Previously determined logarithmic ACewater partitioning coefficients for PAHs ranged from 5.5 to 9.2 (Hale and Werner, 2010) and are approximately three orders of magnitude higher than the sedimentewater partitioning coefficients for PAHs in River Tyne sediment, as shown in Table S4 in the supplementary data. AC is a very strong sorbent of PAHs and it was previously observed that the effect of DOC on the adsorption of PAHs to AC was minimal for this sediment (Hale and Werner, 2010). Under conditions identical to those of the sediment biodegradation microcosms, a 2% dose of AC by dry sediment weight was very effective in further reducing the availability of the already strongly sequestered PAHs. Total PAH uptake by PE passive samplers was reduced by 94% compared to the unamended sediment microcosm, coming close to the analytical limit of detection for most compounds and is shown in Fig. 2.
3.7. Modelling the uptake of PAHs to PE passive samplers Sediment data from Table S4 in the supplementary data and empirical relationships for the PE characteristics (Booij et al.,
2003) were used to calibrate a numerical model (Werner et al., 2006) which was employed to assess the uptake of PAHs to PE passive samplers embedded in sediment microcosms. An example of the model calibration data needed is given for phenanthrene in Table S5 in the supplementary data. Model predictions shown in Fig. 2 indicate the level of PAH uptake anticipated in the absence of biodegradation and AC amendment and is therefore based solely on the geochemical characterisation of the sediment. Even when allowing for a typical factor of two uncertainty in such model predictions (Werner et al., 2006), a comparison of modelled and measured data supports the idea of biodegradation of a portion of the available phenanthrene and other two to four ring PAHs in the unamended sediment microcosm for which uptake was less than predicted by the model. This intrinsic biodegradation potential appears to be suppressed for the nutrient amended microcosm, possibly because nutrient amendment stimulated growth of microorganisms not involved in PAH degradation and this growth may have held down intrinsic PAH degradation. In the case of nutrient and P. putida amendment, the availability is greater than the model prediction for most compounds, likely due to the presence of surfactants in these microcosms. Hence, in this study, an apparent increase in the PAH availability was not accompanied by an increase in PAH degradation and in an environmental application this may exacerbate, instead of mitigate the PAH pollution problem, corroborating what has previously been suggested (Laha and Luthy, 1992).
4.
Conclusions
This work illustrates the remediation challenges caused by the low availability and low free aqueous concentrations of PAHs in River Tyne sediment. Such conditions appear to hinder effective bioremediation however it is possible that different experimental conditions other than those investigated in this sediment microcosm study could have resulted in successful biodegradation. Yet, low cost bioremediation technologies are difficult to engineer and environmental manipulations, in this instance, the addition of nutrients or nutrients and P. putida,
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 2 9 e4 5 3 6
do not always have the desired effect. PE passive samplers used in sediment biodegradation studies allowed the fate of a small available portion of the PAHs to be monitored. This elucidated differences between treatments which would otherwise have been masked by the uncertainty of quantifying a small difference in the total sediment PAH concentration. Uptake of PAHs by passive samplers demonstrated that not all of the PAHs released from sediment were subsequently biodegraded, even for the unamended microcosm where intrinsic biodegradation of phenanthrene was evident. On the other hand, the addition of AC to River Tyne sediment could be an effective means of further stabilising the already strongly sequestered pollutant residue. This treatment would enhance the sediment quality and potential for re-use as a resource, for instance for shore defence or as a geotechnical infill material, instead of handling it as waste material.
Acknowledgements Funding was provided by the UK Engineering and Physical Science Research Council (EPSRC) grant EP/D079055/1. Petrographic analysis of the sediment was kindly provided by Dr. Stavros Kalaitzidis from the Department of Geology at the University of Patras, Greece. Collaboration with Dr Kalaitzidis was enabled by EPSRC grant EP/F012934/1.
Appendix. Supplementary data Supplementary data associated with article can be found in online version at 10.1016/j.watres.2010.06.027.
references
Abbondanzi, F., Bruzzi, L., Campisi, T., Frezzati, A., Guerra, R., Iacondini, A., 2006. Biotreatability of polycyclic aromatic hydrocarbons in brackish sediments: preliminary studies of an integrated monitoring. International Biodeterioration & Biodegradation 57 (4), 214e221. Acinas, S.G., Marcelino, L.A., Klepac-Ceraj, V., Polz, M.F., 2004. Divergence and redundancy of 16S rRNA sequences in genomes with multiple rrn operons. Journal of Bacteriology 186 (9), 2629e2635. Adams, R.G., Lohmann, R., Fernandez, L.A., Macfarlane, J.K., Gschwend, P.M., 2007. Polyethylene devices: passive samplers for measuring dissolved hydrophobic organic compounds in aquatic environments. Environmental Science & Technology 41 (4), 1317e1323. Ahn, S., Werner, D., Luthy, R.G., 2008. Modeling PAH mass transfer in a slurry of contaminated soil or sediment amended with organic sorbents. Water Research 42 (12), 2931e2942. Bamforth, S.M., Singleton, I., 2005. Bioremediation of polycyclic aromatic hydrocarbons: current knowledge and future directions. Journal of Chemical Technology and Biotechnology 80 (7), 723e736. Barnsley, E.A., 1976. Role and regulation of ortho-pathway and meta pathway of catechol metabolism in pseudomonas metbolizing naphthalene and salicylate. Journal of Bacteriology 125 (2), 404e408.
4535
Barnsley, E.A., 1983. Bacterial oxidation of naphthalene and phenanthrene. Journal of Bacteriology 153 (2), 1069e1071. Basu, A., Apte, S.K., Phale, P.S., 2006. Preferential utilization of aromatic compounds over glucose by pseudomonas putida CSV86. Applied and Environment Microbiology 72 (3), 2226e2230. Benson, D.A., Karsch-Mizrachi, I., Lipman, D.J., Ostell, J., Wheeler, D.L., 2008. GenBank. Nucleic Acids Research 36, D25eD30. Booij, K., Hofmans, H.E., Fischer, C.V., Van Weerlee, E.M., 2003. Temperature-dependent uptake rates of nonpolar organic compounds by semipermeable membrane devices and lowdensity polyethylene membranes. Environmental Science & Technology 37 (2), 361e366. Carmichael, L.M., Christman, R.F., Pfaender, F.K., 1997. Desorption and mineralization kinetics of phenanthrene and chrysene in contaminated soils. Environmental Science & Technology 31 (1), 126e132. Cho, Y.-M., Smithenry, D.W., Ghosh, U., Kennedy, A.J., Millward, R.N., Bridges, T.S., Luthy, R.G., 2007. Field methods for amending marine sediment with activated carbon and assessing treatment effectiveness. Marine Environmental Research 64 (5), 541e555. Cornelissen, G., Breedveld, G.D., Naes, K., Oen, A.M.P., Ruus, A., 2006. Bioaccumulation of native polycyclic aromatic hydrocarbons from sediment by a polychaete and a gastropod: freely dissolved concentrations and activated carbon amendment. Environmental Toxicology and Chemistry 25 (9), 2349e2355. Danko, A.S., Luo, M.Z., Bagwell, C.E., Brigmon, R.L., Freedman, D.L., 2004. Involvement of linear plasmids in aerobic biodegradation of vinyl chloride. Applied and Environmental Microbiology 70 (10), 6092e6097. Department for Environment Food and Rural Affairs, 2002. The Use of Action Levels in the Assessment of Dredged Material Placement at Sea and in Estuarine Areas under FEPA(II). European Parliament and Council of the European Union, 2000. Water Framework Directive 2000/60/EC. Official Journal of the European Communities L (327), 1. Ghosh, U., Zimmerman, J.R., Luthy, R.G., 2003. PCB and PAH speciation among particle types in contaminated harbor sediments and effects on PAH bioavailability. Environmental Science & Technology 37 (10), 2209e2217. Hale, S.E., Werner, D., 2010. Modeling the mass transfer of hydrophobic organic pollutants in briefly and continuously mixed sediment after amendment with activated carbon. Environmental Science and Technology 44 (9), 3381e3387. Hawthorne, S.B., Azzolina, N.A., Neuhauser, E.F., Kreitinger, J.P., 2007. Predicting bioavailability of sediment polycyclic aromatic hydrocarbons to Hyalella azteca using equilibrium partitioning, supercritical fluid extraction, and pore water concentrations. Environmental Science & Technology 41 (17), 6297e6304. Huckins, J.N., Manuweera, G.K., Petty, J.D., Mackay, D., Lebo, J.A., 1993. Lipid-containing semipermeable-membrane devices for monitoring organic contaminants in water. Environmental Science & Technology 27 (12), 2489e2496. Keuth, S., Rehm, H.J., 1991. Biodegradation of phenanthrene by arthrobacter-polychromogenes isolated from a contaminated soil. Applied Microbiology and Biotechnology 34 (6), 804e808. Kirby, M.F., Matthiessen, P., Neall, P., Tylor, T., Allchin, C.R., Kelly, C.A., Maxwell, D.L., Thain, J.E., 1999. Hepatic EROD activity in flounder (Platichthys flesus) as an indicator of contaminant exposure in English estuaries. Marine Pollution Bulletin 38 (8), 676e686. Kraaij, R., Mayer, P., Busser, F.J.M., Bolscher, M.V., Seinen, W., Tolls, J., 2003. Measured pore-water concentrations make equilibrium partitioning work e a data analysis. Environmental Science & Technology 37 (2), 268e274.
4536
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 2 9 e4 5 3 6
Kuiper, I., Lagendijk, E.L., Pickford, R., Derrick, J.P., Lamers, G.E.M., Thomas-Oates, J.E., Lugtenberg, B.J.J., Bloemberg, G.V., 2004. Characterization of two Pseudomonas putida lipopeptide biosurfactants, putisolvin I and II, which inhibit biofilm formation and break down existing biofilms. Molecular Microbiology 51 (1), 97e113. Laha, S., Luthy, R.G., 1992. Effects of nonionic surfactants on the solubilization and mineralization of phenanthrene in soilewater systems. Biotechnology and Bioengineering 40 (11), 1367e1380. Landrum, P.F., Robinson, S.D., Gossiaux, D.C., You, J., Lydy, M.J., Mitra, S., Ten Hulscher, T.E.M., 2007. Predicting bioavailability of sediment-associated organic contaminants for Diporeia spp. and Oligochaetes. Environmental Science & Technology 41 (18), 6442e6447. Leys, N.M., Bastiaens, L., Verstraete, W., Springael, D., 2005. Influence of the carbon/nitrogen/phosphorus ratio on polycyclic aromatic hydrocarbon degradation by Mycobacterium and Sphingomonas in soil. Applied Microbiology and Biotechnology 66 (6), 726e736. Lloyd-Jones, G., Laurie, A.D., Tizzard, A.C., 2005. Quantification of the Pseudomonas population in New Zealand soils by fluorogenic PCR assay and culturing techniques. Journal of Microbiological Methods 60 (2), 217e224. Magurran, A.E., 2004. Measuring Biological Diversity. Blackwell Publishing Ltd, Oxford, UK. McLeod, P.B., Van Den Heuvel-Greve, M.J., Allen-King, R.M., Luoma, S.N., Luthy, R.G., 2004. Effects of particulate carbonaceous matter on the bioavailability of benzo a pyrene and 2,20 ,5,50 -tetrachlorobiphenyl to the clam, Macoma balthica. Environmental Science & Technology 38 (17), 4549e4556. Miyasaka, T., Asami, H., Watanabe, K., 2006. Impacts of bioremediation schemes on bacterial population in naphthalene-contaminated marine sediments. Biodegradation 17 (3), 227e235. Muyzer, G., Dewaal, E.C., Uitterlinden, A.G., 1993. Profiling of complex microbial-populations by denaturing gradient gelelectrophoresis analysis of polymerase chain reactionamplified genes-coding for 16s Ribosomal-Rna. Applied and Environmental Microbiology 59 (3), 695e700. Pignatello, J.J., Kwon, S., Lu, Y.F., 2006. Effect of natural organic substances on the surface and adsorptive properties of environmental black carbon (char): attenuation of surface activity by humic and fulvic acids. Environmental Science & Technology 40 (24), 7757e7763. Reichenberg, F., Mayer, P., 2006. Two complementary sides of bioavailability: accessibility and chemical activity of organic contaminants in sediments and soils. Environmental Toxicology and Chemistry 25 (5), 1239e1245.
Reid, B.J., Jones, K.C., Semple, K.T., 2000. Bioavailability of persistent organic pollutants in soils and sediments e a perspective on mechanisms, consequences and assessment. Environmental Pollution 108 (1), 103e112. Rhodes, A.H., Carlin, A., Semple, K.T., 2008. Impact of black carbon in the extraction and mineralization of phenanthrene in soil. Environmental Science & Technology 42 (3), 740e745. Ro¨ling, W.F.M., Milner, M.G., Jones, D.M., Fratepietro, F., Swannell, R.P.J., Daniel, F., Head, I.M., 2004. Bacterial community dynamics and hydrocarbon degradation during a field-scale evaluation of bioremediation on a mudflat beach contaminated with buried oil. Applied and Environmental Microbiology 70 (5), 2603e2613. Stentiford, G.D., Longshaw, M., Lyons, B.P., Jones, G., Green, M., Feist, S.W., 2003. Histopathological biomarkers in estuarine fish species for the assessment of biological effects of contaminants. Marine Environmental Research 55 (2), 137e159. Talley, J.W., Ghosh, U., Tucker, S.G., Furey, J.S., Luthy, R.G., 2002. Particle-scale understanding of the bioavailability of PAHs in sediment. Environmental Science & Technology 36 (3), 477e483. Tran, H., Kruijt, M., Raaijmakers, J.M., 2008. Diversity and activity of biosurfactant-producing Pseudomonas in the rhizosphere of black pepper in Vietnam. Journal of Applied Microbiology 104 (3), 839e851. Tuleva, B.K., Ivanov, G.R., Christova, N.E., 2002. Biosurfactant production by a new Pseudomonas putida strain. Zeitschrift Fur Naturforschung C-a Journal of Biosciences 57 (3e4), 356e360. Vinturella, A.E., Burgess, R.M., Coull, B.A., Thompson, K.M., Shine, J.P., 2004. Use of passive samplers to mimic uptake of polycyclic aromatic hydrocarbons by benthic polychaetes. Environmental Science & Technology 38 (4), 1154e1160. Werner, D., Ghosh, U., Luthy, R.G., 2006. Modeling polychlorinated biphenyl mass transfer after amendment of contaminated sediment with activated carbon. Environmental Science & Technology 40 (13), 4211e4218. Woodhead, R.J., Law, R.J., Matthiessen, P., 1999. Polycyclic aromatic hydrocarbons in surface sediments around England and Wales, and their possible biological significance. Marine Pollution Bulletin 38 (9), 773e790. Yamamoto, S., Harayama, S., 1998. Phylogenetic relationships of Pseudomonas putida strains deduced from the nucleotide sequences of gyrB, ropD and 16S rRNA genes. International Journal of Systematic Bacteriology 48, 813e819. Zimmerman, J.R., Ghosh, U., Millward, R.N., Bridges, T.S., Luthy, R.G., 2004. Addition of carbon sorbents to reduce PCB and PAH bioavailability in marine sediments: physicochemical tests. Environmental Science & Technology 38 (20), 5458e5464.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 7 0 e4 5 8 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Characterization of elemental and structural composition of corrosion scales and deposits formed in drinking water distribution systems Ching-Yu Peng a,*, Gregory V. Korshin a, Richard L. Valentine b, Andrew S. Hill c, Melinda J. Friedman c, Steve H. Reiber d a
Department of Civil and Environmental Engineering, University of Washington, Box 352700, Seattle, WA 98105-2700, USA Department of Civil and Environmental Engineering, University of Iowa, Iowa City, IA 52242-1527, USA c Confluence Engineering, 517 NE 92nd Street, Seattle, WA, USA d HDR Inc. 500 108th Ave NE Suite 1200, Bellevue, WA 98004-5549, USA b
article info
abstract
Article history:
Corrosion scales and deposits formed within drinking water distribution systems (DWDSs)
Received 9 January 2010
have the potential to retain inorganic contaminants. The objective of this study was to
Received in revised form
characterize the elemental and structural composition of extracted pipe solids and
23 May 2010
hydraulically-mobile deposits originating from representative DWDSs. Goethite (a-FeOOH),
Accepted 26 May 2010
magnetite (Fe3O4) and siderite (FeCO3) were the primary crystalline phases identified in
Available online 10 June 2010
most of the selected samples. Among the major constituent elements of the deposits, iron was most prevalent followed, in the order of decreasing prevalence, by sulfur, organic
Keywords:
carbon, calcium, inorganic carbon, phosphorus, manganese, magnesium, aluminum and
Corrosion scales
zinc. The cumulative occurrence profiles of iron, sulfur, calcium and phosphorus for pipe
Composition
specimens and flushed solids were similar. Comparison of relative occurrences of these
Structure
elements indicates that hydraulic disturbances may have relatively less impact on the
Drinking water distribution systems
release of manganese, aluminum and zinc, but more impact on the release of organic carbon, inorganic carbon, and magnesium. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Corrosion scales affect water quality in drinking water distribution systems (DWDSs) in many important ways. The rates at which such scales are generated and chemical composition, structures, morphologies and solubilities of predominant mineralogical phases constituting them are all affected by both the pipe material (cast iron, steel or PVC) on which they are deposited and water chemistry parameters that include pH, temperature, DIC and alkalinity, concentrations of sulfate, chloride and natural organic matter (NOM), dissolved oxygen (DO) and disinfectant type and residual,
presence of corrosion inhibitors (e.g., phosphate), overall conductivity of water, and hydraulic patterns (Sarin et al., 2001; Korshin et al., 1996; Vazquez et al., 2006). Over past several years, the issue of potentially significant accumulation of trace inorganic contaminants (e.g., arsenic, vanadium, lead and others) within DWDSs has gained considerable attention. Reiber and Dostal, 2000; Lytle et al., 2004; Schock et al., 2008 and Gerke et al., 2009 have demonstrated that when these inorganic contaminants are present at concentrations below their respective maximum contaminant levels (MCLs) or even at essentially non-detect levels in water sources, they are capable of accumulating to
* Corresponding author. E-mail address:
[email protected] (C.-Y. Peng). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.05.043
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 7 0 e4 5 8 0
measurable levels on and within deposits that exist in DWDSs. Certain compounds commonly found in corrosion scales and other deposits, including individual phases of iron(III) and manganese oxides, have been shown to adsorb and concentrate trace inorganic contaminants (Sugiyama et al., 1992; Nelson et al., 1995; Fendorf et al., 1997; Larsen and Postma, 1997; Gray et al., 1999; O’Reilly et al., 2001; Trived et al., 2001; Lytle et al., 2004; Cance`s et al., 2005). While the amount of information concerning the properties of corrosion scales, deposits and colloidal particles in DWDSs is remarkable (Tuovinen et al., 1980; Benjamin et al., 1996; Sarin et al., 2001; Teng et al., 2008; Gerke et al., 2008; Borch et al., 2008; Barkatt et al., 2009), there is a need to provide a more complete characterization of their physicochemical properties. This information can provide valuable insight on factors that influence and control the accumulation and co-occurrence of regulated trace inorganic contaminants. Accordingly, the main objective of this study was to characterize corrosion scales and deposits originating from DWDSs with varying finished water chemistries and pipe materials.
2.
Materials and methods
2.1.
Participating utilities
Twenty drinking water utilities that participated in the study were located in the contiguous United States. The selection of participated utilities and detailed information on sampling and analytical approaches can be found in the report by Friedman et al. (2010) on the relevant study funded by Water Research Foundation. Most of the utilities were from the upper Midwest while others were from the Western, Southwestern, and Northeastern regions, where groundwater tends to be softer and less mineralized than in the Midwest. Table 1 provides a summary of the utility participants and certain characteristics of each, as reported by each utility.
2.2.
Sample types and processing
Three types of samples were collected: (1) pipe specimens, either obtained from a recent “live” extraction or from a utility storage area (referred to as “boneyard” specimens); (2) hydraulically-mobile deposit material collected during hydrant flushing events; and (3) distribution system water samples. Where appropriate, distribution system water samples were collected to correspond to each solid sample (i. e., at a site near the location where the solid sample was obtained). General water quality parameters (pH, temperature, alkalinity, disinfectant residual, and turbidity) were measured at the same time and location as water sample collection. Samples provided by utility participants were sent to the Environmental Engineering and Science Laboratory (EES) of the University of Iowa for processing, distribution and analyses for radionuclides (to be reported elsewhere). Water samples were shipped to the Department of Civil and Environmental Engineering (CEE) of the University of Washington. Upon receipt, the samples were filtered through a 33-mm Millex-HA syringe filter (Millipore Corporation, Bedford MA)
4571
with a 0.45-mm nominal pore size to remove particulate matter. The filtrate was acidified to reach a 1% nitric acid concentration and placed in 15 mL conical polypropylene test tubes. The samples were spiked with internal standards (45Sc, 74 Ge, and 103Rh) and stored at 4 C until analyzed.
2.3.
Solid samples collection
72 Solid samples were collected from 20 drinking water utilities. Of those, 26 were hydrant flush solids and 46 were pipe specimens (including 34 live pipe extractions and 12 boneyard samples). The hydrant flush samples were obtained during conventional flushing. They provided an opportunity to assess the composition of hydraulically-mobile solids released due to hydraulic disturbances. In contrast with hydrant flush solids, scale from pipe specimens can be operationally considered as hydraulically-inert material. Removal and characterization of scale allow for an assessment of total accumulation of inorganic compounds and contaminants, particularly in cases of adhering scales that are not susceptible to removal by flushing. Solid samples examined in this study are summarized in Table 1. To obtain pipe specimens, deposit material was carefully removed from the exposed pipe surface. To obtain solids mobilized during hydrant flush events, a net assembly consisting of twin hydrant nets and/or pantyhose was used to retain the particulates. All collected solid material was dried at 103 C for 24 h and weighed to determine its dry mass. In the case of pipe specimens, a portion of mixed dried sample was crushed using a mortar and pestle, passed through a number 50 sieve (300-mm mesh) and homogenized. The crushed/sieved material was digested (as described below) and analyzed to determine its elemental composition. For selected samples, subsets of both crushed and uncrushed material were used examined using X-Ray Diffraction (XRD) and Scanning Electron Microscopy (SEM) to determine their mineralogy and morphology. Determination of the elemental composition was possible for 35 of 46 pipe specimens and 23 of 26 of hydrant flush solids. The determination of elemental sulfur and carbon content was carried out for the subsets of 48 and 36 samples, respectively.
2.4.
Analytical procedures
All digestions of solid samples were performed at the EES using USEPA Method 3050B (Acid Digestion of Sediments, Sludges and Soils) (U.S.EPA method 3050B). For 58 samples that had adequate mass for processing, the fraction of sample mass that was digested by the above procedure ranged from 24 to 96%, with an average standard deviation of 78% 15% (Supplementary Information Table S1). Aliquots of the digests were sent to the CEE to determine their elemental composition. Ten of the solid samples were chosen for morphological and surface elemental composition using SEM and energy dispersive spectroscopy (EDS) technique. SEM/EDS measurements were performed with a JEOL-7000F high-resolution SEM instrument (JEOL Corporation, Japan). EDS data were acquired in two modes. The first mode allowed examining the entire surface of the sample, while the second mode
4572
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 7 0 e4 5 8 0
Table 1 e Summary of utility participants and solid samples examined in this study. Utility identifier
Region
Service Water population sourcesa
W
Midwest
7000
GW
CL
Midwest
28,000
GW
SA
West
60,000
GW, SW
CH RW IN
West West West
11,000 6300 57,000
GW GW GW, SW
CC
West
1900
GW
DN CA PC WDB WA B
West West West Northeast Northeast West
1,200,000 100,000 8000 1200 6000 493,000
SW GW, SW GW, SW GW GW SW, GW
G
Midwest
5000
GW
AZ BC J NC
West West Midwest Midwest
245 28,000 145,000 200
GW GW GW GW
ST
Midwest
15,000
GW
K
Midwest
8000
GW
Primary treatment and posttreatment applications Electrodialysis reversal, chloramines, poly-PO4 Fluoridation, PO4 blend, free chlorine Free chlorine, Fe/Mn removal, pH adjustment, fluoride Free chlorine Free chlorine Free chlorine, Fe/Mn/As removal, fluoride, ortho-PO4 Free chlorine, Fe/Mn removal, pH adjustment, fluoride, ortho-PO4 Conventional treatment, chloramines Free chlorine, Fe/Mn removal Free chlorine, Fe/Mn/As removal Free chlorine Free chlorine, pH adjustment Conventional treatment, chloramines Free chlorine, cation exchange, pH adjustment, fluoride Free chlorine Free chlorine PO4 Blend, chloramines Cation exchange, poly-PO4, fluoride, free chlorine Cation exchange, PO4 blend, pH adj., fluoride, free chlorine Chloramines, HMO Filter Process, pH adjustment
Pipe materialb
Solid samples provided by the utility 3 Pipe specimen, 5 hydrant flush 5 pipe specimen, 2 hydrant flush 3 Pipe specimen, 2 hydrant flush 1 Pipe specimen 2 pipe specimen 4 Pipe specimen 6 Pipe specimen 2 2 2 1 4 4
Pipe specimen Pipe specimen pipe specimen Pipe specimen Pipe specimen Pipe specimen
2 Hydrant flush 1 Pipe specimen 2 Pipe specimen 10 Hydrant flush 1 Hydrant flush 2 Pipe specimen, 2 hydrant flush 2 Pipe specimen, 2 hydrant flush
Cast iron Cast iron 1 PVC, 4 Cement-Lined Iron steel Galvanized Iron 1 Cement-Lined Iron, 3 Ductile Iron 4 Cast iron, 2 Cement-Lined Iron Cast iron 1 Steel, 1 Cast iron Galvanized Iron HDPE Cast iron 2 Ductile Iron, 1 cast iron, 1 Cement-Lined Iron 1 Cast iron, 1 Cement-Lined Iron PVC PVC Cast iron PVC 2 Cast iron, 2 Cement-Lined Iron 1 Ductile Iron, 3 cast iron
a GW ¼ Groundwater; SW ¼ Surface water. b For hydrant flush samples, pipe material refers to the type of pipe used to distribute water in the flushing zone.
corresponded to localized spots that were selected primarily on the basis of apparent morphological differences. Prior to SEM/EDS analysis, the requisite amount of the solid was placed on a 9.5-mm aluminum specimen mount (Ted Pella Inc., Redding CA) using double-coated 9-mm conductive carbon pads (Ted Pella Inc., Redding CA). The sample was vacuum sputter-coated to deposit a thin surface conductive layer with an SPI sputter coater (Structure Probe Inc., West Chester PA). XRD measurements were performed for the aforementioned ten selected samples to identify predominant mineralogical phases. Analysis of the first group of solid samples (consisting of CC-A, CC-D, CH-A, J-B, and J-E) were carried out using a Philips PW1830 X-ray diffractometer (Philips, Netherlands). Analysis of the second group of solid samples (consisting of RW-A, RW-B, PC-A, PC-B and J-J) were carried out using a Siemens D5000 X-ray diffractometer (Siemens Corporation, New York NY). Ni-filtered Cu-Ka radiation (l of ˚ ) was used to perform crystallographic analysis in 1.5406 A both cases. The range of 2q values was 10 e80 with a 0.05 step size. The scanning speed was 2 per second. XRD patterns were identified using Jadeþ software (version 6). Diffraction data were compared against reference patterns from the 1995 version of International Center for Diffraction Data (ICDD).
2.5.
Analytical parameters and methods
Inorganic elements of treated water and digested solid samples were quantitatively analyzed by the method of inductively coupled plasma-mass spectroscopy (ICP-MS) in accord with Standard Method 3125. The following isotopes were targeted for analysis: 27Al, 42Ca, 56Fe, 24Mg, 55Mn, 31P, 28Si, and 66Zn. Analyses were carried out with a PerkinElmer ELAN DRC-e ICP-MS instrument equipped with an AS 93 Plus autosampler (PerkinElmer Instruments, Shelton CT). Atomization was achieved using a MicroMist nebulizer with baffled cyclonic spray chamber (PerkinElmer Instruments, Shelton CT). Data processing and acquisition were carried out using ELAN instrument software (version 3.3). Excluding iron, all inorganic elements were analyzed using the standard mode (involving a dynamic bandpass tuning parameter of 0.25). Iron measurements were made in the Dynamic Reaction Cell mode to remove interfering ions. This mode involved use of ammonia as a reaction gas (0.5 mL/min) and a dynamic bandpass tuning parameter of 0.50. A certified reference material (CRM) (River Water Reference Material for Trace Metals SLRS-4) was purchased from the National Research Council of Canada (Ottawa, Canada) to
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 7 0 e4 5 8 0
4573
evaluate ICP-MS performance. Results of ICP-MS analysis on this CRM are summarized in Table S2 in the Supplementary Information section. On the average, the deviation from expected values was 3.6% indicating a reasonable level of accuracy. Recoveries of known additions were also conducted. They ranged from 85% to 120%. Determination of total carbon (TC), total inorganic carbon (TIC) and total sulfur in crushed undigested samples was carried at the EPA laboratory in Cincinnati, OH, using a LECO model CS230 combustion furnace instrument (LECO Corporation, St. Joseph MI). TIC concentrations were determined by a modified ASTM D513 method. Total organic carbon (TOC) was calculated as the difference between TC and TIC values.
was identified in the hydrant flush samples, which may be indicative of its precipitation from the bulk distributed water. Quartz SiO2 was also frequently found. Its presence may be attributed to carryover from the source water or as treatment breakthrough. Hydroxyapatite Ca5(PO4)3(OH) was observed in cases where the utility applied orthophosphate or phosphate blend to control corrosion. It should be noted that while the possibility of the presence of some artifacts caused by sample processing (partial conversion of Fe(II) to Fe(III) caused by drying, Sarin et al., 2001) cannot be ruled out, XRD characterization of the samples used in this study provides an important insight into the nature of prevalent solid phases formed in drinking water distribution systems (Borch et al., 2008; Gerke et al., 2008).
3.
Results and discussion
3.3.
3.1.
Treated water chemistry
The elemental composition of DWDS samples is discussed below in the context of the occurrence of the common matrix elements, notably iron, sulfur, total organic carbon (TOC), calcium, inorganic carbon (TIC), phosphorous, manganese, magnesium, aluminum and zinc. This combination of major matrix components was determined based on the approach developed in our recently completed study undertaken under the auspices of Water Research Federation (Friedman et al., 2010). The data for silicon (Si) will not be considered since that element was not necessarily dissolved during the digestion procedure employed in this work. Since statistical analysis of the data showed that the concentrations of these elements were not normally distributed (Friedman et al., 2010), the results at selected percentiles (e.g., median) are emphasized over average values and standard deviations. Table 4 provides a statistical summary of the concentrations of common elements constituting the solids. In most cases, the reporting units are micrograms of element per gram of deposit (mg/g), or parts-per-million. When the median result for a given element exceeds 10,000 mg/g, the results are presented as weight percent (wt%). For reference, 10,000 mg/g is equivalent to 1.0 wt%. 61% and 83% of Samples processed in this study were pipe specimens and hydrant flush solids, respectively, formed on unlined cast iron. Thus, the data reported here are more representative of drinking water distribution systems in which unlined cast iron pipes predominate over other pipe materials.
Treated water conditions were ascertained through a combination of “snapshot” site-specific distribution system sampling (at locations where solid samples were obtained) and utility-provided records of entry-point and system monitoring results (Friedman et al., 2010). The compilation of distribution system water quality observations for each utility and sample site is provided in Table 2. It should be noted that the water quality conditions reported here reflect only the data obtained from either the water sampling performed specifically for this study or utility-provided water quality data representing entry-point sampling. Of the 20 utility participants, 17 utility participants had Fe concentrations at or above 0.06 mg/L (i.e., 20% of the secondary MCL) in entry-point and/or distribution system water samples. The median iron concentration in the treated water at sampling locations was 0.25 mg/L. A total of 11 utility participants reported and/or were found to have manganese present at concentrations exceeding 0.01 mg/L (i.e., 20% of the secondary MCL) in entry-point and/or distribution system water samples; however, the median Mn concentration in treated water at sampling locations was only 0.5 mg/L. Six utilities had dedicated iron and/or manganese removal processes (e.g., greensand filtration, permanganate-enhanced direct filtration, hydrous manganese oxide (HMO) filtration process) at problem sources. Five utilities used polyphosphate (either alone or as part of an ortho/poly blend) to sequester the soluble, reduced forms of these metals and/or to prevent excessive calcite precipitation.
3.2.
Morphological examination of corrosion scales
The morphological properties of selected samples of corrosion scales were examined using SEM/EDS and XRD. The examined samples typically lacked morphologically significant features (Supplementary Information Fig. S1). EDS analysis indicated that most frequently detected elements found on the surfaces were Fe, O, C, Si, S and Ca. XRD showed that goethite (a-FeOOH), magnetite (Fe3O4) and siderite (FeCO3) were major phases present in the samples (Table 3). This observation was in accordance with the findings of Sarin et al. (2001) and Barkatt et al., 2009. Calcite CaCO3
3.3.1.
Deposit composition-common matrix elements
Iron
Iron (Fe) was the most prevalent inorganic constituent in practically all samples. Thirty three of the 35 pipe specimens that contained enough mass for processing were composed of unlined iron or steel, and 19 of 23 hydrant flush samples that contained enough mass were obtained from unlined iron pipes. The median Fe concentration was 31.7 wt% and the 10th and 90th percentile Fe concentration were 11.8 wt% and 40.6 wt%, respectively. This was in agreement with the XRD data confirming the prominence of goethite a-FeOOH, magnetite Fe3O4, siderite FeCO3 and in some cases troilite FeS. Fig. 1(a) illustrates the cumulative iron occurrence profiles for all deposit samples and the different sample types (pipe specimens or hydrant flush solids). The Fe percentile profiles
4574
Table 2 e Distribution system water quality e general water quality parameters and other inorganic elements. Associated samples
Data source
pH
Temp ( C)
W CL CL CL CL CL SA SA SA SA CH RW IN CC DN CA PC WDB WA B B G G AZ BC J J NC ST ST ST ST K
A-H A, B C D, E F G A B, C D E A A, B A-D A-F A-B A-B A-B A-B A-D A, B, D C A B A A, B A-D, G-J E, F A A B C D A-D
Sample Sample Sample Survey Sample Sample Sample Sample Sample Sample Sample Sample Sample Sample Survey Sample Sample Survey Sample Sample Sample Sample Sample Sample Sample Survey Survey Sample Sample Sample Sample Sample Survey
7.7 8.2 8.5 NAa 8.3 7.9 8.1 7.4 8.0 8.1 8.0 7.1 7.4 7.6 7.8 7.3 7.6 7.6 7.2 8.1 7.8 7.9 NA 7.3 7.2 7.6 7.6 7.2 7.4 7.4 7.5 7.4 8.0
24 18 20 NA 19 19 19 17 16 18 19 15 20 14 12 18 2 NA NA 23 18 22 NA 31 37 NA NA 20 28 24 21 26 NA
a NA ¼ No data available.
Alkalinity (mg/L CaCO3) 147 238 244 NA 233 235 65 95 48 68 64 76 204 177 49 135 136 NA NA 119 115 277 276 267 150 289 280 278 265 236 232 261 268
Disinfectant residual (mg/L) 0.60 Comb. 0.68 Free 0.83 Free NA Free 0.42 Free 0.68 Free 0.02 Free 0.04 Free 0.01 Free 0.07 Free 0.12 Free 0.51 Free 0.65 Free 0.81 Free 1.35 Comb. 1.90 Free 0.3 Free NA Free < 0.1 Free 2.14 Comb. 1.84 Comb. 0.6 Free NA Free 0.3 Free 0.2 Free NA Comb. NA Comb. 0.13 Free 0.71 Free 0.97 Free 1.04 Free 0.58 Free 3.50 Comb.
Al (mg/L) 61 0.5 0.2 NA 1.1 0.7 5.1 0.3 1.7 1.0 2.1 8.4 4.4 <0.14 30 0.1 59 NA 507 145 134 22 3.1 <0.14 <0.14 NA NA <0.14 0.6 0.4 <0.14 0.5 NA
Ca (mg/L) 65 52 52 NA 55 60 19 10 9 10 23 17 64 45 26 44 106 19 12 27 29 2 2 310 155 91 59 60 26 52 43 50 63
Fe (mg/L) 0.27 0.14 0.15 NA 0.15 0.18 0.05 0.03 0.04 0.06 0.22 0.24 0.28 0.28 < 0.05 0.27 12.1 < 0.03 0.03 1.4 2.0 <0.001 <0.001 0.96 0.45 1.22 0.21 0.16 0.07 0.15 0.11 0.14 0.02
Mg (mg/L) 22 24 29 NA 28 25 4 6 4 3 5 7 30 14 NA 15 28 NA 3 5 9 1 1 93 40 38 20 33 16 35 34 33 28
Mn (mg/L) 0.04 0.06 0.03 NA 0.4 0.1 <0.01 0.1 1.2 0.3 0.3 0.5 6.6 3.9 < 6.0 0.2 0.02 50.0 11.1 0.1 0.8 0.2 0.1 29.9 0.1 55.4 7.8 30.0 <0.01 <0.01 <0.01 1.9 9.0
Si (mg/L) 4 4 3 NA 4 4 14 15 5 15 3 12 3 10 NA 9 5 NA 7 6 6 3 3 10 13 NA 4.1 3 4 4 5 4 6.9
S (mg/L) 161 19 11 NA 15 33 3 5 8 2 7 4 24 25 NA 30 69 NA 12 20 27 2 2 462 144 NA NA 12 9 9 9 9 NA
P (mg/L) 0.45 0.40 0.57 NA 0.58 0.51 0.25 0.06 0.07 0.3 0.01 0.12 0.13 0.46 NA 0.37 0.04 NA 0.01 <0.002 0.09 0.01 0.01 0.02 0.01 NA NA 0.31 0.12 0.01 0.01 0.06 NA
Zn (mg/L) 3.1 2.4 28.1 NA < 0.06 11.1 23.0 9.8 < 0.06 < 0.06 3.6 3.9 2.3 1.6 < 5.0 158 97.3 10.0 6.0 1.0 22.3 <0.06 <0.06 4.4 47.2 18 < 20 <0.06 <0.06 <0.06 <0.06 <0.06 NA
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 7 0 e4 5 8 0
Utility identifier
4575
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 7 0 e4 5 8 0
0.4 wt% and 3.3 wt%, respectively. It was the third most common component of the corrosion scales. The occurrence of organic carbon may be a result of autochthonous process (biofilm growth) and sorption of NOM present in source waters. Fig. 2(b) illustrates the cumulative organic carbon occurrence profiles for all sample types. TOC levels in hydrant flush solids were consistently higher that those in the pipe specimens, with median TOC concentrations of 1.7 wt% and 0.7 wt%, respectively. The observed dissimilarity between these two sample types could be ascribed to various factors. For instance, hydrant flush solids can be hypothesized to retain more sorbed/occluded NOM or have higher rate of microbiological processes in them. It can be also hypothesized that some loss of labile organic carbon may have occurred in boneyard samples that were exposed to air for prolonged periods of time.
Table 3 e Crystal phase identification of pipe deposits. Sample ID
Pipe material
X-ray diffraction results
Pipe specimen sample CC-A Cast iron CC-D Cast iron CH-A Steel PC-A Galvanized iron PC-B Galvanized iron RW-A Galvanized iron RW-B Galvanized iron
Magnetite, quartz Siderite, quartz, hydroxyapatite Goethite, magnetite, quartz Goethite, magnetite, siderite, troilite Goethite, siderite Magnetite, troilite Magnetite, troilite
Hydrant flush samples J-B Cast iron J-E Cast iron J-J Cast iron
Goethite, siderite, calcite, hydroxyapatite Calcite, dolomite, quartz Goethite, magnetite, ferrihydrite, calcite, quartz
3.3.4.
for pipe specimens and hydrant flush solids were similar, with median Fe concentrations of 32.7 wt% and 28.4 wt%, respectively.
3.3.2.
Sulfur
Sulfur (S) was the second-most prevalent inorganic constituent in the solids, with a median concentration of 11,100 mg/g (1.1 wt%). The 10th and 90th percentile S concentrations were 0.15 wt% and 2.8 wt%, respectively. Sulfur occurrence in the scales is likely to be due to the formation of troilite, a common under-layer component of iron corrosion scale (Benjamin et al., 1996). The XRD pattern for several samples (e.g., J-E, PC-A, RW-B) confirmed its presence. Fig. 2(a) illustrates the cumulative sulfur occurrence profiles for all sample types. The profiles for pipe specimens and hydrant flush solids are very similar. The median S concentrations for pipe specimens and hydrant flush solids are 1.0 wt% and 1.1 wt%, respectively.
3.3.3.
Organic carbon
The median concentration of organic carbon (TOC) was 9800 mg/g (0.98 wt%), while its 10th and 90th percentiles were
Calcium
Calcium (Ca) was the fourth most common element, with a median concentration of 7700 mg/g (0.8 wt%). The 10th and 90th percentile Ca concentrations were 420 mg/g and 8.8 wt%, respectively. Calcium occurrence in deposits is likely to take place via the deposition of calcite, dolomite CaMg(CO3)2, and hydroxyapatite Ca5(PO4)3OH. Each of these minerals was identified in XRD analyses. Hydroxyapatite was observed only in the samples from utilities that applied phosphate as part of their treatment process. While calcite was found in samples associated with “hard” waters, soluble Ca2þ and its complexes can associate with metal-oxide substrates due to surface adsorption (Ali and Dzombak, 1996; Weng et al., 2005). This may explain calcium occurrence in cases where the water was under-saturated with respect to calcium-based mineral phases. Fig. 1(b) illustrates the cumulative calcium occurrence profiles for all deposit samples and the different sample types. The profiles for pipe specimens and hydrant flush solids are similar, with the median Ca concentrations 0.9 wt% and 0.74 wt%, respectively.
3.3.5.
Inorganic carbon
Inorganic carbon (TIC) was the fifth most prevalent element, with a median concentration of 2500 mg/g. The 10th and 90th percentile TIC concentrations were 50 mg/g and 4.6 wt%, respectively. The occurrence of inorganic carbon in deposits is due to the presence of siderite, calcite, dolomite (Table 3) and
Table 4 e Statistical summary of elemental occurrence in deposit samples. Common element
No. of samples
Average result
Standard deviation
Minimum result
10th Percentile
Median result
90th Percentile
Maximum result
Al (mg/g) Ca (mg/g) Fe (wt%) Mg (mg/g) Mn (mg/g) S (wt%) Si (mg/g) P (mg/g) Zn (mg/g) TIC (mg/g) TOC (mg/g)
58 58 58 58 58 48 58 58 58 36 36
1630 29300 28.5 3190 7320 1.4 167 2250 1370 13300 19000
3150 56500 11.8 6860 31200 1.7 234 2520 3980 23200 34100
32 <0.33 0.1 49 100 0.05 <0.22 100 3.2 0.01 0.01
120 420 11.8 110 290 0.15 <0.22 450 24 50 4140
620 7700 31.7 640 790 1.1 90 1400 230 2500 9800
3400 87500 40.6 6950 7000 2.8 390 4300 1900 45800 32600
20300 252 700 46.8 37 900 232 500 10.9 1330 12 600 19 700 108 500 206 700
4576
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 7 0 e4 5 8 0
a
100%
All solid samples Hydrant flush solids Pipe specimens
Sample Percentile
80%
60%
40%
20%
0% 10,000
100,000
1,000,000
Iron concentration, µg/g
b
100%
All solid samples Hydrant flush solids Pipe specimens
Sample Percentile
80%
60%
40%
20%
0% 100
1,000
10,000
100,000
1,000,000
10,000
100,000
Calcium concentration, µg/g
c 100%
All solid samples Hydrant flush solids Pipe specimens
Sample Percentile
80%
60%
40%
20%
0% 10
100
1,000
Magnesium concentration, µg/g
Fig. 1 e Cumulative occurrence profiles for (a) iron, (b) calcium, and (c) magnesium in corrosion scales and deposits.
surface adsorption/co-precipitation reactions involving bicarbonate and carbonate species. XRD patterns showed that the aforementioned mineral phases were present in several samples. Fig. 2(c) illustrates the cumulative TIC occurrence profiles for all deposit samples and the different sample types. The Ca levels in hydrant flush solids were higher than those in pipe specimens across the entire range. The median TIC concentrations for pipe specimens and hydrant flush solids are 500 mg/g and 3000 mg/g, respectively.
3.3.6.
Phosphorus
Phosphorus (P) was the sixth most prevalent element, with a median concentration of 1400 mg/g. The 10th and 90th percentile phosphorus concentrations were 450 mg/g and
4300 mg/g (0.43 wt%), respectively. Phosphorus occurrence in the corrosion scale can be associated with the formation of hydroxyapatite Ca5(PO4)3OH (Table 3), adsorption of orthophosphate onto various mineral surfaces, and phosphorus uptake/accumulation in biofilm and cellular material. The accumulation of phosphorus in the scales requires that it be present in the treated water. The total phosphorus concentration in water samples collected for this study ranged from non-detect (MQL of 0.001 mg/L) to 0.6 mg/L, with a median of 0.1 mg/L. Seven of the 20 utility participants reported adding phosphate-based chemicals in their treatment process. Several other utility participants have moderate levels of phosphorus (of unknown speciation) originating in one or more of their sources of supply. Despite that,
4577
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 7 0 e4 5 8 0
a
100%
Sample Percentile
80%
All solid samples Hydrant flush solids Pipe specimens
60%
40%
20%
0% 100
1,000
10,000
100,000
1,000,000
Sulfur concentration, g/g
b
100%
Sample Percentile
80%
All solid samples Hydrant flush solids Pipe specimens
60%
40%
20%
0% 1000
10000
100000
1000000
Organic carbon concentration, g/g
c
100%
Sample Percentile
80%
All solid samples Hydrant flush solids Pipe specimens
60%
40%
20%
0% 10
100
1,000
10,000
100,000
Inorganic carbon concentration, g/g
d
100%
All solid samples Hydrant flush solids Pipe specimens
Sample Percentile
80%
60%
40%
20%
0% 100
1,000
10,000
100,000
Phosphorus concentration, g/g
Fig. 2 e Cumulative occurrence profiles for (a) sulfur, (b) TOC, (c) TIC, and (d) phosphorus in corrosion scales and deposits.
4578
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 7 0 e4 5 8 0
little correlation between phosphorus levels in water and solids was observed (Supplementary Information Fig. S2). Fig. 2(d) illustrates the cumulative phosphorus occurrence. The profiles for pipe specimens and hydrant flush solids are similar. The median phosphorus concentrations for pipe specimens and hydrant flush solids are 1360 mg/g and 1600 mg/ g, respectively.
the manganese found in the treated water. Fig. 3(a) illustrates the cumulative manganese occurrence profiles for all deposit samples and the different sample types. The results for pipe specimens were higher in manganese than hydrant flush solids. The median manganese concentrations for pipe specimens and hydrant flush solids are 940 mg/g and 610 mg/g, respectively.
3.3.7.
3.3.8.
Manganese
Manganese (Mn) was the seventh most abundant element found in the samples, with a median concentration of 790 mg/ g. The 10th and 90th percentile Mn concentrations were 290 mg/g and 7000 mg/g (0.7 wt%), respectively. Manganese occurrence in deposits is likely to be due to the formation and deposition of manganese oxyhydroxide solids starting from
a
100%
Magnesium
Magnesium (Mg) was the eighth most abundant element found in deposit samples, with a median concentration of 640 mg/g. The 10th and 90th percentile Mg concentrations were 110 mg/g and 6950 mg/g (0.7 wt%), respectively. Magnesium occurrence in deposits is expected to be due to the formation and deposition of minerals such as dolomite (Table 3). Fig. 1(c)
All solid samples Hydrant flush solids Pipe specimens
Sample Percentile
80%
60%
40%
20%
0% 100
1,000
10,000
100,000
1,000,000
Manganese concentration, g/g
b
100%
All solid samples Hydrant flush solids Pipe specimens
Sample Percentile
80%
60%
40%
20%
0% 10
100
1,000
10,000
100,000
Aluminum concentration, g/g
c
100%
All solid samples Hydrant flush solids Pipe specimens
Sample Percentile
80%
60%
40%
20%
0% 1
10
100
1,000
10,000
100,000
Zinc concentration, g/g
Fig. 3 e Cumulative occurrence profiles for (a) manganese, (b) aluminum, and (c) zinc in corrosion scales and deposits.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 7 0 e4 5 8 0
illustrates the cumulative magnesium occurrence profiles. The results for hydrant flush solids were higher than pipe specimens, with the median Mg concentrations 800 mg/g and 340 mg/g, respectively.
3.3.9.
Aluminum
Aluminum (Al) was the ninth most common element found in deposit samples, with a median concentration of 620 mg/g. The 10th and 90th percentile aluminum concentrations were 120 mg/g and 3400 mg/g, respectively. Aluminum presence in deposits can be due to the deposition of alumina Al2O3, gibbsite Al(OH)3, precipitation of amorphous aluminum hydroxide, and formation of aluminosilicates. Surface adsorption/co-precipitation reactions involving free Al3þ and its complexes may also account for its occurrence. Sources of aluminum may include the treated water, either due to natural occurrence in source water and/or the application and treatment “breakthrough” of aluminum-based coagulants. Fig. 3(b) illustrates the cumulative aluminum occurrence profiles for all samples. The profiles for pipe specimens and hydrant flush solids are dissimilar, with the Al levels in pipe specimens being higher across the entire range. The median Al concentrations for pipe specimens and hydrant flush solids are 640 mg/g and 555 mg/g, respectively.
3.3.10. Zinc Zinc (Zn) was the tenth most abundant element found in deposit samples, with a median concentration of 230 mg/g. The 10th and 90th percentile zinc concentrations were 24 mg/g and 1900 mg/g, respectively. Sources of zinc may include the treated water, either due to natural occurrence in source water, applications of zinc orthophosphate and “inner” sources such as its presence as in galvanized pipe and as a component in copper-based alloys. Internal corrosion of galvanized iron piping appears to be the primary source of zinc in many samples. Indeed, comparison between cast iron and galvanized iron specimens showed that Zn concentration was much higher in the latter case, with median Zn concentrations being 185 mg/g and 8422.5 mg/g (Supplementary Information Table S2). Fig. 3(c) illustrates the cumulative Zn occurrence profiles for all deposit samples and the different sample types. The profiles for pipe specimens and hydrant flush solids are dissimilar, with the results for pipe specimens being higher across the entire range. The median zinc concentrations for pipe specimens and hydrant flush solids are 290 mg/g and 110 mg/g, respectively.
4.
Conclusions
Characteristics of corrosion scales formed in drinking water distribution systems predominated by unlined cast iron pipes and deposits mobilized during hydrant flushing events were determined using SEM/EDS, XRD and ICP/MS. XRD data showed that goethite (a-FeOOH), magnetite (Fe3O4) and siderite (FeCO3) were the primary crystalline phases identified in most of the samples. Among the major constituent elements of the scales, iron was most prevalent by a considerable margin, followed, in the order of decreasing prevalence, by sulfur, organic carbon, calcium, inorganic carbon, phosphorus, manganese, magnesium, aluminum and zinc.
4579
The nature of relatively abundant organic carbon found in the scales remains to be determined. The cumulative occurrence profiles of iron, sulfur, calcium and phosphorus for pipe specimens and hydrant flush solids were similar. For TOC, TIC and magnesium, the cumulative occurrence profiles showed that hydrant flush solids have consistently higher levels of these components compared with pipe specimens. On the other hand, the cumulative occurrence profiles for manganese, aluminum and zinc indicated that pipe specimens tended to have higher concentrations of these elements than hydrant flush solids. Comparison of relative occurrences of these elements indicates that hydraulic disturbances may have relatively less impact on the release of manganese, aluminum and zinc. However, observations concerning differences of concentrations of selected elements in pipe specimens and hydrant flush solids need to be confirmed in further exploration of hydraulically immobile and mobile solids originating from the same systems. Zinc concentrations in the scales formed on galvanized iron were much higher than those formed on cast iron, with internal corrosion suspected of being the major sources of zinc in the former case.
Acknowledgments This study was supported by Water Research Foundation (Project Number 3118) and the USEPA. The authors would like to thank the WRF project manager Dr. Jian Zhang and the personnel of the EPA laboratory, Cincinnati, OH for carrying out analyses for carbon and sulfur in solid samples. The content and conclusions are the views of the authors and do no necessarily reflect the views of the funding agency.
Appendix. Supplementary data The supplementary data associated with this article can be found in the on-line version at doi:10.1016/j.watres. 2010.05.043.
references
Ali, M.A., Dzombak, D.A., 1996. Effects of simple organic acids on sorption of Cu2þ and Ca2þ on goethite. Geochimica et Cosmochimica Acta 60 (2), 291e304. U.S. EPA method 3050B, Acid digestion of sediments, sludges and soils. http://www.epa.gov/waste/hazard/testmethods/sw846/ pdfs/3050b.pdf. Barkatt, A., Pulvirenti, A.L., Adel-Hadadi, M.A., Viragh, C., Senftle, F.E., Thorpe, A.N., Grant, G.R., 2009. Composition and particle size of superparamagnetic corrosion products in tap water. Water Research 43 (13), 3319e3325. Benjamin, M.M., Sontheimer, H., Leroy, P., 1996. Corrosion of Iron and Steel. In Internal Corrosion of Water Distribution Systems. Cooperative Research Report. AWWA Research Foundation, Denver, CO, pp. 29e70.
4580
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 7 0 e4 5 8 0
Borch, T., Camper, A.K., Biederman, J.A., Butterfield, P.W., Gerlach, R., Amonette, J.E., 2008. Evaluation of characterization techniques for iron pipe corrosion products and iron oxide thin films. Journal of Environmental Engineering-ASCE 134 (10), 835e844. Cance`s, B., Juillot, F., Morin, G., Laperche, V., Alvarez, L., Proux, O., Hazemann, J.-L., Brown Jr., G.E., Calas, G., 2005. XAS evidence of As(V) association with iron oxyhydroxides in a contaminated soil at a former arsenical pesticide processing plant. Environmental Science & Technology 39 (24), 9398e9405. Fendorf, S., Eick, M.J., Grossl, P., Sparks, D.L., 1997. Arsenate and chromate retention on goethite: I. Surface structure. Environmental Science & Technology 31 (2), 315e320. Friedman, M.J., Hill, A.S., Reiber, S.H., Valentine, R.L., Larsen, G., Young, A., Korshin, G.V., Peng, C.-Y., 2010. Assessment of Inorganics Accumulation in Drinking Water System Scales and Sediments. Water Research Foundation, Denver, CO. Gerke, T.L., Maynard, J.B., Schock, M.R., Lytle, D.L., 2008. Physiochemical characterization of five iron tubercles from a single drinking water distribution system: possible new insights on their formation and growth. Corrosion Science 50 (7), 2030e2039. Gerke, T.L., Scheckel, K.G., Schock, M.R., 2009. Identification and distribution of vanadinite (Pb5(V5þO4)3Cl) in lead pipe corrosion by-products. Environmental Science & Technology 43 (12), 4412e4418. Gray, C.W., McLaren, R.G., Roberts, A.H.C., Condron, L.M., 1999. Solubility, sorption and desorption of native and added cadmium in relation to properties of soils in New Zealand. European Journal of Soil Science 50 (1), 127e137. Korshin, G.V., Ferguson, J.F., Perry, S., 1996. Influence of natural organic matter on corrosion of copper in potable waters. Journal of the American Water Works Association 88 (7), 36e47. Larsen, F., Postma, D., 1997. Nickel mobilization in a groundwater well field: release by pyrite oxidation and desorption from manganese oxides. Environmental Science & Technology 31 (9), 2589e2595. Lytle, D.A., Sorg, T.J., Frietch, C., 2004. Accumulation of arsenic in drinking water distribution systems. Environmental Science & Technology 38 (20), 5365e5372.
Nelson, Y.M., Lo, W., Lion, L.W., Shuler, M.L., Ghiorse, W.C., 1995. Lead distribution in a simulated aquatic environment: effects of bacterial biofilms and iron oxide. Water Research 29 (8), 1934e1944. O’Reilly, S.E., Strawn, D.G., Sparks, D.L., 2001. Residence time effects on arsenate adsorption/desorption mechanisms on goethite. Soil Science Society of America Journal 65 (1), 67e77. Reiber, S., Dostal, G., 2000. Well water disinfection sparks surprises. Opflow 26 (3), 1e6. Sarin, P., Snoeyink, V.L., Bebee, J., Kriven, W.M., Clement, J.A., 2001. Physico-chemical characteristics of corrosion scales in old iron pipes. Water Research 35 (12), 2961e2969. Schock, M.R., Hyland, R., Welch, M., 2008. Occurrence of contaminant accumulation in lead pipe scales from domestic drinking water distribution systems. Environmental Science & Technology 42 (12), 4285e4291. Sugiyama, M., Hori, T., Kihara, S., Matsui, M., 1992. A geochemical study on the specific distribution of barium in lake Biwa, Japan. Geochimica et Cosmochimica Acta 56 (2), 597e605. Teng, F., Guan, Y.T., Zhu, W.P., 2008. Effect of biofilm on cast iron pipe corrosion in drinking water distribution system: corrosion scales characterization and microbial community structure investigation. Corrosion Science 50 (10), 2816e2823. Trived, P., Axe, L., Tyson, T.A., 2001. XAS studies of Ni and Zn sorbed to hydrous manganese oxide. Environmental Science & Technology 35 (22), 4515e4521. Tuovinen, O.H., Button, K.S., Vuorinen, A., Carlson, L., Mair, D.M., Yut, L.A., 1980. Bacterial, chemical, and mineralogical characteristics of tubercles in distribution system pipelines. Journal of the American Water Works Association 72 (11), 626e635. Vazquez, F.A., Heaviside, R., Tang, Z., Taylor, J.S., 2006. Effect of free chlorine and chloramines on lead release in a distribution system. Journal of the American Water Works Association 98 (2), 144e154. Weng, L.P., Koopal, L.K., Hiemstra, T., Meeussen, J.C.L., Van Riemsdijk, W.H., 2005. Interactions of calcium and fulvic acid at the goethiteewater interface. Geochimica et Cosmochimica Acta 69 (2), 325e339.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 1 1 e4 4 2 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Climate change induced salinisation of artificial lakes in the Netherlands and consequences for drinking water production Matthijs Bonte*, John J.G. Zwolsman KWR Watercycle Research Institute, PO Box 1072, 3430BB Nieuwegein, The Netherlands
article info
abstract
Article history:
In this paper we present a modelling study to investigate the impacts of climate change on
Received 12 February 2010
the chloride concentration and salinisation processes in two man-made freshwater lakes
Received in revised form
in the Netherlands, Lake IJsselmeer and Lake Markermeer. We used a transient compart-
25 May 2010
mental chloride and water balance model to elucidate the salinisation processes occurring
Accepted 2 June 2010
under present conditions and assess future salinisation under two climate forcing
Available online 9 June 2010
scenarios. The model results showed that the Rhine River is the dominant determinant for the chloride concentration in both lakes, followed by drainage of brackish groundwater
Keywords:
from the surrounding polders. The results further show that especially during dry years,
Chloride
seawater intrusion through the tidal closure dam is an important source of chloride to Lake
Modelling
IJsselmeer. The results from the climatic forcing scenarios show that Lake IJsselmeer is
Salinisation
especially vulnerable to climate-induced salinisation whereas effects on Lake Markermeer
Water supply
are relatively small. Peak chloride concentrations at the raw water intake of the Andijk
Climate change
drinking water facility on Lake IJsselmeer are projected to increase to values above 250 mg/l in the most far-reaching climate change scenario Wþ in 2050 for dry years. This is well above the maximum allowable concentration of 150 mg/l for chloride in drinking water. Modelling showed that climate change impacts the chloride concentrations in a variety of ways: 1) an increasing occurrence of low river flows from summer to autumn reduces the dilution of the chloride that is emitted to the Rhine with a constant load thereby increasing its concentration; 2) increased open water evaporation and reduced rainfall during summer periods and droughts increases the chloride concentration in the water; and 3) rises in sea level increase seawater intrusion through the tidal closure dam of Lake IJsselmeer. The processes described here are likely to affect many other tidal rivers or lakes and should be considered when planning future raw water intake stations for drinking water production or agricultural water supply. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Salinisation due to poor land use practices, excessive groundwater extractions and rising sea levels is increasingly putting a strain on the availability of freshwater resources worldwide. Salinisation, defined here as spatially
or temporally increasing concentrations of chloride and sodium, can result in decreasing agricultural yields (Rozema and Flowers, 2008), negative impacts on aquatic ecosystems (Kaushal and Gene, 2009) and loss of suitable water resources for drinking water production (Delpla et al., 2009).
* Corresponding author. Tel.: þ31 (0)30 60 69 761; fax: þ31 (0)30 60 61 165; E-mail address:
[email protected] (M. Bonte). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.004
4412
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 1 1 e4 4 2 4
Lake IJsselmeer and Lake Markermeer form the largest artificial freshwater system in both the Netherlands and northwestern Europe and are fed primarily by the Rhine River. Both lakes are an important source of freshwater for drinking water production, agriculture and industry in the Netherlands. During the second half of the previous century, salinisation due to salt emissions mainly from potash and brown coal mining activities in the catchment area, caused the water quality of the Rhine River and Lake IJsselmeer to deteriorate with chloride concentrations in excess of 400 mg/l (Molt, 1961; Van Der Weijden and Middelburg, 1989). Growing concern regarding the water quality of the Rhine led to a series of treaties and plans to address the issue, the most important being the Rhine Salt Treaty (1976) and the Rhine Action Plan (1987) which aimed at reducing salt emissions and general improvement of river water quality, respectively (Dieperink, 2000). Over previous decades, monitoring data showed the effectiveness of these treaties in that the average annual chloride concentration gradually declined to values below 100 mg/l. Though emission reduction clearly had a positive impact on the average concentrations, peak values near 200 mg/l are still observed in the Rhine near the DutcheGerman border (at Lobith) and Lake IJsselmeer (at Andijk) during periods of low flow (Zwolsman and Bokhoven, 2007). Hydrological modelling has shown that the frequency and duration of such low flow periods are likely to increase as a result of climate change (Van Deursen, 2006). This suggests that the impacts of climate change are of great importance for the future water supply of the Netherlands. While quantitative numerical assessments of the impacts of climate change on flow conditions of river systems and large lakes are well documented in literature (Parmet et al., 1995; Middelkoop et al., 2001; Middelkoop and Kwadijk, 2001; Shabalova et al., 2003; Van Deursen, 2006; Te Linde, 2007; Leander et al., 2008; Jonkeren et al., in press), effects on water quality have received far less attention. Most studies examining climate effects on water quality have a qualitative or semi-quantitative nature (Senhorst and Zwolsman, 2005; Zwolsman and Bokhoven, 2007; van Vliet and Zwolsman, 2008). Zwolsman and Bokhoven (2007) for example analysed water quality in the Rhine during relatively dry years with low river flows and found a strong increase in concentrations of chloride, nutrients and heavy metals when compared to average hydrological conditions. More quantitative assessments of climate change effects on water quality reported in literature mainly deal with eutrophication (Arheimer et al., 2005; Cugier et al., 2005; Evans, 2005; Loos et al., 2009; Schauser and Chorus, 2009). Arheimer et al. (2005) showed that climate change may have a detrimental effect on the water quality of Lake Ringsjo¨n in Sweden (comparable in depth and residence time to Lake Markermeer) with concentrations of nitrogen, phosphate and cyanobacteria increasing by 20, 50 and 80% respectively. Here, we present a modelling study to investigate the impacts of climate change and associated hydrological shifts on the chloride concentration and salinisation processes in the Lake IJsselmeer region. Chloride is especially relevant for drinking water production as it is a compound that is not removed by conventional drinking water treatment systems
in the Netherlands. In this paper we present the construction, calibration and validation of a transient compartmental water and chloride model. We used this model to elucidate the salinisation processes occurring under present conditions and to assess future salinisation under two climate forcing scenarios.
2.
Study site
Lakes IJsselmeer and Markermeer are part of a larger system of connected freshwater lakes referred to as the IJsselmeer region, which includes the lakes IJsselmeer proper, Markermeer, Gooimeer, Eemmeer, IJmeer and Veluwemeer (Fig. 1). This paper focuses on lakes IJsselmeer and Markermeer which cover over 80% of the area of the IJsselmeer region. Table 1 shows the main physiographical characteristics of the two lakes. Lake IJsselmeer was created in 1932 by the construction of a 30 km long tidal closure dam, called the “Afsluitdijk”, separating the Wadden Sea from the former northern Rhine estuary, called the “Zuiderzee”. Initially, Lake IJsselmeer was a large brackish water lake that gradually freshened as relatively fresh river water flushed the former estuary. In 1942 and 1968, large land reclamation projects resulted in two islands in Lake IJsselmeer, the Noordoostpolder and the Flevopolder. In 1976, the lakes IJsselmeer and Markermeer were physically separated by the completion of a dam, the “Houtribdijk”. Two sluices in the “Houtribdijk” control water transfer between the two lakes. Water levels in the two lakes are controlled to preset winter and summer level through a series of sluices used to discharge water gravitationally to the Wadden Sea and North Sea during low tide. A water balance constructed for the years 2002e2004 shows that Lake IJsselmeer receives most of its water from the IJssel River (70% on average), with the remainder coming from rainfall, exchange with Lake Markermeer and a number of polders that drain directly to the lake (Syncera Water, 2006). The IJssel River is a branch of the Rhine River receiving a fixed fraction of about 11% of total Rhine river discharge. The main inflows of chloride into Lake IJsselmeer include the IJssel River itself, drainage from the surrounding polders and seepage from the Wadden Sea through the tidal closure dam. The main inputs of water and chloride to Lake Markermeer are exchange with Lake IJsselmeer (predominantly Rhine River water), rainfall and polder drainage.
3.
Methods
3.1.
Numerical model
In order to simulate the chloride dynamics of lakes IJsselmeer and Markermeer, we built a compartmental cell model for water and chloride that simulates the chloride concentration on a daily interval over the period of 1997e2007. The model consists of three main cells representing lakes IJsselmeer, Markermeer and the combined lakes Gooi- and Eemmeer (Fig. 2) of which the latter is included as a flux boundary and not used to simulate chloride concentrations.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 1 1 e4 4 2 4
4413
Fig. 1 e Map showing the Lake IJsselmeer region and the Rhine catchment
Inflowing water and chloride are instantaneously mixed in the cells representing Lake Markermeer and lakes Gooi- and Eemmeer. Lake IJsselmeer was subdivided into a series of cells in which water and chloride are instantaneously mixed in order to better approximate chloride transport in this part of the system. Transport of chloride between the three main cells occurs only through advection whereas between the cells in Lake IJsselmeer chloride transport occurs through both advection and diffusion. The diffusion term is used to simulate wind-induced mixing processes, a concept originally proposed for Lake Tjeukemeer by Leenen (Leenen, 1982a; Leenen, 1982b). The complete conceptual model is illustrated in Fig. 2. The water balance of each cell in the model is described by: X
Qin EA Qout ¼ A
dh dt
(1)
P where Qin is the sum of the individual water inflow terms to a cell [m3/day] (such as rainfall, river inflow, polder drainage), E is the evaporation flux [m/day], A is the area of a cell [m2], Qout is the outflow of water from a cell [m3/day], h is the water level in a cell [m] and t represents time [day]. The chloride balance of each cell is described by: X dc dh Qin cin Qout c ¼ A h þ c (2) dt dt where cin is the chloride concentration of a certain inflow term to a cell [g/day]. The water level time derivative in Eq. (2) is eliminated with Eq. (1), which gives: X
dc X Qin cin ¼ Ah þ Qin EA c dt
(3)
The solution of Eq. (3) subject to initial condition c ¼ c0 at t ¼ t0 and assuming that the change in water level between t0 and t is small, is: P P P Qin EA Qin cin Qin EA c ¼ c0 exp t 1 exp t þP ; Qin EA Ah Ah X Q in sEA ð4aÞ P c ¼ c0 þ
Qin cin X t; Q in ¼ EA Ah
(4b)
Eq. (4) is applied using daily gauging data for Qin, cin, E and h as input data and the calculated chloride concentration of one time step as the initial concentration of the next. Eq. (4) is used going ‘with the flow’, i.e., we start at the upstream end (point of inflow of the IJssel River into Lake IJsselmeer) and solve each cell going downstream. This is required as the outflow concentration of the upstream cell is necessary for the inflow concentration of the downstream cell. Diffusion of chloride between the cascading cells in Lake IJsselmeer is simulated with a numerical approximation using a Taylor series expansion of the diffusion equation, as described by Appelo and Postma (Appelo and Postma, 2005): cx;corrected ¼ cx þ
Df Dt ðcx1 2$cx þ cxþ1 Þ ðDx2 Þ
¼ mixf $cx1 þ mixf $cxþ1 þ ð1 2mixf Þ$cx
(5)
with: mixf ¼
Df Dt ðDx2 Þ
(6)
4414
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 1 1 e4 4 2 4
Table 1 e Physiographic characteristics of lakes IJsselmeer and Markermeer
2
Area (km ) Average depth (m) Residence time (month) Desired winter water level (m-ref) Average historic winter water level (m-ref) Desired summer water level (m-ref) Average historic summer water level (m-ref)
IJsselmeer
Markermeer
1080 4.7 3e4 0.40 0.25
700 3.9 15e18 0.40 0.33
0.20 0.19
0.20 0.20
where cx,corrected is the chloride concentration corrected for diffusion and cx, cx1 and cxþ1 represent the chloride concentrations in the cell under consideration calculated with Eq. (4), the cell upstream and the cell downstream, respectively. Df represents the diffusion coefficient [m2/day] and Dx is the cell length. It should be noted that Df does not represent the free diffusion of chloride in water, but instead is used to simulate wind-induced mixing processes. This transport term is increasingly important when through flow in Lake IJsselmeer is negligible as it causes peak concentrations to be levelled out over time. The number of cells used to simulate Lake IJsselmeer also has influence on the magnitude of dispersion. This is the result of numerical dispersion in the model occurring if the
cell size is large compared to the product of the water velocity and the length of a time step (the travel length of a given body of water in one time step). The effect of numerical dispersion and the simulated diffusion can be combined in an apparent dispersion factor (Appelo and Postma, 2005): mixf ðDxÞ2 Dx vDt þ v (7) DL ¼ Df þ anum v ¼ Dt 2 where DL is the apparent dispersion factor [km2/day], anum is the numerical dispersivity [km], v is the velocity of water flow [km/day], Dx is the cell size, perpendicular to the direction of water flow [km] and Dt is the size of the time step [day] which in our model is 1 day. Eq. (7) shows that for a constant time step and cell size (as is the case in our model) the numerical dispersion increases with flow velocity. This is contrary to the effect of the diffusion term in the model. The number of cells (N ) and the mix factor (mixf ) are both parameters used in model calibration. Calibration of the model was accomplished by minimising the mean absolute error (MAE) between the measured and simulated chloride concentrations in the western part of Lake IJsselmeer (at the Andijk water intake station) and a monitoring point in Lake Markermeer (Markermeer Middle, see Fig. 2 for locations). The MAE is calculated as: P jcsimulated cmeasured j (8) MAE ¼ n where csimulated and cmeasured are the simulated and measured chloride concentrations, respectively, and n is the number of
Fig. 2 e Conceptual model of the Lake IJsselmeer region, comprising Lake IJsselmeer proper, Lake Markermeer and lakes Gooi & Eem.
Table 2 e Adopted relative changes on water and chloride budget terms to simulate the effects of climate forcing Month
Rainfalla Evaporationb
IJssel chloride concentrationd
Dike seepage Dike seepage and Polders Polder chloride Polders Polder and sluice sluice losses e discharge e load e discharge e chloride load losses e dischargee chloride concentration IJsselmeerf IJsselmeerf Markermeerf Markermeerf
rise and no changes to air circulation in W-Europe), 2050 þ1% þ8% 2% þ2% þ7% 2% þ2% þ5% 1% þ2% þ3% 1% þ3% þ1% 0% þ3% þ1% 0% þ50% þ3% 0% 0% þ3% 1% 0% þ3% 0% 0% þ2% þ2% 1% þ2% þ4% 2% þ2% þ7% 2%
Scenario Wþ (2 C temperature rise and changed air circulation in W-Europe), 2050 January þ15% þ3% þ14% 4% February þ14% þ3% þ18% 5% March þ10% þ5% þ16% 4% April þ2% þ7% þ12% 3% May 7% þ10% þ2% 0% June 15% þ13% 12% þ5% þ70% July 21% þ16% 25% þ13% August 22% þ17% 34% þ21% September 21% þ15% 37% þ24% October 6% þ11% 33% þ19% November þ6% þ6% 19% þ7% December þ13% þ3% þ4% 1% a b c d e f
0%
þ1%
þ1%
0%
1%
0%
1%
þ1%
1%
þ2%
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 1 1 e4 4 2 4
Scenario G (1 C temperature January þ4% February þ4% March þ3% April þ3% May þ2% June þ2% July þ2% August þ3% September þ2% October þ3% November þ3% December þ4%
IJssel dischargec
Based on Climate Scenario’s by KNMI (Van Den Hurk et al., 2007) and transformed to daily time series using on-line tool by Bakker (Bakker and Bessembinder, 2007; Bakker, 2008); As a, with interpolation to monthly values by Lenderink (Lenderink, 2006); Based on Rhineflow modelling calculations (Van Deursen, 2006); Based on empirical chloride-discharge model (Davis and Zobrist, 1978; Van Der Weijden and Middelburg, 1989; Zwolsman and Bokhoven, 2007), assuming a constant load; Linear annual average change derived from projected sea level rise in the year 2050; Annual average change based on Modflow/MOCDENSE3D variable density groundwater flow modelling (Delsman et al., 2009)
4415
4416
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 1 1 e4 4 2 4
observations. Although the MAE is an adequate parameter for optimising the model, it does not hold any information on the performance of the model compared to the observed variation in chloride concentration. For this purpose we use the Mean Relative Error (MRE), which is calculated as: MAE ¼
MAE MAE 100% ¼ 100% rangeðcÞ maxðcÞ minðcÞ
(9)
where range(c) represents the range in observed chloride concentrations calculated as the difference between the highest observed concentration, max(c), and lowest observed concentration, min(c).
3.2.
Input data
Input data for the model was comprised of: i) gauged flow and chloride concentration data of the IJssel River obtained from Rijkswaterstaat (RWS), the water management agency of the Ministry of Transport, Public Works and Water Management; ii) gauged flow and analysed chloride concentration data of the polder areas directly draining to Lakes IJsselmeer and Markermeer obtained from the regional water boards; iii) gauged rainfall and reference evaporation data obtained from the Royal Dutch Meteorological Institute (KNMI); iv) gauged water level data of the three main lakes obtained from RWS; and v) estimates on seawater intrusion from the Wadden Sea into Lake IJsselmeer due to seepage through the tidal closure dam and through direct intrusion through the shipping sluices in the Afsluitdijk dam, both obtained from RWS. Data was collected for the period 1997e2007. Flow data from the IJssel River were obtained from RWS who derived the discharge from water level data and a Qeh relationship. Flow data from the contributing polder areas were calculated by the water boards from the operating hours of the polder dewatering pumps and water level data using the pump curve (flow-head relationship). Chloride concentrations of water from the IJssel River and polder discharge (determined by argentometric titration) were supplied by RWS and the water boards respectively. In the Netherlands, the reference evaporation is calculated from incoming solar radiation and temperature using the Makkink method (Rosenberry et al., 2007). The reference evaporation is expected to be representative for short green grass vegetation. Conversion to evaporation for other land uses is done with an evaporation correction factor (ECF). In our model, the ECF is a calibration parameter.
3.3.
Climate forcing data transformation
To assess the effects of climate change, we transformed the input data for the period 1997e2007 to the reference year of 2050 using two climate change scenarios developed by the Royal Dutch Meteorological Institute (KNMI) (Van Den Hurk et al., 2006; Van Den Hurk et al., 2007). These two scenarios, named G and Wþ are the least and most far-reaching of the four climate change scenarios derived by KNMI. This means that the results of our simulations provide a bandwidth for the expected impact of climate change on chloride concentration in the Lake IJsselmeer region. The four KNMI scenarios are based on two variables: the increase in global temperature in
the year 2050 (either 1 or 2 C for the G or W scenarios respectively) and whether the air circulation patterns in Western Europe will change (Gþ, Wþ) or remain the same (G, W). We transformed all input data mentioned in Section 3.2 except the water level data. In our reference scenario, we assume no changes in current water management practice and thus no changes to the water levels in the lakes from the present situation. Table 2 presents an overview of the climate transformation of the input data by monthly relative change.
3.3.1.
Rainfall and evaporation
Rainfall data was transformed using an on-line transformation tool developed by KNMI (Bakker and Bessembinder, 2007; Bakker, 2008). This tool transforms both mean precipitation values as well as extreme precipitation values associated with the Wþ climate change scenario, such as the wet day frequency and the 10-year design storm. The G and Wþ scenarios describe relative changes in reference evaporation for each season. These values were interpolated to monthly values (Lenderink, 2006) and we further interpolated them to 36 periods of ten days using a fast Fourier transformation. We assume that the ECF remains constant under the different climate change scenarios.
3.3.2.
Water and chloride discharge of the Rhine
River works dating back to 1771 have caused a set fraction of 11% of the flow from the Rhine to be diverted to the IJssel River. We therefore assume that climate-induced changes to the flow regime are linearly transferred to the IJssel River. In the present situation, the Rhine River has a stable base flow during summer caused by a high snowmelt component (up to 80%). However, increasing atmospheric temperature is expected to change the Rhine’s hydrological behaviour from a snowmelt-dominated river to a more rain-fed river, with higher peak flows during winter and lower base flow during summer and autumn. Modelling the impact of climate change according to the Wþ scenario on the flow regime of the Rhine was carried out by Van Deursen using the RhineFLOW model and by Te Linde using the HBV model (Van Deursen, 2006; Te Linde, 2007). The results of these two modelling approaches are similar. We used the van Deursen’s results from the RhineFLOW model which delivers discharge correction factors on a ten-day interval. We applied these correction factors to the measured daily discharge. The correction factors show an increase in winter discharge of up to 20% and a decrease in summer discharge of up to 40%. The chloride concentration of the Rhine and IJssel rivers is strongly dependent on discharge. A frequently reported approach to model the chloride concentration in literature is to assume a base concentration which is constant in time and independent of flow magnitude and a constant chloride load which results in a chloride concentration depending on the flow, according to (Davis and Zobrist, 1978; Van Der Weijden and Middelburg, 1989; Zwolsman and Bokhoven, 2007): Lrhine cRhine ¼ cbase;rhine þ 1000 Q
(10)
where crhine is the chloride concentration of the Rhine [mg/l], cbase,rhine is the base chloride concentration [mg/l], Lrhine is the constant load [kg/s] and Q is river flow [m3/s]. As described
4417
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 1 1 e4 4 2 4
Fig. 3 e Results of the model calibration for Lake Markermeer and Lake IJsselmeer. The simulated chloride concentration is shown with the solid line, the measured chloride concentration is represented by the dots. Best model fit is achieved with ECF [ 1.28; N [ 22 cells and mixf [ 0.21.
earlier, the Rhine Action Plan has dramatically improved the water quality of the Rhine River and caused a decrease in both the base concentration and the chloride load. However, the future development of chloride emissions in the Rhine catchment is highly uncertain. Positive developments to reduce chloride emissions include the closing of the potash salt industry in the Alsace region and the subsequent remediation of the salt stockpile (completed in 2008). Contrarily, planned developments such as the establishment of CO2 storage caverns in rock salt formations, involving discharge of highly saline brine to the Rhine, will likely have a negative influence (Baggelaar and Van der Meulen, 2009). Given the uncertainty, and based on a review of the chloride emission and concentrations of the Rhine in the period 1997e2008, we have estimated the present-day chloride load at 60 kg/s, being the average of the period 2007e2008, and the base chloride concentration at 50 mg/l, being the average of the period 1997e2009 (Bonte and Zwolsman, 2009). In other words, we adopted a ‘stand still’ scenario for future chloride emissions in
the Rhine Catchment which can be considered as a conservative (pessimistic) approach. It can be seen from Table 2 that the transformed Rhine discharge combined with the chloridedischarge relation (Eq. 10) leads to an increase of the chloride concentration in the dry period (AugusteOctober) of 19e24% for climate scenario Wþ while under the G scenario the chloride concentration remains more or less stable. This is due to the decrease of the river flow during summereautumn in the Wþ scenario, leading to a reduction in dilution of the constant chloride load.
3.3.3.
Polder discharge
The polder areas surrounding lakes IJsselmeer and Markermeer are subject to seepage of brackish groundwater due to their lowlying nature, up to 6 m below mean sea level. In polders close to the North Sea, an increase in seepage flux can be expected due to rising sea levels. However, the larger part of the polder area draining towards lakes IJsselmeer and Markermeer is situated east of the lakes and no impact of sea level rise is expected here.
Table 3 e Summary of calibration and validation statistics for various gauging stations in lakes Markermeer and IJsselmeer Calibration
Mean observed concentration (mg/l) Mean absolute error (mg/l) Range (mg/l) Mean relative error (%) Correlation between measured and simulated (R2)
Validation
Markermeer eMiddle
IJsselmeer eAndijk
Kornwerd (west)
Vrouwezand (central)
Ketelmeer (east)
127 6 93 7% 0.88
119 12 143 8% 0.85
130 29 273 11% 0.84
110 16 226 7% 0.73
91 16 185 8% 0.83
4418
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 1 1 e4 4 2 4
The relative changes in the polder discharge flux and polder discharge chloride load were calculated using a variable density groundwater flow and solute transport model based on the MOCDENSE3D finite difference code (Delsman et al., 2009). The model was used to simulate the effects of sea level rise, changing rainfall and evaporation as well as land reclamation on the chloride concentration and drainage flux from the polders based on a yearly time step. The results in Table 2 differentiate between polder areas draining to Lake IJsselmeer and Lake Markermeer and show that the effects on the lakes under both climate scenarios are approximately 1% change compared to the current situation. This is considered to be a negligible change.
3.3.4.
Seawater intrusion
The pre-fixed water levels in lakes IJsselmeer and Markermeer are chosen to facilitate gravity discharge of water to the Wadden Sea and North Sea, to provide the surrounding polders with fresh water in summer and to maintain safe conditions during the winter storm season. As the sea level is expected to rise and the water level in the lakes is kept constant, the seepage flux and direct intrusion of sea water through the sluices (“sluice losses”) in the tidal closure dam are expected to increase. Although it is currently under debate whether or not the design water level in Lake IJsselmeer
should be raised along with rising sea levels, for the purpose of this paper we will assume that the design water level remains constant. In order to calculate the relative change in seawater intrusion, we assumed that the increase of the intrusion flux of sea water to Lake IJsselmeer is linearly related to the average head difference between sea water and lake water: DQ2050 ¼
Dh2050 100% MSL MIJL
(11)
Where DQ2050 is the relative change in seawater intrusion, Dh2050 is the expected sea level rise in the year 2050, MSL is the current mean sea level and MIJL is the mean water level in Lake IJsselmeer. The sea level rise in the year 2050 is expected to be between 0.15e0.25 m and 0.20e0.35 m compared to current MSL for the G and Wþ climate change scenarios respectively (Van Den Hurk et al., 2007). The current mean sea level at the tidal closure dam (Den Oever) is þ0.1 m-ref (calculated from sea level data at 10 min interval for 2008 obtained from RWS). The average design water level of Lake IJsselmeer is 0.3 m-ref (Table 1). This yields an increase in seawater intrusion across the tidal closure dam of 50% and 70% for climate change scenario’s G and Wþ respectively. It can be noted that this is the largest relative change in water and chloride budget terms shown in Table 2.
Fig. 4 e Validation of the transient water and chloride balance model by comparing measured and simulated chloride concentration for three chloride gauging sites in Lake IJsselmeer. The locations of the gauging sites are shown in Fig. 1.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 1 1 e4 4 2 4
4419
Fig. 5 e Contribution of the different sources of water to the simulated chloride concentration at Lake IJsselmeereAndijk and Markermeer Middle.
4.
Results & discussion
4.1.
Model calibration
Calibration parameters of the model included the number of cells and the mix factor used in Lake IJsselmeer and the Evaporation Correction Factor (ECF) used to convert reference evaporation to open water evaporation. Prior to the model calibration we performed a sensitivity analysis to quantify the effect of parameter variation on the MAE. The sensitivity analysis was comprised by a model run with a base calibration parameter set followed by a number of runs in which one of the calibration parameters was either increased or decreased by 20%. The sensitivity analysis showed that the simulated chloride concentration in Lake Markermeer depends mostly on the ECF while the chloride concentration in Lake IJsselmeer depends mostly on the mix factor and the number of cells.
Based on the relative sensitivities, we calibrated the model in two steps: First we calculated the best fit model for Lake Markermeer by increasing the ECF in a stepwise fashion, then we calculated the best fit model for both lakes with the best fit ECF and with a stepwise increase in the number of cells (N) and the mix factor from 1 to 50 and 0 to 0.35, respectively. We determined the best fit to be with ECF ¼ 1.28, N ¼ 22 and mixf ¼ 0.21. The result of the calibration is shown in Fig. 3. The calibration statistics are presented in Table 3.With this parameter set, MAE’s of 12 and 3 mg/l were calculated for the gauging sites IJsselmeereAndijk and MarkermeereMiddle, respectively. The MRE’s at these two sites are 7% at IJsselmeereAndijk and 8% at MarkermeereMiddle. This is considered adequate when compared to the uncertainty in the climate change represented by the difference between the G and Wþ scenarios. The calibrated ECF of 1.28 is slightly higher than the ECF of 1.25 in summer and 1.00 in winter used by RWS for regional
4420
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 1 1 e4 4 2 4
day and v ¼ 2700 m/day (based on an average through flow through Lake IJsselmeer of 38 106 m3/day) yields an apparent dispersion factor of DL ¼ 2.8 km2/day. This value can be compared to the dispersion factor calculated with the following relationship from (Lam and Jaquet, 1976): DL ¼ 0:153L1:3
(12)
where L is the length of the lake [km]. Substituting L ¼ 67 km in Eq. (12) gives a dispersion factor of 36.2 km2/day. Leenen derived a more physically based estimate of the dispersion factor used to simulate wind mixing processes (Leenen, 1982a): rffiffiffiffiffiffiffiffiffiffi r (13) DL ¼ Uw L cd a rw
Fig. 6 e Effects of climate change scenarios G and WD on chloride concentrations at the IJsselmeereAndijk and Markermeer Middle gauging sites in 2050. MAC chloride [ drinking water standard of 150 mg/l.
water balance calculations in the Netherlands. However, this higher value is in agreement with the findings of Van Loon and Droogers who show that the combination of a Makkink reference evaporation and an ECF of 1.25 yields lower open water evaporation rates than other more physically based calculation methods such as the De BruineKeijman method (Van Loon and Droogers, 2006). The latter method is considered to be a more appropriate way of estimating open water evaporation in moderate climates, but requires data that are not regularly collected by Dutch meteorological stations. Substituting N ¼ 22 and mixf ¼ 0.21 in Eq. (7) combined with a cell size Dx ¼ 3045 m (based on a total length of 67 km), Dt ¼ 1
where Uw is the average wind velocity [m/s], cd is the drag coefficient [dimensionless], ra and rw are the density of air and water, respectively [kg/m3]. Filling in Uw ¼ 5.4 m/s (average wind velocity at the Den Helder meteorological station over the period 1997e2008 (data KNMI)), cd ¼ 0.001 (used by Leenen), ra ¼ 1.8 kg/m3 en rw ¼ 1000 kg/m3 gives a dispersion factor of 41.9 km2/day. Both values for DL are an order of magnitude higher than the calibrated value of 2.8 km2/day. This difference is likely due to the fact that the length in both Eqs. (12) and (13) is based on the size of the lake while the magnitude of dispersion is based on the size of wind-induced eddies. For this reason, the fetch length of Lake IJsselmeer under the dominant wind direction is probably a more appropriate length for Eqs. (12) and (13). The dominant wind direction in the Netherlands is southwest, which gives an average fetch length of around 15 km. This yields dispersion factors of 5.2 and 9.4 km2/day for Eqs. (12) and (13), respectively. These values are of a similar order of magnitude as the dispersion factor found through calibration.
4.2.
Model verification
We validated the model on measured chloride time series data from three other gauging sites in Lake IJsselmeer. As Lake Markermeer is simulated with only one cell, a spatial validation was not possible there. Fig. 4 presents the results of the validation for the three chloride time series in Lake IJsselmeer (locations of the gauging sites are shown in Fig. 1). Table 3 presents the validation statistics (MAE’s and MRE’s) for these sites. The validation statistics show that although the MAE’s are higher than
Table 4 e Summary of the results of the climate change scenario modelling for Lake IJsselmeereAndijk and Lake MarkermeereMiddle Chloride concentration (mg/l)
Difference in concentration between climate change scenario and reference situation (mg/l)
Reference
Scenario G
Scenario Wþ
Scenario G
Scenario Wþ
81 105 158
79 105 177
77 121 267
3 1 19
7 16 108
MarkermeereMiddle Minimum 108 Average 119 Maximum 135
106 118 135
116 127 140
3 2 0
1 7 15
IJsselmeereAndijk Minimum Average Maximum
4421
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 1 1 e4 4 2 4
Table 5 e Calculated probability and duration of drinking water limit (MAC) exceedance at IJsselmeereAndijk and Lake MarkermeereMiddle for reference situation and climate change scenarios Daily probability of exceedance of MAC
IJsselmeereAndijk MarkermeereMiddle
Reference
Scenario G
Scenario Wþ
Reference
Scenario G
Scenario Wþ
2.5% <0.1%
3.1% <0.1%
14.3% <0.1%
103 e
124 e
178 e
calculated for the calibration sites, the MRE’s are of comparable magnitude except for the Kornwerd gauging location. If we compare the general trend between the different gauging sites we see an increasing chloride gradient from upstream locations (Ketelmeer), to central and downstream locations in Lake IJsselmeer (Vrouwezand and Kornwerd). At the Kornwerd gauging site, the simulated chloride concentration is structurally overestimated. This is likely due to the fact that water samples are collected 1 m below the water surface and thus may not be representative for the depth averaged chloride concentration as simulated by the model. This error can be particularly pronounced near the tidal closure dam, where vertical chloride gradients can be expected due to salt water seepage.
4.3.
Maximum duration of MAC exceedence (days)
Sources of chloride to the lakes
To gain an understanding of the salinisation processes occurring in Lake IJsselmeer and Lake Markermeer, the calibrated model was used to determine the contribution of the different water sources to the chloride concentration at Andijk water intake station (Lake IJsselmeer) and at Markermeer Middle. This was done by first running the calibrated model only including water and chloride input from the IJssel
River, followed by a number of model runs where one source of water and chloride is consecutively added to the model. The results of this analysis are presented in Fig. 5 which shows the stacked contribution to the chloride concentration at the two monitoring sites over a ten-year period (1997e2008). The simulation shows that in both lakes the IJssel River is the main driver of the chloride concentration. In Lake Markermeer, discharges from the surrounding polder areas increase the chloride concentration while the net rainfall (rainfall minus evaporation) and inflow from lake Gooi & Eem have a decreasing effect on the chloride concentration. This clearly reflects the differences in chloride concentrations of the individual water sources to the lake. In Lake IJsselmeer (Andijk), the main sources of chloride besides the IJssel (or Rhine) River are seawater intrusion across the tidal closure dam via both seepage and direct intrusion through the shipping sluices and drainage from the Flevoland polder. In the relatively dry summer and fall of 2003, seawater intrusion across the tidal closure dam was simulated to have a relatively large contribution of around 50 mg/l to the chloride concentration at station Andijk. This input of chloride from the Wadden Sea, together with the high chloride concentrations in the Rhine at that time, caused the chloride concentration at Andijk to exceed the drinking water standard
Fig. 7 e Contribution of the different sources of chloride to the changing chloride concentration at Lake IJsselmeereAndijk and Markermeer Middle for climate change scenarios G and WD in the year 2050.
4422
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 1 1 e4 4 2 4
of 150 mg/l in the fall of 2003. The simulation also shows that in all other simulated years, the contribution of seawater intrusion to the chloride concentration at Andijk is far less. This difference is probably due to higher discharge from Lake IJsselmeer to the Wadden Sea during non-dry years, which flushes out the seawater introduced at the tidal closure dam. It is interesting to note that in Lake IJsselmeer, the effect of rainfall and evaporation varies with the season while in Lake Markermeer this budget term always has a freshening effect. This difference is due to the difference in residence time of water between the two lakes: 3e4 months in Lake IJsselmeer and 15e18 months in Lake Markermeer.
4.4.
Impact of climate change on salinisation processes
The results of the climate change simulations are compared to the reference scenario based on present climate and hydrology and average salt emissions for 2007e2008. The calculated chloride concentrations in the reference situation are lower than the concentrations actually measured for the period 1997e2007 because of the strongly reduced chloride emissions during this period. Fig. 6 presents the simulated chloride concentrations at the gauging sites IJsselmeereAndijk and Markermeer Middle in the year 2050 for the two climate change scenarios (G and Wþ) and the reference scenario. A summary of the changes in chloride concentrations at the two gauging sites for the two climate change scenarios is presented in Table 4. This table presents minimum, average and maximum simulated chloride concentrations at the two gauging sites for the two climate scenarios, as well as the minimum, average and maximum differences between the scenario models and the reference model. The results clearly show that the largest effects occur at the Andijk gauging site for climate scenario Wþ. The chloride concentration at Andijk under scenario Wþ shows an average increase of 16 mg/l (compared to the reference situation), and a maximum increase of up to 108 mg/l under dry conditions (year 2003, transformed to 2050). The effects of climate change on Lake Markermeer are relatively small: the average and maximum chloride concentration increase in scenario Wþ amount to 7 and 15 mg/l, respectively. For scenario G, concentrations in Lake IJsselmeer increase slightly on average while for Lake Markermeer the chloride concentrations show no difference from the reference scenario. Table 5 presents data on the probability and duration that the drinking water standard (150 mg/l) for chloride is exceeded. The daily probability of exceeding the chloride standard for drinking water in Lake IJsselmeer (Andijk) increases from 2.5% for the reference scenario to 3.1% and 14.3% for climate change scenarios G and Wþ respectively. The maximum duration of the exceedance increases from 103 days in the reference scenario to 178 days in the Wþ scenario. Similar to the chloride source analyses presented in Fig. 5 for the present situation, Fig. 7 presents the stacked changes in chloride concentration due to: i) changing hydrology of the Rhine; ii) changing rainfall and precipitation patterns; iii) changing intrusion from the Wadden Sea; and iv) changing salt input from polders directly draining to Lake IJsselmeer. From this figure it can be seen that a large fraction of the chloride change in Lake IJsselmeer is due to the increased seepage of salt water from the Wadden Sea and the changing
hydrology of the Rhine. Rising sea levels (and the associated increase in head difference between Wadden Sea and Lake IJsselmeer) cause more chloride to enter Lake IJsselmeer across the tidal closure dam. The modelling suggests that the reduction of through flow in Lake IJsselmeer during dry summers as a result of reduced Rhine flows causes this brackish water to be transferred further upstream in Lake IJsselmeer. Moreover, the reduced summer flows also cause less dilution of the chloride load of the Rhine River, resulting in a higher chloride concentration in the river water which propagates in the IJssel branch and Lake IJsselmeer. Fig. 7 shows that the small increase in chloride concentration in Lake Markermeer is due to increasing concentrations in the Rhine as well as increasing summer evaporation and decreasing rainfall. As previously mentioned, chloride concentrations in Lake Markermeer are far less influenced by climate change. This is for a large part due to the timing and direction of water transfers between Lake Markermeer and Lake IJsselmeer. In general, water is transferred from Lake IJsselmeer to Lake Markermeer in spring and early summer when Rhine flow is high and the chloride concentration of Lake IJsselmeer is lower than that of Lake Markermeer. The largest increase in chloride concentrations in Lake IJsselmeer occurs in autumn and early winter when the river flow is low and, during which, water is transferred from Lake Markermeer to Lake IJsselmeer. The results of the climate change scenario simulations provide important information that can be used to optimise water management efforts to keep both Lake IJsselmeer and Lake Markermeer sufficiently fresh for drinking water production. The results show that in order to keep Lake IJsselmeer sufficiently fresh for year round drinking water production (limit 150 mg/l chloride), a further reduction of salt emissions to the Rhine or a reduction of seepage from the Wadden Sea through technical engineering measures is required. The modelling shows that Lake Markermeer is actually a more suitable source for drinking water production than Lake IJsselmeer as it is less susceptible to climate change induced salinisation. The possibility to control water transfers from Lake IJsselmeer to Lake Markermeer also makes it a better raw water source from the perspective of other pollutants present in river water, especially during chemical spills and other calamities on the Rhine River. As a result, Lake Markermeer can be considered a water conservation reservoir for the Lake IJsselmeer region.
5.
Conclusions
1. The transient chloride concentration of lakes IJsselmeer and Markermeer can be adequately simulated with a simple, compartmental cell water and chloride balance model using a single cell for Lake Markermeer and a compartmental cell flow tube with a diffusion term representing wind-forced mixing for Lake IJsselmeer. 2. The model can be used as a tool to elucidate the salinisation processes occurring in the Lake IJsselmeer region and to target management strategies aiming to reduce salinisation. The model was used to determine the contribution of different water sources to the chloride concentration at different gauging sites. The model results showed that the Rhine River is the dominant determinant for the chloride
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 1 1 e4 4 2 4
concentration in both lakes, followed by drainage from the surrounding polders. The results further show that especially during dry years, seawater intrusion through the tidal closure dam is an important source of chloride to Lake IJsselmeer. It is noted that this transient modelling approach provides a clear advantage over a steady state or yearlyaveraged water and chloride balance as it was shown that the relative impact of the different salinisation processes varies considerably during the year. 3. Modelling of climatic forcing of the water and chloride balance for Lakes Markermeer and IJsselmeer shows that the latter is especially vulnerable to a climate-induced increase in salinisation. Modelling suggests that peak chloride concentrations at the Andijk gauging site on Lake IJsselmeer can increase by up to 108 mg/l in a dry year based on the climate change scenario Wþ in 2050. This implies serious consequences for drinking water production. The dominant drivers for the increased chloride concentrations are the changing hydrology of the Rhine River, which is expected to change from a snowmelt to rainfall dominated river, and the rising sea level, which is expected to increase salt water intrusion across the tidal closure dam. 4. The processes described here are likely to affect many other tidal rivers or lakes and should be considered when planning future raw water intake stations. Climate change is shown here to impact water quality (i.e., chloride) in a number of ways: 1) reduced river flows during summer and autumn impact the dilution of chemical substances that enter the river at a constant load, thereby increasing their concentration, 2) in estuarine environments, salt water intrusion can be expected to reach further upstream due to both rising sea levels and reducing river flows, and 3) increased open water evaporation and reduced rainfall during summer periods and droughts will further increase concentrations of chloride and other substances in the water.
Acknowledgements This research was financially supported by the Joint Research Programme of the Dutch drinking water sector (BTO), the European Commission through the FP6 Water and Global Change (WATCH) project and the Delft Cluster project. We would like to thank the following organisations for supplying data: Rijkswaterstaat, RIWA Rijn and the water boards Hoogheemraadschap Hollands Noorderkwartier, Amstel Gooi en Vecht, Zuiderzeeland, Vallei en Eem and Fryslaˆn. We thank Nick Gorski for proof reading the manuscript and Kees Maas for his help with the mathematical equations.
references
Appelo, C.A.J., Postma, D., 2005. Geochemistry, groundwater and pollution, second ed. A.A. Balkema, Leiden, ISBN 04 1536 428 0, 649 pp. Arheimer, B., Andre´asson, J., Fogelberg, S., Johnsson, H., Pers, C.B., Persson, K., 2005. Climate change impact on water quality: model results from southern Sweden. Ambio 34 (7), 7.
4423
Baggelaar, P.K., Van der Meulen, E.C.J., 2009. Historical and future developments in the chloride load from Lobith to Andijk. RIWA Rijn, Nieuwegein (in dutch)111 pp. Bakker, A., 2008. Transformation of observed time series for climate change impact assessment, In: Proc. EGU2008, European Geosciences Union General Assembly 2008, Vienna. Bakker, A., Bessembinder, J., 2007. Precipitation time series for the KNMI’06 climate change scenario. H2O 40 (22), 45e47 (in Dutch). Bonte, M., Zwolsman, J.J.G., 2009. Climate change and freshening of the Rhine. H2O 20, 29e31 (in Dutch). Cugier, P., Billen, G., Guillaud, J.F., Garnier, J., Me´nesguen, A., 2005. Modelling the eutrophication of the Seine Bight (France) under historical, present and future riverine nutrient loading. Journal of Hydrology 304 (1e4), 381e396. Davis, J., Zobrist, J., 1978. The Interrelationships among chemical parameters in rivers e analysing the effect of natural and anthropogenic sources. Progress in Water Technology 10 (5/6), 13. Delpla, I., Jung, A.-V., Baures, E., Thomas, M.C.a.O, 2009. Impacts of climate change on surface water quality in relation to drinking water production. Environment International 35 (8), 1225e1233. Delsman, J., Borren, W., Essink, G.O., Stuurman, R., Verkaik, J., 2009. Changes in the Groundwater Flow System of Lake Markermeer. Deltares, Utrecht (in Dutch) 91 pp. Dieperink, C., 2000. Successes in the international cooperation in the Rhine catchment area. Physics and chemistry of the earth, part b: hydrology. Oceans and Atmosphere 25 (3), 341e347. Evans, C.D., 2005. Modelling the effects of climate change on an acidic upland stream. Biogeochemistry 74 (1), 21e46. Jonkeren, O., Jourquin, B. and Rietveld, P. Modal-split effects of climate change: The effect of low water levels on the competitive position of inland waterway transport in the river Rhine area. Transportation Research Part A: Policy and Practice, in press. Kaushal, S.S., Gene, E.L., 2009. Chloride. Encyclopedia of Inland Waters. Academic Press, Oxford, ISBN 978-0-12-370626-3, pp. 23e29. Lam, D.C.L., Jaquet, J.M., 1976. Computation of Physical transport and regeneration of phosphorus in Lake Erie, fall 1970. Journal of the Fisheries Research Board of Canada 33 (3), 550e563. Leander, R., Buishand, T.A., van den Hurk, B.J.J.M., de Wit, M.J.M., 2008. Estimated changes in flood quantiles of the river Meuse from resampling of regional climate model output. Journal of Hydrology 351 (3e4), 331e343. Leenen, J., 1982a. Wind induced diffusion in a shallow lake, a case study. Aquatic Ecology 16 (2), 231e240. Leenen, J.D., 1982b. Hydrology of Tjeukemeer. Hydrobiologia 95 (1), 199e203. Lenderink, G., 2006. KNMI ’06 Scenario’s: Interpolation to Monthly Changes. Royal Netherlands Meteorological Institute, De Bilt, 9 pp. Loos, S., Middelkoop, H., van der Perk, M., van Beek, R., 2009. Large scale nutrient modelling using globally available datasets: a test for the Rhine basin. Journal of Hydrology 369 (3e4), 403e415. Middelkoop, H., Daamen, K., Gellens, D., Grabs, W., Kwadijk, J.C.J., Lang, H., Parmet, B.W.A.H., Scha¨dler, B., Schulla, J., Wilke, K., 2001. Impact of climate change on hydrological regimes and water resources management in the Rhine basin. Climatic Change 49 (1), 105e128. Middelkoop, H., Kwadijk, J.C.J., 2001. Towards integrated assessment of the implications of global change for water management e the Rhine experience. Physics and chemistry of the earth, part b: hydrology. Oceans and Atmosphere 26 (7e8), 553e560. Molt, E.L., 1961. Pollution of Rhine Water. Moormans Periodieke Pers N.V., Den Haag, De Rijn (in Dutch)pp. 46e71.
4424
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 1 1 e4 4 2 4
Parmet, B., Kwadijk, J., Raak, M., Zwerver, S., Rompaey, R.S.A.R. van, Kok, M.T.J., Berk, M.M., 1995. Impact of climate change on the discharge of the river Rhine. Studies in Environmental Science, pp. 911e918, Elsevier ISSN 0166-1116. Rosenberry, D.O., Winter, T.C., Buso, D.C., Likens, G.E., 2007. Comparison of 15 evaporation methods applied to a small mountain lake in the northeastern USA. Journal of Hydrology 340 (3e4), 149e166. Rozema, J., Flowers, T., 2008. ECOLOGY: crops for a Salinized World. Science 322 (5907), 1478e1480. Schauser, I., Chorus, I., 2009. Water and phosphorus mass balance of Lake Tegel and Schlachtensee e a modelling approach. Water Research 43 (6), 1788e1800. Senhorst, H.A.J., Zwolsman, J.J.G., 2005. Climate change and effects on water quality: a first impression. Water Science and Technology 50 (5), 6. Shabalova, M.V., Deursen, W.P.A.van., Buishand, T.A., 2003. Assessing future discharge of the river Rhine using regional climate model integrations and a hydrological model. Climate Research 23 (3), 233e246. Syncera Water, 2006. Beheersverslag Rijkswateren IJsselmeergebied 2002e2004. IJ Rapport 2006-4. ISBN: 90 369 1348 9, 187 pp. Te Linde, A., 2007. Effect of climate change on the rivers Rhine and Meuse. Applying the KNMI 2006 scenarios using the HBV model. Project nr. Q4286. WL, Delft Hydraulics, Delft, 44 pp.
Van Den Hurk, B., Tank, A.K., Lenderink, G., Ulden, A. van, Oldenborgh, G.J. van, Katsman, C., Brink, H. van den., Keller, F., Bessembinder, J., Burgers, G., Komen, G., Hazeleger, W., Drijfhout, S., 2007. New climate change scenarios for the Netherlands. Water Science & Technology 56 (4), 27e33. Van Den Hurk, B., Tank, A.K., Lenderink, G., Ulden, A.van, Oldenborgh, G.J.van, Katsman, C., Brink, H. van den., Keller, F., Bessembinder, J., Burgers, G., Komen, G., Hazeleger, W., Drijfhout, S., 2006. KNMI Climate Change Scenarios 2006 for the Netherlands. Royal Netherlands Meteorological Institute, De Bilt, 82 pp. Van Der Weijden, C.H., Middelburg, J.J., 1989. Hydrogeochemistry of the River Rhine: long term and seasonal variability, elemental budgets, base levels and pollution. Water Research 23 (10), 1247e1266. Van Deursen, W., 2006. Rapportage Rhineflow/Meuseflow Nieuwe KNMI scenario’s 2050. Carthago Consultancy, Rotterdam, 7 p. Van Loon, A., Droogers, P., 2006. Calculating the open water evaporation. Report for the Waterboard Fryslan. Future Water, Wageningen (in Dutch), 24 pp. van Vliet, M.T.H., Zwolsman, J.J.G., 2008. Impact of summer droughts on the water quality of the Meuse river. Journal of Hydrology 353 (1e2), 1e17. Zwolsman, J.J.G., Bokhoven, A.J.v, 2007. Impact of summer droughts on water quality of the Rhine River - a preview of climate change? Water Science & Technology 56 (4), 10.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 9 e4 3 9 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Compositional changes in the hydrophobic acids fraction of drainage water from different land management practices Corinna M.P. Byrne a, Michael H.B. Hayes a,*, Rajeev Kumar b, Etelvino H. Novotny c, Gary Lanigan d, Karl G. Richards d, Deirdre Fay d, Andre J. Simpson b,** a
Department of Chemical and Environmental Sciences, University of Limerick (UL), Ireland Department of Chemistry, University of Toronto, Scarborough, Toronto, Ontario, M1C 1A4, Canada c UL and Embrapa Solos, R. Jardim Botaˆnico, 1024, CEP 22460-000, Rio de Janeiro-RJ, Brazil d Teagasc Environmental Research Centre, Johnston Castle, Wexford, Ireland b
article info
abstract
Article history:
Dissolved organic matter (DOM) can play a key role in many environmental processes,
Received 9 February 2010
including carbon cycling, nutrient transport and the fates of contaminants and of agro-
Received in revised form
chemicals. Hydrophobic acids (Ho), the major components of the DOM, were recovered
27 May 2010
from the drainage waters from well-drained (WDS) and poorly-drained (PDS) Irish grass-
Accepted 31 May 2010
land soils in lysimeters, amended with N fertiliser (F) and with bovine urine (U) and were
Available online 9 June 2010
studied using 1D and 2D solution-state Nuclear Magnetic Resonance (NMR) spectroscopy. The Diffusion Edited (DE) 1H NMR spectra indicated that the Ho consisted largely of larger
Keywords:
molecules, or of molecules that formed rigid aggregates, and the 1D and the 2D (Hetero-
Grassland
nuclear Multiple Quantum Coherence e HMQC, the Total Correlation Spectroscopy e
Dissolved organic matter
TOCSY, and the Nuclear Overhauser Effect e NOESY) spectra indicated that the samples
Hydrophobic acids
were composed of lignin residues, carbohydrates, protein/peptides, and aliphatic compo-
Drainage water
nents derived from plant waxes/cuticular materials and from microbial lipids. The F
Fertiliser
amendments increased the concentrations of Ho in the waters by 1.5 and 2.5 times those in
Urine
the controls in the cases of WDS and PDS, respectively. The lignin-derived components
Solution-state NMR
were increased by 50% and 300% in the cases of the Ho from the WDS and PDS, respectively. Applications of F þ U decreased the losses of Ho, (compared to the F amendments alone) and very significantly decreased those of the lignin-derived materials, indicating that enhanced microbial activity from U gave rise to enhanced metabolism of the Ho components, and especially of lignin. In contrast the less biodegradable aliphatic components containing cuticular materials increased as the result of applications of F þ U. This study helps our understanding of how management practices influence the movement of C between terrestrial and aquatic environments. ª 2010 Elsevier Ltd. All rights reserved.
* Corresponding author. Tel.: þ353 61 202631; fax: þ353 61 202572. ** Corresponding author. Tel.: þ1 416 287 7547; fax: þ1 416 287 7279. E-mail addresses:
[email protected] (M.H.B. Hayes),
[email protected] (A.J. Simpson). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.05.055
4380
1.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 9 e4 3 9 0
Introduction
Dissolved organic matter (DOM) is a complex, heterogeneous mixture found in all natural waters, and it represents the largest fraction of mobile carbon (C) on earth. It provides an intimate link between the terrestrial and aquatic environments (Lam et al., 2007). Soil derived DOM can play a key role in many environmental processes, including carbon cycling, nutrient transport and the fates of contaminants and of agrochemicals (Qualls and Haines, 1991; Royer et al., 2007; Zsolnay, 2003). Despite its obvious importance, the structural components of soil DOM and the variations of these components with different land management practices have not been well resolved (Royer et al., 2007). Temperate grassland ecosystems, which comprise 32% of the earth’s natural vegetation (Frank and Dugas, 2001), can be considered to have a significant role in the uptake of atmospheric CO2 and in balancing the global C budget (Batjes, 1998). Grassland, the dominant ecosystem in Ireland, represents 90% of agricultural land and 56% of the total land area (Jaksic et al., 2006). Article 3.4 of the Kyoto protocol (UNFCCC, 1998) makes provision for the use of soil C stock changes in grazing lands to offset greenhouse gas (GHG) emissions and to facilitate the achievement of emissions reduction targets (Byrne et al., 2007). On that basis, there is a need to better understand the organic components leached from these carbon stocks under different management practices. Soils in long-term pasture are in a steady-state with regard to soil organic matter (SOM) content. Carbon accumulation in grassland ecosystems occurs mostly below ground and changes in soil organic C (SOC) stocks may result from changes in land uses management (Soussana et al., 2004). Grassland C stocks represent at least 10% of the global total, and some sources suggest up to 30% of that total (Scurlock and
Hall, 1998). The stocks of SOM result from the balance between inputs and outputs of C. Inputs are primarily from leaf and root detritus. Outputs are dominated by the efflux of carbon dioxide (CO2) and of methane (CH4) from the soil surface and by the hydrologic leaching of dissolved and particulate C (Davidson and Janssens, 2006). The pool of SOM is of particular interest because even small changes in flux rates into or out of such a large pool could lead to the accumulation of significant quantities of greenhouse gases (Billings and Ziegler, 2008). Although land use and related management practices are known to affect the amounts and compositions of SOM and soil properties, their influences on the amounts and compositions of DOM have not been extensively studied (Chantigny, 2003). Various aspects of the effects of elevated nitrogen (N) deposition and of N fertilization have been studied, yet little is known about their effects on DOM turnover (Kalbitz et al., 2000). The same is true for organic amendments such as urine. The OM in amendments is biodegradable and is generally readily transformed by soil microbes. That may result in transient increases in the soil DOM (Chantigny, 2003). Amendment with slurry has been found to increase nitrogen (N) immobilisation through increased microbial activity (Hoekstra et al., 2009). This may lead to an increase in carbon mineralization and a decrease in DOM export. However, to our knowledge, detailed studies have not been carried out on changes in DOM compositions following mineral fertilization and organic amendments. Soil hydrology is also likely to affect DOM dynamics. Differences have been found between DOM fractions isolated from different drainage regimes (Hayes et al., 1997), and research has shown that DOC exports were 33 Kg/ha lower from drained than from undrained plots (McTiernan et al., 2001). In this study we characterise, in detail, the Hydrophobic acids (Ho) from the DOM formed from two soils, one well-
Table 1 e Analyses of the well-drained and of poorly-drained soils. Soil
Depth (cm)
Total C
Organic C
Total N
C\N ratio
% Sand
% Silt
% Clay
WD
0e10 10e20 20e30 30e40 40e50 50e60 60e70 70e80 80e90 90e100
3.22 2.52 1.42 1.59 1.19 0.69 0.17 0.32 0.22 0.19
3.18 2.33 1.43 1.5 1.12 0.66 0.14 0.27 0.19 0.15
0.3 0.24 0.14 0.13 0.08 0.05 0.02 0.03 0.02 0.02
10.7 10.5 10.1 12.2 14.9 13.8 8.5 10.7 11.0 9.5
45.2 44.0 48.6 41.4 40.0 42.8 46.6 46.7 50.2 42.6
20.4 27.9 28.0 33.1 42.5 43.2 35.7 32.4 36.4 44.7
12.0 12.3 12.4 14.3 9.2 7.5 6.5 6.1 5.9 1.8
PD
0e10 10e20 20e30 30e40 40e50 50e60 60e70 70e80 80e90 90e100
4.36 2.72 0.83 0.34 0.25 0.22 0.14 0.14 0.12 0.11
4.23 2.7 0.77 0.29 0.22 0.18 0.11 0.11 0.09 0.08
0.35 0.25 0.09 0.04 0.03 0.03 0.03 0.03 0.03 0.03
12.5 10.9 9.2 8.5 8.3 7.3 4.7 4.7 4.0 3.7
24.8 25.6 27.2 30.1 31.0 31.1 34.6 34.2 35.8 33.8
35.0 35.2 34.7 17.8 34.6 34.2 34.5 35.3 34.4 36.6
25.2 26.3 30.0 45.6 28.7 16.1 25.3 24.0 23.9 24.3
4381
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 9 e4 3 9 0
drained (WDS) and the second poorly-drained (PDS), each amended with fertiliser, and with fertiliser and urine. The emphasis is on the characterisation of the components of the Ho released in the drainage water from these soils using advanced Multidimensional nuclear magnetic resonance spectroscopy (NMR) techniques that are widely used to study structures and interactions in environmental chemistry (Simpson and Brown, 2005; Thrippleton and Keeler, 2003).
2.
Materials and methods
2.1.
Source of samples
Intact soil monoliths lysimeters (0.8 m diameter by 1 m deep) were sampled from a well-drained (WDS) Brown Podzolic soil (Haplic Podzol (Anthric)) (FAO, 2007) and from a poorlydrained (PDS) Gley (Luvic Stagnosol (Eutric, Siltic)) (FAO, 2007) were installed in 2004 in lysimeters in a pasture field at the Teagasc Environmental Research Centre (ERC), Johnstown Castle, Wexford, Ireland. The sand, silt, and clay contents of the soils are given in Table 1. The soils were collected as undisturbed monoliths and installed according to an established protocol (Cameron et al., 1992). Briefly this involved isolating a 1 m by 1 m soil column and then carefully, reciprocally, pushing a 0.8-m HDPE pipe through the soil column. When the pipe reached 1 m a cutting plate was hydraulically pushed beneath the lysimeter to cut it from the soil beneath. To prevent edge flow liquid petrolatum was injected between the soil and the HDPE pipe. The lysimeters were inverted and 5 cm of fine gravel inserted at the base of the soil and a base plate with drainage outlet was welded to the pipe. The completed lysimeters were installed in a field lysimeter facility under natural rainfall and meteorological conditions. Each soil was sown with perennial ryegrass (Lolium perennae L.). In order to replicate typical Irish grazed grassland activities, some of the lysimeter soils were amended with fertiliser and some with both fertiliser and bovine urine, and unamended soils served as controls as described in Stark et al. (2007). With the exception of the controls, the lysimeter soils received in 2004 and 2005, 291 kg N ha1 yr1 as fertiliser and 310 kg N ha1 yr1 as urine (Table 2). Treatments were applied in a randomised complete block design with 3 replicates per treatment. Herbage was harvested regularly to correspond with a 28-day rotation of livestock. A series of pipes transported the drainage water (DW) from each lysimeter to storage vessels housed below ground level. Drainage water samples, 200 L from each treatment and control, were collected from the lysimeter facility between June and December, 2005.
the entire sample had passed through the column. The resin was then desalted with distilled water until effluent conductivities were <100 mS cm1. Back elution was carried out using 0.1 M NaOH and the centre cut eluates were Hþ exchanged (Amberlite IR-120, Hþ-form; Rohm and Haas, Philadelphia), and then freeze dried to give the XAD-8 hydrophobic (Ho) acids.
2.3. Solution-state NMR spectroscopy experimental details Samples (40 mg) were dissolved in 600 mL of deuterium oxide (D2O) and titrated to pH 12 using NaOD to ensure complete solubility. Additional samples (40 mg) were dissolved in 600 mL DMSO-d6. Samples were analysed using a Bruker Avance 500 MHz NMR spectrometer equipped with a 1He19Fe15Ne13C 5 mm, quadruple resonance inverse probe with actively shielded zgradient (QXI). 1D solution-state 1H NMR spectra were obtained with 128 scans, a recycle delay of 2 s, 16 384 time domain points, and an acquisition time of 0.79 s. Water suppression was achieved using PURGE (Simpson and Brown, 2005). Spectra were apodized through multiplication with an exponential decay corresponding to 1 Hz line broadening, and a zero-filling factor of 2. Diffusion edited (DE) spectra were obtained using a bipolar pulse longitudinal encodeeencode sequence. Scans (1600) were collected using a 2.5 ms, 49 G/cm, sine-shaped gradient pulse, a diffusion time of 200 ms, 16 384 time domain points, 0.82 s acquisition time, and a sample temperature of 298 K. Heteronuclear multiple quantum coherence (HMQC) spectra were obtained in phase-sensitive mode using echo/ anti-echo gradient selection and a 1J 1He13C value of 145 Hz. Scans (512) were collected for each of the 128 increments in the F1 dimension. A total of 1048 data points were collected in F2, and a relaxation delay of 1 s was employed. The F2 dimension was multiplied by an exponential function corresponding to a 10 Hz line broadening and a zero-filling factor of 2. The F1 dimension was processed using a sine-squared function with a p/2 phase shift and a zero-filling factor of 2. Total correlation spectroscopy (TOCSY) spectra were acquired in the phase-sensitive mode, using time proportional phase incrimination (TPPI). TOCSY NMR experiments were carried out using 512 scans with 128 time domain points in the
Table 2 e Nutrient application rates to lysimeters. Treatment
Nutrient application rates kg/ha Inorganic Fertiliser
2.2.
Isolation of hydrophobic acids from drainage waters
The Ho were isolated from the drainage waters using previously described procedures (Hayes et al., 2008; Malcolm and MacCarthy, 1992). Waters were filtered under pressure (69 kPa) through 0.2 mm Sartorius (Goettingen, Germany) cellulose acetate membrane filters. The filtrates were adjusted to pH 2 (HCl) and applied to XAD-8 resin [(poly)methylmethacrylate] (Rohm and Haas, Philadelphia). Two column volumes of 0.01 M HCl were pumped through to ensure that
a
Control Fertiliser only Fertiliser & urine
b
Cow Urine
Urea
CAN
N
P
K
0 58 58
0 233 233
0 0 310
0 0 0.8
0 0 465
2004 At grass sowing all lysimeters received a basal application of NPK of 37, 37 and 74 kg/ha, respectively. a Urea (46% N) manufactured by Goulding. b CAN- Calcium Ammonium Nitrate (27% N) manufactured by Goulding.
4382
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 9 e4 3 9 0
F1 dimension and 1048 time domain points in the F2 dimension. A mixing time of 60 ms was used with a relaxation delay of 1 s. Processing of both dimensions used a sine-squared function with a p/2 phase shift and a zero-filling factor of 2. Nuclear Overhauser Effect Spectroscopy (NOESY) was obtained with the elimination of zero-quantum interference (Thrippleton and Keeler, 2003). NOESY NMR experiments were carried out using 256 scans with 128 time domain points in the F1 dimension and 1048 time domain points in the F2 dimension. A mixing time of 250 ms was used with a relaxation delay of 1 s. Zero-quantum suppression was achieved through the use of an adiabatic-pulse/gradient pair during the mixing time (Thrippleton and Keeler, 2003). Both dimensions were processed using a sine-squared function with a p/2 phase shift and a zero-filling factor of 2.
3.
Results and discussion
DOM in soil is composed of humic substances and a variety of specific identifiable organic compounds, including carbohydrates and peptides. In this study the hydrophobic acid fraction was isolated using an XAD-8 resin technique (Leenheer, 1981), and is the dominating constituent of bulk dissolved organic matter (DOM) in soil solutions (Asakawa et al., 2006).
3.1. acids
Characterisation of the drainage water hydrophobic
Two solvent systems were used for the NMR analysis of the Ho; D2O/NaOD and DMSO-d6. D2O or D2O/NaOD systems are commonly used for studies of DOM (Hertkorn et al., 2006; Kaiser et al., 2003; Kim et al., 2003; Lam et al., 2007; Simpson, 2001; Smejkalova and Piccolo, 2008) and the D2O/NaOD system in this study enabled comparisons with previous studies. DOM samples in the protonated form (achieved here through exchange with the IR-120 cation exchange resin) are completely soluble in DMSO. DMSO is a dipolar aprotic solvent; hence signals from exchangeable protons, for example, NeH, can be observed. Thus DMSO provides excellent complimentary information for structural studies, especially for protein/peptide components, and in many cases it provides spectra with better defined resonances (Simpson, 2001). Our samples were completely soluble in both solvent systems used. 1D and 2D NMR spectroscopy techniques were used to observe compositional differences in the Ho components in the drainage waters. Fig. 1A shows the 1H NMR spectrum in DMSO-d6 for the Ho isolated from the poorly-drained soil (PDS) treated with fertiliser. Major structural components present include aromatics, lignin (Lig), carbohydrates (Carb), proteins/
Aliphatic DMSO
Aromatic/amide
Anomeric protons in Carb
Lig OCH3/ Carb
WC/L
A P
P P
P
P
Lig
B SC N-H
Phe
CH3
-protons
C Methoxyl Aromatic
D Fig. 1 e (A), 1H NMR spectrum in DMSO-d6 for Ho isolated from the PDS treated with Fertiliser. (B), Diffusion edited 1H NMR spectrum in DMSO-d6 for the Ho. (C), Cultured soil microbes. (D), Organosolv Lignin. Assignments include lignin (Lig), carbohydrates (Carb), protein/peptides (P), waxes, cuticles and lipids (WC/L), protein/peptide side chains (SC), phenylalanine (Phe) and amide (NeH).
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 9 e4 3 9 0
4383
Fig. 2 e Various 2D NMR spectra of the Ho isolated from the PDS treated with Fertiliser. (A), HMQC Spectrum, main assignments can be summarised as 1, p-hydroxybenzoate aromatics in lignin (Kelleher and Simpson, 2006; Simpson et al., 2004); 2, phenylalanine in peptides (Kelleher and Simpson, 2006; Simpson et al., 2007a); 3, aromatic lignin units (Kelleher et al., 2006; Simpson et al., 2004); 4, anomeric protons in carbohydrates (Kelleher and Simpson, 2006; Lam et al., 2007); 5, methine in carbohydrates (Kelleher and Simpson, 2006; Lam et al., 2007); 6, methylene units in carbohydrates (Kelleher and Simpson, 2006; Lam et al., 2007); 7, a-protons in peptides and proteins (Kelleher and Simpson, 2006; Simpson et al., 2007a; Simpson et al., 2007b); 8, methoxyl in lignin (Kelleher and Simpson, 2006; Simpson et al., 2003, 2004); 9, aliphatic linkages including signals from various lipids and plant cuticles (Deshmukh et al., 2003, 2005; Simpson et al., 2003, 2007b), and side chain protons in peptides (Kelleher and Simpson, 2006; Simpson et al., 2007a); 10, N-acetyl and/or O-acetyl carbohydrates (Hertkorn et al., 2006; Lam et al., 2007); 11, methylene units in aliphatic chains (Kelleher et al., 2006; Simpson et al., 2001, 2003); 12, methyl groups, a small contribution in this region will be from terminal CH3 in lipids, though the majority of signals are from peptides (Kelleher et al., 2006; Simpson et al., 2003, 2007a). (B), is an expanded region of the HMQC. The intense lignin methoxyl signal is clearly evident in region 8. (C), is the TOCSY spectrum which supports assignments made from the 1D and HMQC spectra. Key assignments: aromatic couplings (Kelleher and Simpson, 2006; Simpson et al., 2004); Pamide [ amideLabg couplings in peptides (Kelleher and Simpson, 2006; Kingery et al., 2000; Simpson et al., 2007a,b); Pa, a-protons coupling to amino acid side chains (Kelleher and Simpson, 2006; Kingery et al., 2000; Simpson et al., 2007a,b); couplings in carbohydrates (Carb) and aliphatic couplings (Deshmukh et al., 2003, 2005; Kelleher et al., 2006; Simpson et al., 2003). (D), is the NOESY spectrum that confirms the strong contribution of P, peptides/proteins with crosspeaks from a-protons in amino acid side chains. The most important assignment is the through space interaction between aromatic rings and methoxyl groups indicative of lignin (Lig) (Simpson, 2001).
peptides (P) and aliphatic units. Fig. 1B is the diffusion edited (DE) NMR spectrum of the same sample. Signals from larger molecules or rigid molecular associations can be further emphasised by the use of diffusion editing. Diffusion editing “spatially encodes” molecules at the start and then “refocuses” these at the end of the experiment. Species that diffuse
and exhibit a high degree of motion during the experiment are not refocused and are essentially gated from the final spectrum (Simpson et al., 2007b). Thus the spectrum produced contains only signals from larger molecules or rigid molecular associations. Because the majority of the signals remain after diffusion editing, it can be considered that the components in
4384
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 9 e4 3 9 0
the Ho are likely to be larger molecules or very stable aggregates (Simpson, 2002). Main chain methylene signals at e1.3 ppm are consistent with aliphatic structures from plantderived waxes/cuticles (Deshmukh et al., 2003) that have previously been identified in humic extracts (Kelleher and Simpson, 2006; Kelleher et al., 2006; Simpson et al., 2003), and to contributions from microbial lipids (Simpson et al., 2007a). In this DE spectrum, the CH3 signal at e0.8 ppm is likely to be mainly from methylated amino acid side chain residues (Simpson et al., 2007a). This is further dealt with in discussion of Fig. 1C. There is considerable overlap in the 1D NMR resonances. However it has been possible to confirm the suggested assignments by an array of 2D NMR experiments, including HMQC, TOCSY, and NOESY. Applications of 2D NMR for studies of natural organic matter (NOM), and interpretations of the data have been discussed extensively in the literature (Cardoza et al., 2004; Simpson, 2001; Simpson et al., 2001). Briefly, 2D NMR experiments provide increased spectral dispersion as well as additional connectivity information allowing detailed assignments of the chemical functionalities and structural components present (Lam et al., 2007). Fig. 2A shows the Heteronuclear Multiple Quantum Coherence (HSQC) spectrum for the Ho isolated from the PDS that was treated with fertiliser. The HMQC experiment detects one bond 1He13C connectivities in an organic structure (Simpson, 2001). When considered together, the cross-peaks form a specific pattern that can be thought of as the “molecular fingerprint” of a specific structure or class of structure (Kelleher and Simpson, 2006). The HMQC NMR spectrum identifies a range of chemical functionalities present (assignments and references are given in the Figure caption) and suggests that the Ho are a mixture of predominately lignin, protein, carbohydrates, and lipids/cuticlar waxes (Deshmukh et al., 2005, 2003; Kelleher and Simpson, 2006; Lam et al., 2007; Simpson et al., 2007a,b). This is further supported by the TOCSY (Fig. 2C) and NOESY (Fig. 2D) data. All these components have been assigned previously for NOM (Deshmukh et al., 2003; Hertkorn et al., 2006; Kelleher and Simpson, 2006; Kelleher et al., 2006; Lam et al., 2007; Simpson, 2001; Simpson et al., 2003, 2007a,b). Signals due to N-acetyl and/or O-acetyl, previously seen in freshwater DOM (Hertkorn et al., 2006; Lam et al., 2007) are evident in region 10 (Fig. 2A,B). Acetyl groups (Lam et al., 2007), often found in peptidoglycan from microbial cell walls (Simpson et al., 2007b) and in protein (Simpson et al., 2007a) could indicate microbial inputs. The microbial contributions
Table 3 e Exports of hydrophobic acids (Ho) in the drainage water from the well-drained and of poorlydrained soils under different treatment applications. Ho losses mg L1
Treatment
Control Fertiliser Fertiliser & urine
Well-drained soil (WDS)
Poorly-drained soil (PDS)
1.62 2.42 2.25
1.54 3.78 1.87
are most clearly evident from comparisons between spectra for microbial biomass cultured from soil (Simpson et al., 2007a) and that for the Ho in this study. Fig. 1B and C compare the DE spectrum of the Ho with that obtained for microbes cultured from a Canadian dark grey Chernozem soil. The microbes on which Fig. 1C is based were isolated from a different soil to that from which the Ho for Fig. 1B was obtained. The microbes were cultured in a minimal medium with glucose and acetate as carbon sources using a “double spiking approach” (Simpson et al., 2007a). Previous studies have shown that soil microbes give a relatively similar NMR spectrum, irrespective of the soil type from which they are isolated (Simpson et al., 2007a), and the spectrum shown in Fig. 1C shows the extent to which the microbial contributions contribute to the Ho. Comparison of the two spectra indicate that signals from microbial biomass, mainly peaks labelled P, are clearly apparent in the NMR spectrum of the Ho. Characteristic resonances seen for protein/peptide, namely amide (NeH), phenylalanine (Phe), a-protons from amino acid side chains, and methylated side chains are easily distinguishable in both the Ho acid and in the microbial biomass. Furthermore, the region labelled “SC” in Fig. 1C represents the side chain resonances from proteins and peptides. This region can generally be considered as a “fingerprint” region representing the type of peptide/protein present (Simpson et al., 2007a). The side chain region in the Ho acid matches well with that of the microbes. The similarities between the Ho spectrum and that of the microbes, highlights the input of microbial biomass to the Ho isolated from the drainage waters. Components from plant biomass, in addition to microbial inputs, are also in evidence. There are clear indications for lignin-derived components. While these signals are very clear in the HMQC and NOESY data (Fig. 2A, D), they are still apparent in the 1D spectra. Fig. 1D displays the DE spectrum for a lignin standard (organosolv lignin, Sigma Aldrich). The large resonance centered at e3.7 ppm is characteristic of the methoxyl of lignin. Comparison of the spectrum for Ho (1B) with that of the Organosolv lignin (1D) clearly indicates that the apex of the central region of the lignin peak (labelled Lig) in the Ho is from the methoxyl of lignin (Simpson et al., 2007b). This is also confirmed by the intensity of the methoxyl signal in the HMQC data (Fig. 2A). Additionally, aromatic resonances from lignin ate6.3e7 ppm (Lig), are evident in the Ho (Fig. 1B) and these partially overlap with the signal for aromatic residues in proteins/peptides. Thus it can be concluded that the Ho is likely to be a mixture of soil derived plant and microbial materials that have previously been identified in a range of NOM samples (Hertkorn et al., 2006, 2002; Kelleher and Simpson, 2006; Kelleher et al., 2006; Lam et al., 2007; Simpson, 2001, 2002; Simpson et al., 2001, 2003, 2004, 2007a,b).
3.2. Investigation into the effects of the various treatment regimes Results have varied with regard to studies of the effects of N on OM decomposition. Concentrations and fluxes of DOC from the forest floor remained unchanged for field additions of N (Currie et al., 1996; McDowell et al., 1998) whereas the DOC
4385
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 9 e4 3 9 0
H2O
(CH2)n
(CH2)n Lig
CH3
P*
CH3 Lig
Aromatic
P*
Aromatic
D
A
P
P
B
C
E
F
Fig. 3 e 1H NMR spectra for Ho in D2O, differing by soil and treatment. (A), WDS Control; (B), WDS Fertiliser; (C), WDS Fertiliser D Urine; (D), PDS Control; (E), PDS Fertiliser; and (F), PDS Fertiliser D Urine. Simple assignments for spectra indicate strong contributions from aromatic functionalities, from P, proteins/peptides; Lig, lignin; Carb, Carbohydrate; (CH2)n, aliphatic methylene units consistent with aliphatic structures from plant-derived waxes, cuticles and lipids, in addition to contributions from microbial lipids; (CH3), could be due to methylated amino acid side residues plus contributions from terminal methyl groups from plant-derived residues. P* could contain contributions from other molecules such as Refractory carboxyl-rich alicyclic molecules (CRAM).
release rate was found to have decreased by 20% following N fertilization of a forest soil (Cronan et al., 1992). N addition as urea resulted in the increased release of water-soluble OC from a forest soil (Homann and Grigal, 1992). Exports of hydrophobic acids in the drainage water from the well-drained and of poorly-drained soils under different treatment applications are shown in Table 3. Both of the control soils had similar exports of Ho acids in their DW. However, the application of fertiliser gave rise to large increases. Exports of Ho were 1.5 times greater from the WDS, and were almost 2.5 times greater from the PDS. This positive correlation between N fertiliser application and total Ho exported in the cases of both soils may have resulted from increased OM inputs arising from increase grassland productivity. This is proportional to N application (McTiernan et al., 2001), and leads to greater returns of OM to the soil via leaf and root decay (Parsons et al., 1991). The additional OM from the increased plant growth would be a potential source of the Ho that would be transported from the plot by rainwater (McTiernan et al., 2001). In addition urea- and ammoniumbased fertilisers temporarily solubilise SOM and can, as the result of an increase in soil pH, induce a marked increase in DOC content (Chantigny, 2003; Myers and Thien, 1988). However, this effect has been found to be short-lived (Clay et al., 1995).
The NMR spectra obtained for samples after dissolving in DMSO-d6, shown in Fig. 4, are better resolved but contain the same major structural components seen in D2O (Fig. 3). The contribution of peptides to the Ho is more evident in the DMSO-d6 spectra, as seen by the double “hump” ate4e4.4 ppm (a-protons) and by the large amide and methyl resonances (Simpson et al., 2007b). This is most clear in the DE spectra in DMSO (see Fig. 5). The DE spectra are dominated by lignin and microbial signatures indicating that these are the largest of the components in the sample. Regardless of solvent used, the NMR spectra indicate that there is an increase in the lignin contribution to the Ho (Figs. 3 and 4: A vs. B, D vs. E) as the result of fertiliser applications. Absolute quantification from such complex 1D spectra is very difficult, as discussed by Simpson et al. (2007b). However, relative quantification of the methoxyl signal is possible from the 2D HMQC spectra. Absolute quantification is not possible because the signal intensity in the HMQC employed in this study is proportional to the one bond coupling constant (1J 1He13C). The intensity of the methoxyl signal with respect to the total intensity of all peaks in the HMQC (with the exclusion of the DMSO peak) provides an estimation of the abundance of lignin in each sample. This, in turn permits the relative increases/decreases in lignin contents in the different samples to be estimated.
4386
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 9 e4 3 9 0
DMSO
(CH2)n
DMSO
(CH2)n
CH3 P*
CH3
Lig Aromatic/amide
P
Aromatic/amide
A
D
B
E
C
Lig
F
Fig. 4 e 1H NMR spectra for Ho in DMSO-d6, differing by soil and treatment. (A), WDS Control; (B), WDS Fertiliser; (C), WDS Fertiliser D Urine; (D), PDS Control; (E), PDS Fertiliser; and (F), PDS Fertiliser D Urine. Assignments are the same as given in Fig. 3.
Semi-quantitative analysis indicates that, compared to the control, treatment of the soil with fertiliser increased the lignin-derived components in the WD Ho by ca 50%. An increase of 300% was found in the case of the PD Ho. The increases in lignin-derived materials are likely to have resulted from the increased vegetative growth arising from the fertiliser-N amendments. Grazing can result in the deposition to soils of large quantities of urine-N (400e1200 Kg N ha1), and the effects of urine on changes in DOM compositions are not well understood (Rooney et al., 2006). Ho was collected from lysimeter soils amended with both fertiliser-N and urine-N. Applications of fertiliser plus urine (F þ U) caused less Ho losses than the treatment with fertiliser alone, (Table 3) but greater than from the control. 1H NMR spectra in both D2O and DMSO-d6 solvents show a significant decrease in the lignin-derived signal in the Ho isolated from both F þ U treated soils (Figs. 3 and 4: B vs. C, E vs. F). This correlates well with the semi-quantitative analysis that suggested a decrease of 70% (in comparison to the control) in the lignin-derived OM signal for the WDS Ho as the result of treatment of the soil with F þ U. A decrease of 3% was found in the case of the PDS Ho as the result of a similar soil treatment. It is probable that this decrease in C export in the drainage water from the F þ U treated soils resulted from increased microbial activity in the soil from the addition of urine. Under the aerobic conditions that prevailed in the WDS F þ U, the lignin appears to have undergone greater oxidation. Soil respiration was found to be higher from a soil treated with cow urine as the
result of an immediate and significant increase in microbial metabolic activity (Lovell and Jarvis, 1996). Urine contains only small concentrations (0.01%) of soluble carbon (Kishan et al., 1989); however, solubilisation of soil organic C has been shown to take place following urine applications (Monaghan and Barraclough, 1993), and that soluble carbon could provide substrate for increased microbial metabolism (Lovell and Jarvis, 1996). Soils treated with varying concentrations of synthetic sheep urine had greater levels of microbial activity than untreated soils (Rooney et al., 2006). Urine deposition has been shown to alter substantially soil microbial communities, in terms of bacterial and fungal counts and respiration rates (Williams et al., 2000). Differences in microbial biomass activity between grassland types are related to differences in substrate availability (Bardgett et al., 1998; Williams et al., 2000). A strong correlation between N immobilisation and C mineralization has been found (Barrett and Burke, 2000). Rapid stabilization of N was facilitated by an active microbial community and the availability of a readily mineralisable C substrate. It is likely that increased microbial activity induced by the addition of urine promoted the decomposition of the lignin-derived DOC (observed in the NMR spectra in this study) leading to the decrease in the DOC concentration in the drainage water. Conversely, cuticular coatings/leaf waxes are known to be highly recalcitrant and to accumulate over time during the degradation of plants (Kelleher et al., 2006). The relative contributions from aliphatic components compared to the lignin components in the DE-NMR spectra for both PDS and
4387
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 9 e4 3 9 0
Lig
WC/L
11
P
10
12
C P P P
L
DMSO
6
P
2
C
A
1
3
5
DMSO 7
8
9
4
D
B
E
C
F
Fig. 5 e Diffusion edited 1H NMR spectra for Ho in DMSO-d6, differing by soil and treatment. (A), WDS Control; (B), WDS Fertiliser; (C), WDS Fertiliser D Urine; (D), PDS Control; (E), PDS Fertiliser; and (F), PDS Fertiliser D Urine. Assignments are the same as shown in Fig. 3 in addition to WC/L, which refers to waxes, cutins and/or lipids. More specific assignments shown for spectrum D refer to: 1, amide; 2, phenylalanine; 3, aromatics in lignin; 4, anomeric protons in carbohydrates; 5, a-protons (peptides); 6, methoxyl (lignin); 7, carbohydrate protons; 8, methylene adjacent to a carbonyl; 9, N-acetyl and/or O-acetyl group in peptidoglycan; 10, aliphatic methylene units b to an acid or ester; 11, aliphatic methylene; 12, CH3. Changes in the relative abundances of Lignin OCH3 and aliphatic methylene are highlighted by the arrows.
WDS increased with applications of F þ U (Fig. 5, C and F, see arrows). That would correspond to an accumulation of aliphatic components in the Ho. Such may result from a decrease in the more readily degradable fraction (i.e. lignin), resulting in higher concentrations of the ‘less digestible’ cuticular fraction in the soil. Treatment of a grassland soil with sheep urine was found to have led to an increase in the dead or decomposing root mass from 2.2% in the untreated soil (control) to 6.3% in the urine treated soil (Shand et al., 2002). They considered that part of the DOC in the soil solution from beneath the urine patches came from roots damaged by the high concentrations of ammonia (NH3). That could explain the greater contribution of methylene units, possibly from suberin in the root material, to the spectra of the Ho isolated from the DW of the soils treated with F þ U (Fig. 5: C and F, see arrows). On the other hand, the signals consistent with protein/microbial contributions are still dominant in the spectra. Such would be expected as both the urea and N should stimulate microbial activity. There are similarities in the Ho exported from the control soils. The various treatment regimes, however, had greater effects on the PDS. As mentioned, the application of fertiliser caused a greater increase in the exports of Ho from the PDS (Table 3). That could arise, in the case of the poorly-drained
soil, from the decreased aeration that would impede biological oxidation to carbon dioxide (CO2) of the increased organic matter (resulting from the application of fertiliser) (McTiernan et al., 2001). On the other hand, rapid decomposition of organic materials may have taken place in the WDS resulting in the removal of less DOM. In contrast, the F þ U application caused a decrease in the Ho from both the PDS and from the WDS, in comparison to the application of fertiliser alone. That could have arisen from increased microbial activity as a result of the urine additions, leading to a greater metabolism of the SOM and leaving less material available to contribute to the DOM. In summary, the main effects of the varying treatment regimes on the Ho composition from both soils are still not completely resolved. The contribution of lignin components (peak labelled Lig or 6) increased with applications of fertiliser and decreases with fertiliser plus urine addition. The most likely causes of the effects is that the F þ U applications lead to an increase in microbial activity causing microbial utilisation of the more degradable lignin components. Irrespective of the causes of these changes it appears that land management practices significantly alter the composition of dissolved organic matter released into drainage water.
4388 3.3.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 9 e4 3 9 0
Agricultural/environmental significance
Results from the multidimensional solution-state NMR analysis, indicate that the components of Ho in the drainage water of typical Irish grassland soils are complex mixtures of both plant and microbial-derived materials. Strong contributions from lignin and of peptides/proteins of microbial origins were evident in all spectra. Treatment with fertiliser (F) resulted in an increase in the Ho export from both the WDS and the PDS, and an increase in the lignin contribution to the compositions of the Ho. This is thought to result directly from elevated OM inputs to the soil as the result of increased dry matter production through fertilization. Enhanced microbial activity is brought about by inputs of labile C (Lovell and Jarvis, 1996). Increased microbial activity, stimulated by the addition of urine, could result in a degradation of the increased OM input brought about by fertilization. That is reflected by a lower lignin contribution to the Ho isolated from the fertiliser and urine treatment. The drainage regime affected the responses of each soil to the treatments. The decreased aeration in the PDS, compared to the WDS, resulted in a lesser decomposition of the increased OM input in the Ho (McTiernan et al., 2001). In contrast, the fertiliser plus urine application gave rise to a decrease in the Ho from the PDS, compared to the treatment with fertiliser alone. A plausible explanation for this might be that the urine may have been transported more slowly through the PDS resulting in a higher level of microbial activity, increased decomposition, and a lower export of Ho. Growing concern about climate change has increased interest in the role of DOM in the global carbon cycle (Kalbitz and Kaiser, 2008). This study provides further information on the extent and the composition of the organic C lost from soils through transport in drainage water from Irish grassland. Additions of plants with high lignin content have been proposed as a means of building C stocks (Paustian et al., 1997) in order to sequester C. Aromatic compounds from lignin are considered to be the most stable components of DOM (Kalbitz and Kaiser, 2008). Our study indicates, however, that the stimulation of microbial activity by the addition of urine decreases the recalcitrance of the lignin components in the DOM. Investigations of the compositions and the extents to which Ho is lost from soils, as influenced by management practices and the processes involved, will help our understanding of the movement of C between the terrestrial and aquatic environments. Such information is important because it provides an insight into an area of the carbon cycle about which little is known.
4.
Conclusions
Hydrophobic acids (Ho) were isolated from drainage waters and characterised using solution-state NMR. The main conclusions from this study can be summarised as follows: 1. Multidimensional solution-state NMR analysis indicates that the components of the Ho from the drainage water of typical Irish grassland soils are complex mixtures of both plant and microbial-derived materials;
2. Treatment with fertiliser (F) increased the Ho export from both well-drained (WDS) and poorly-drained (PDS) soils, and increased the lignin contribution to the compositions of the Ho. This possibly resulted from elevated OM inputs to the soil as the result of increased dry matter production through fertilization. Application of a fertiliser plus urine (F þ U) mixture resulted in smaller losses of Ho and decreased the lignin-derived signal. This is likely to be attributable to an increase in microbial activity arising from the urine application; 3. The drainage regime affected the responses of each soil to the treatments. Application of fertiliser caused a greater increase in the exports of Ho from the PDS. That reflected the decreased aeration in the PDS, resulting in a lesser decomposition of the increased OM input in the HO. The F þ U application gave rise to a decrease in the Ho from the PDS, compared to the treatment with fertiliser alone. The urine may have been transported more slowly through the PDS resulting in a higher level of microbial activity, increased decomposition, a lower export of Ho, and a lower lignin contribution to the Ho. 4. Our study shows that the stimulation of microbial activity by the addition of urine decreases the recalcitrance of the lignin components.
Acknowledgement We thank the Teagasc, Ireland Walsh fellowship scheme, the Environmental Protection Agency, Ireland, for funding this research, and the International Humic Substances Society for a Training Bursary award to CMPB for a research period in the laboratory of AJS. AJS thanks NSERC (Discovery and Strategic Programs) and an Early Researcher Award (Ontario Government) for providing support.
references
Asakawa, D., Mochizuki, H., Yanagi, Y., Suzuki, T., Nagao, S., Fujitake, N., 2006. Changes in elemental composition, molecular weight and 1H NMR spectra of the waterextractable hydrophobic acid fraction in Cambisol with season and soil depth. Soil Science and Plant Nutrition 52 (3), 361e370. Bardgett, R.D., Keiller, S., Cook, R., Gilburn, A.S., 1998. Dynamic interactions between soil animals and microorganisms in upland grassland soils amended with sheep dung: a microcosm experiment. Soil Biology and Biochemistry 30 (4), 531e539. Barrett, J.E., Burke, I.C., 2000. Potential nitrogen immobilization in grassland soils across a soil organic matter gradient. Soil Biology and Biochemistry 32, 1707e1716. Batjes, N.H., 1998. Mitigation of atmospheric CO2 concentrations by increased carbon sequestration in the soil. Biology and Fertility of Soils 27 (3), 230e235. Billings, S.A., Ziegler, S.E., 2008. Altered patterns of soil carbon substrate usage and heterotrophic respiration in a pine forest with elevated CO2 and N fertilization. Global Change Biology 14 (5), 1025e1036. Byrne, K.A., Kiely, G., Leahy, P., 2007. Carbon sequestration determined using farm scale carbon balance and eddy
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 9 e4 3 9 0
covariance. Agriculture Ecosystems and Environment 121 (4), 357e364. Cameron, K.C., Smith, N.P., McLay, C.D.A., Fraser, P.M., Mcpherson, R.J., Harrison, D.F., Harbottle, P., 1992. Lysimeters without edge flow e an improved design and sampling procedure. Soil Science Society of America Journal 56 (5), 1625e1628. Cardoza, L.A., Korir, A.K., Otto, W.H., Wurrey, C.J., Larive, C.K., 2004. Applications of NMR spectroscopy in environmental science. Progress in Nuclear Magnetic Resonance Spectroscopy 45 (3e4), 209e238. Chantigny, M.H., 2003. Dissolved and water-extractable organic matter in soils: a review on the influence of land use and management practices. Geoderma 113 (3e4), 357e380. Clay, D.E., Clay, S.A., Liu, Z., Harper, S.S., 1995. Leaching of dissolved organic-carbon in soil following anhydrous ammonia application. Biology and Fertility of Soils 19 (1), 10e14. Cronan, C.S., Lakshman, S., Patterson, H.H., 1992. Effects of disturbance and soil amendments on dissolved organiccarbon and organic acidity in Red pine forest floors. Journal of Environmental Quality 21 (3), 457e463. Currie, W.S., Aber, J.D., McDowell, W.H., Boone, R.D., Magill, A.H., 1996. Vertical transport of dissolved organic C and N under long-term N amendments in pine and hardwood forests. Biogeochemistry 35 (3), 471e505. Davidson, E.A., Janssens, I.A., 2006. Temperature sensitivity of soil carbon decomposition and feedbacks to climate change. Nature 440 (7081), 165e173. Deshmukh, A.P., Simpson, A.J., Hadad, C.M., Hatcher, P.G., 2005. Insights into the structure of cutin and cutan from Agave americana leaf cuticle using HRMAS NMR spectroscopy. Organic Geochemistry 36 (7), 1072e1085. Deshmukh, A.P., Simpson, A.J., Hatcher, P.G., 2003. Evidence for cross-linking in tomato cutin using HR-MAS NMR spectroscopy. Phytochemistry 64 (6), 1163e1170. FAO, 2007. I.U.S.S. Working Group WRB World Reference Base for Soil Resources 2006, First Update 2007, Rome. Frank, A.B., Dugas, W.A., 2001. Carbon dioxide fluxes over a northern, semiarid, mixed-grass prairie. Agricultural and Forest Meteorology 108 (4), 317e326. Hayes, T.M., Hayes, M.H.B., Skjemstad, J.O., Swift, R.S., 2008. Compositional relationships between organic matter in a grassland soil and its drainage waters. European Journal of Soil Science 59 (4), 603e616. Hayes, T.M., Watt, B.E., Hayes, M.H.B., Clapp, C.E., Scholfield, D., Swift, R.S., Skjemstad, J.O., 1997. In: Hayes, M.H.B., Wilson, W. S. (Eds.), Humic Substances, Peats and Sludges. Health and Environmental Aspects. Royal Society of Chemistry., Cambridge, UK, pp. 107e120. Hertkorn, N., Benner, R., Frommberger, M., Schmitt-Kopplin, P., Witt, M., Kaiser, K., Kettrup, A., Hedges, J.I., 2006. Characterization of a major refractory component of marine dissolved organic matter. Geochimica et Cosmochimica Acta 70 (12), 2990e3010. Hertkorn, N., Permin, A., Perminova, I., Kovalevskii, D., Yudov, M., Petrosyan, V., Kettrup, A., 2002. Comparative analysis of partial structures of a peat humic and fulvic acid using oneand two-dimensional nuclear magnetic resonance spectroscopy. Journal of Environmental Quality 31 (2), 375e387. Hoekstra, N.J., Lalor, S.T.J., Richards, K.G., O’Hea, N., Lanigan, G.J., Dyckmans, J., Schulte, R.P.O., Schmidt, O., 2009. Slurry 15NH4N recovery in herbage and soil: effects of application method and timing. Plant and Soil. doi:10.1007/s11104-009-0210-z. Homann, P.S., Grigal, D.F., 1992. Molecular-weight distribution of soluble organics from laboratory-manipulated surface soils. Soil Science Society of America Journal 56 (4), 1305e1310.
4389
Jaksic, V., Kiely, G., Albertson, J., Oren, R., Katul, G., Leahy, P., Byrne, K.A., 2006. Net ecosystem exchange of grassland in contrasting wet and dry years. Agricultural and Forest Meteorology 139 (3e4), 323e334. Kaiser, E., Simpson, A.J., Dria, K.J., Sulzberger, B., Hatcher, P.G., 2003. Solid-state and multidimensional solution-state NMR of solid phase extracted and ultrafiltered riverine dissolved organic matter. Environmental Science and Technology 37 (13), 2929e2935. Kalbitz, K., Kaiser, K., 2008. Contribution of dissolved organic matter to carbon storage in forest mineral soils. Journal of Plant Nutrition and Soil Science-Zeitschrift Fur Pflanzenernahrung und Bodenkunde 171 (1), 52e60. Kalbitz, K., Solinger, S., Park, J.H., Michalzik, B., Matzner, E., 2000. Controls on the dynamics of dissolved organic matter in soils: a review. Soil Science 165 (4), 277e304. Kelleher, B.P., Simpson, A.J., 2006. Humic substances in soils: are they really chemically distinct? Environmental Science and Technology 40 (15), 4605e4611. Kelleher, B.P., Simpson, M.J., Simpson, A.J., 2006. Assessing the fate and transformation of plant residues in the terrestrial environment using HR-MAS NMR spectroscopy. Geochimica et Cosmochimica Acta 70 (16), 4080e4094. Kim, S., Simpson, A.J., Kujawinski, E.B., Freitas, M.A., Hatcher, P.G., 2003. High resolution electrospray ionization mass spectrometry and 2D solution NMR for the analysis of DOM extracted by C-18 solid phase disk. Organic Geochemistry 34 (9), 1325e1335. Kingery, W.L., Simpson, A.J., Hayes, M.H.B., Locke, M.A., Hicks, R. P., 2000. The application of multidimensional NMR to the study of soil humic substances. Soil Science 165 (6), 483e494. Kishan, J., Khan, M.Y., Lal, M., 1989. Effect of feeding and fasting on urinary nitrogen and carbon excretion in Desi cows. Indian Journal of Animal Sciences 59 (11), 1465e1467. Lam, B., Baer, A., Alaee, M., Lefebvre, B., Moser, A., Williams, A., Simpson, A.J., 2007. Major structural components in freshwater dissolved organic matter. Environmental Science and Technology 41 (24), 8240e8247. Leenheer, J.A., 1981. Comprehensive approach to preparative isolation and fractionation of dissolved organic-carbon from natural-waters and wastewaters. Environmental Science and Technology 15 (5), 578e587. Lovell, R.D., Jarvis, S.C., 1996. Effects of urine on soil microbial biomass, methanogenesis, nitrification and denitrification in grassland soils. Plant and Soil 186 (2), 265e273. Malcolm, R.L., MacCarthy, P., 1992. Quantitative-evaluation of XAD-8 and XAD-4 resins used in tandem for removing organic solutes from water. Environment International 18 (6), 597e607. McDowell, W.H., Currie, W.S., Aber, J.D., Yano, Y., 1998. Effects of chronic nitrogen amendments on production of dissolved organic carbon and nitrogen in forest soils. Water Air and Soil Pollution 105 (1e2), 175e182. McTiernan, K.B., Jarvis, S.C., Scholefield, D., Hayes, M.H.B., 2001. Dissolved organic carbon losses from grazed grasslands under different management regimes. Water Research 35 (10), 2565e2569. Monaghan, R.M., Barraclough, D., 1993. Nitrous-oxide and dinitrogen emissions from urine-affected soil under controlled conditions. Plant and Soil 151 (1), 127e138. Myers, R.G., Thien, S.J., 1988. Organic-matter solubility and soil reaction in an ammonium and phosphorus application zone. Soil Science Society of America Journal 52 (2), 516e522. Parsons, A.J., Orr, R.J., Penning, P.D., Lockyer, D.R., Ryden, J.C., 1991. Uptake, cycling and fate of nitrogen in grass clover swards continuously grazed by sheep. Journal of Agricultural Science 116, 47e61. Paustian, K., Andren, O., Janzen, H.H., Lal, R., Smith, P., Tian, G., Tiessen, H., Van Noordwijk, M., Woomer, P.L., 1997.
4390
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 9 e4 3 9 0
Agricultural soils as a sink to mitigate CO2 emissions. Soil Use and Management 13 (4), 230e244. Qualls, R.G., Haines, B.L., 1991. Geochemistry of dissolved organic nutrients in water percolating through a forest ecosystem. Soil Science Society of America Journal 55 (4), 1112e1123. Rooney, D., Kennedy, N., Deering, L., Gleeson, D., Clipson, N., 2006. Effect of sheep urine deposition on the bacterial community structure in an acidic upland grassland soil. Applied and Environmental Microbiology 72 (11), 7231e7237. Royer, I., Angers, D.A., Chantigny, M.H., Simard, R.R., Cluis, D., 2007. Dissolved organic carbon in runoff and tile-drain water under corn and forage fertilized with hog manure. Journal of Environmental Quality 36 (3), 855e863. Scurlock, J.M.O., Hall, D.O., 1998. The global carbon sink: a grassland perspective. Global Change Biology 4 (2), 229e233. Shand, C.A., Williams, B.L., Dawson, L.A., Smith, S., Young, M.E., 2002. Sheep urine affects soil solution nutrient composition and roots: differences between field and sward box soils and the effects of synthetic and natural sheep urine. Soil Biology and Biochemistry 34 (2), 163e171. Simpson, A., 2001. Multidimensional solution state NMR of humic substances: a practical guide and review. Soil Science 166 (11), 795e809. Simpson, A.J., 2002. Determining the molecular weight, aggregation, structures and interactions of natural organic matter using diffusion ordered spectroscopy. Magnetic Resonance in Chemistry 40, S72eS82. Simpson, A.J., Brown, S.A., 2005. Purge NMR: effective and easy solvent suppression. Journal of Magnetic Resonance 175 (2), 340e346. Simpson, A.J., Burdon, J., Graham, C.L., Hayes, M.H.B., Spencer, N., Kingery, W.L., 2001. Interpretation of heteronuclear and multidimensional NMR spectroscopy of humic substances. European Journal of Soil Science 52 (3), 495e509. Simpson, A.J., Kingery, W.L., Hatcher, P.G., 2003. The identification of plant derived structures in humic materials using three-dimensional NMR spectroscopy. Environmental Science and Technology 37 (2), 337e342.
Simpson, A.J., Lefebvre, B., Moser, A., Williams, A., Larin, N., Kvasha, M., Kingery, W.L., Kelleher, B., 2004. Identifying residues in natural organic matter through spectral prediction and pattern matching of 2D NMR datasets. Magnetic Resonance in Chemistry 42 (1), 14e22. Simpson, A.J., Simpson, M.J., Smith, E., Kelleher, B.P., 2007a. Microbially derived inputs to soil organic matter: are current estimates too low? Environmental Science and Technology 41 (23), 8070e8076. Simpson, A.J., Song, G.X., Smith, E., Lam, B., Novotny, E.H., Hayes, M.H.B., 2007b. Unraveling the structural components of soil humin by use of solution-state nuclear magnetic resonance spectroscopy. Environmental Science and Technology 41 (3), 876e883. Smejkalova, D., Piccolo, A., 2008. Aggregation and disaggregation of humic supramolecular assemblies by NMR diffusion ordered spectroscopy (DOSY-NMR). Environmental Science and Technology 42 (3), 699e706. Soussana, J.F., Loiseau, P., Vuichard, N., Ceschia, E., Balesdent, J., Chevallier, T., Arrouays, D., 2004. Carbon cycling and sequestration opportunities in temperate grasslands. Soil Use and Management 20, 219e230. Stark, C.H., Richards, K., Fay, D., Dennis, S.J., Sills, P., Staples, V., 2007. A decision support system for sustainable grazing management regimes e the effect of urine application timing and soil type on N loss to the environment. In: Holden, N.M., Hochstrasser, T., Schulte, R.P.O., Walsh, S. (Eds.), Making Science Work on the Farm: a Workshop on Decision Support Systems for Irish Agriculture. AGMET, Dublin, pp. 114e119. Thrippleton, M.J., Keeler, J., 2003. Elimination of zero-quantum interference in two-dimensional NMR spectra. Angewandte Chemie-International Edition 42 (33), 3938e3941. UNFCCC, 1998. Kyoto Protocol to the United Nations Framework Convention on Climate Change. Williams, B.L., Grayston, S.J., Reid, E.J., 2000. Influence of synthetic sheep urine on the microbial biomass, activity and community structure in two pastures in the Scottish uplands. Plant and Soil 225 (1e2), 175e185. Zsolnay, A., 2003. Dissolved organic matter: artefacts, definitions, and functions. Geoderma 113 (3e4), 187e209.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 5 1 e4 4 6 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Contribution of combined sewer overflows to trace contaminant loads in urban streams Philip Weyrauch a, Andreas Matzinger a,*, Erika Pawlowsky-Reusing b, Stephan Plume a, Do¨rthe von Seggern c, Bernd Heinzmann d, Kai Schroeder a, Pascale Rouault a a
Kompetenzzentrum Wasser Berlin, Cicerostrasse 24, 10709 Berlin, Germany Berliner Wasserbetriebe, Netz- und Anlagenbau, Neue Ju¨denstrasse 1, 10864 Berlin, Germany c Berlin Senate Department of Health, Environment and Consumer Protection, Bru¨ckenstrasse 6, 10179 Berlin, Germany d Berliner Wasserbetriebe, Research and Development, Cicerostrasse 24, 10709 Berlin, Germany b
article info
abstract
Article history:
The present study examines the contribution of combined sewer overflows (CSO) to loads
Received 10 November 2009
and concentrations of trace contaminants in receiving surface water. A simple method to
Received in revised form
assess the ratio of CSO to wastewater treatment plant (WWTP) effluents was applied to the
11 May 2010
urban River Spree in Berlin, Germany. The assessment indicated that annual loads are
Accepted 6 June 2010
dominated by CSO for substances with removal in WWTP above w95%. Moreover, it
Available online 12 June 2010
showed that substances with high removal in WWTP can lead to concentration peaks in the river during CSO events. The calculated results could be verified based on eight years of
Keywords:
monitoring data from the River Spree, collected between 2000 and 2007. Substances that
Combined sewer overflows
are well removed in WWTP such as NTA (nitrilotriacetic acid) were found to occur in
CSO
significantly increased concentration during CSO, while the concentration of substances
EDTA
that are poorly removable in WWTP such as EDTA (ethylenediaminetetraacetic acid)
NTA
decreased in CSO-influenced samples due to dilution effects. The overall results indicate
Priority pollutants
the potential importance of the CSO pathway of well-removable sewage-based trace
River spree
contaminants to rivers. In particular, high concentrations during CSO events may be relevant for aquatic organisms. Given the results, it is suggested to include well-removable, sewage-based trace contaminants, a substance group often neglected in the past, in future studies on urban rivers in case of combined sewer systems. The presented methodology is suggested for a first assessment, since it is based solely on urban drainage data, which is available in most cities. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Urban streams are affected by trace contaminants from wastewater treatment plants (WWTP), stormwater effluents and e in the case of combined sewer system e combined sewer overflows (CSO). In particular, pollutants that persist in WWTP are an ongoing issue. For instance persistent
carbamazepine, sulfamethoxazole and diclofenac have been identified recently as top priority pharmaceutical residues by the Global Water Research Coalition (2008). A second group of substances that receive increasing attention are both persistent and well-removable contaminants in stormwater runoff from impervious urban surfaces, which can have a significant impact on receiving waters (Grapentine et al., 2008; Hwang
* Corresponding author. Tel.: þ49 30 53653824; fax: þ49 30 53653888. E-mail address:
[email protected] (A. Matzinger). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.011
4452
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 5 1 e4 4 6 2
and Foster, 2008). Especially polycyclic aromatic hydrocarbons (PAH), polychlorinated biphenyls (PCB) and trace metals were identified as priority pollutants from stormwater effluents (Eriksson et al., 2007). In comparison, well-removable substances, which are predominantly found in municipal wastewater (in the following referred to as sewage-based substances), have received little attention, since they are retained in WWTP and e if measured e only rarely detected at high concentrations in river monitoring programs. However, they can be contained in CSO, which bypass WWTP, in concentrations exceeding those in treated wastewater (Buerge et al., 2006; Phillips and Chalmers, 2009). As a result, an important CSO contribution of such non-persistent substances would be expected in surface waters after rain storms. The general effect was confirmed by Benotti and Brownawell (2007), who found that concentrations in a bay influenced by WWTP and CSO remained high for well-removable paracetamol (CAS 103-90-2) and caffeine (CAS 58-08-2) after a major storm event, while concentrations decreased clearly for all other sewage-based trace organics. Similarly, Buerge et al. (2006) showed that wellremovable caffeine can be used as a tracer for untreated wastewater in streams. Their results imply that the impact of rain events on the CSO share of total annual loads of such well-removable substances may be significant. Apart from the annual contaminant influx, peak concentrations in water bodies, such as rivers or reservoirs, might be caused by CSO. Potential peak concentrations during CSO could be an important issue, since non-persistent substances typically show high biological activity. For instance the synthetic hormone ethinylestradiol (CAS 57-63-6), which is removed in WWTP by w90% (Joss et al., 2004), can have an endocrinal effect on aquatic organisms already at very low concentrations and short exposures (Caldwell et al., 2008). The impact of trace substances from CSO in receiving rivers is difficult to assess, since quasi-continuous measurements would be required during storm events. However, even occasional measurements of trace substances during CSO events in receiving rivers or CSO require a high sampling and analytical effort and are, therefore, rarely performed (Welker, 2007). This article presents two simple methods, which allow (a) the estimation of CSO contribution to annual mass balances of sewage-based substances of varying WWTP removal in receiving rivers and (b) the estimation of a critical WWTP removal fraction, above which peak concentrations are expected during CSO in the receiving river. Both methods do not require actual measurements of concentrations but are based on volumetric flows via WWTP and CSO, data which are typically available today for most larger cities. In a second step, methods (a) and (b) are exemplified for the Berlin urban drainage system. The results are compared with a unique dataset of trace contaminant concentrations in the receiving River Spree during CSO. A particular focus lies on sewage-based trace contaminants EDTA and NTA, but a selected number of other substances of varying origin (stormwater versus sewage) as well as hygienic parameters are included. Finally local and general implications of sewagebased trace contaminants from CSO are discussed. Abbreviations and mathematical variables are explained in Table 1.
2.
Materials and methods
2.1.
Study site
The combined sewer system in the centre of Berlin, Germany, drains an area of w100 km2 with 1.4 million inhabitants (Table 2). This corresponds to about 20% of the total drained area of Berlin, but almost 40% of the city’s population. The remainder of the area is drained by a separated sewer system (Fig. 1). In the combined sewer area, wastewater is collected together with stormwater and pumped to the WWTP, which are mostly located at the edge or beyond the city limits. If the storage volume of the combined sewer system is exceeded during storm events, combined sewage overflows to the River Spree and its side channels via 179 CSO discharge points over a stretch of w18 km (Table 2). CSO were found to occur typically for rain events (separated by at least 6 h without rain) exceeding a total amount of 4.7 mm for the most sensitive areas (Riechel, 2009). On average this comparably low margin was exceeded 37 times per year between 2000 and 2007, with large interannual variability (Fig. 2). For instance, 48 rain events with a total height > 4.7 mm occurred in exceptionally wet year 2007 compared to 30 events in dry year 2003. In terms of volume, currently about seven million cubic meters combined sewage enter the River Spree each year, with an average sewage to stormwater ratio of 1:11 (unpublished data, Berliner Wasserbetriebe). In order to meet legislative requirements a long term pollution control plan has been set up, which involves different measures to upgrade the combined sewer system. Until 2020 these measures will be realized including, e.g., full utilization of the static in-pipe storage capacities by heightening of CSO crests (under consideration of flood prevention), implementation of actuators like weirs, sluices and throttles for real-time control and construction of additional stormwater tanks. Whereas effluents from WWTP are measured, total CSO loads and contained raw sewage were available from simulations with the urban drainage network model KOSMO, which requires input data on precipitation, canalization network, hydrological data, inhabitant density and sewage loads (Schmitt, 1993). For the Berlin case study, hydrology and pollutant loads were simulated based on a verified simplified canalization network for the years 1961e1981 (unpublished data, Berliner Wasserbetriebe, 2008). Simulation results were available for two scenarios, (i) before realization of rehabilitation measures and (ii) after complete realization of rehabilitation measures in 2020. Table 3 contains estimated raw sewage effluents from CSO based on simulations for the time period 2000e2007 and for 2020. The receiving River Spree flows through the combined sewer area before it joins the River Havel, which leaves Berlin in the South West (Fig. 1). Average monthly flow rates before the confluence (at the monitoring station in Fig. 1) vary between 12 m3 s1 in summer and 42 m3 s1 in spring (longtime averages published by the Senate of Berlin). Apart from CSO the River Spree receives stormwater effluents upstream of the city centre and the treated wastewater of WWTP 1 (Fig. 1). WWTP 2 was also discharging into the River Spree, before it was decommissioned in March 2003 after having gradually decreased its operation since November 2002. As
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 5 1 e4 4 6 2
4453
Table 1 e Explanation of abbreviations and variables in the text. Abbreviation
Explanation
General abbreviations CSO PNEC S-CSO S-noCSO WWTP Water quality parameters BOD5 DEHP DO EDTA EE2 NH4-N NTA SMX Variables criv,dry,i criv,i criv,wet,i cww,i QCSO Qriv Qriv,dry Qriv,wet QWWTP
Combined sewer overflow Predicted no effect concentration Grab samples taken at monitoring station during CSO influence Grab samples taken at monitoring station without CSO influence Wastewater treatment plant Five-day biological oxygen demand bis(2-ethylhexyl)phthalate Dissolved oxygen Ethylenediaminetetraacetic acid Ethinylestradiol Ammonium Nitrilotriacetic acid Sulfamethoxazole Concentration of substance i in receiving river at dry weather [mg m3] Concentration of substance i in receiving river [mg m3] Concentration of substance i in receiving river during CSO influence [mg m3] Concentration of substance i in untreated sewage (excluding stormwater) [mg m3] Flow of raw sewage contained in CSO (e.g., w9% for Berlin) [m3 s1] Outflow of the considered river stretch [m3 s1] Dry weather Qriv [m3 s1] Qriv during CSO (¼wet weather) [m3 s1] Flow of treated sewage contained in WWTP effluent (e.g., in Berlin w93% on average for WWTP 1 and WWTP 2, w7% is stormwater) [m3 s1] Volume of considered receiving river stretch [m3] CSO share of the total load (to a river stretch) of a sewage-based substance i [] Critical removal fraction in WWTP above which concentration increase is expected in the receiving river during CSO [] Removal fraction in WWTP for substance i []
Vriv xCSO,i hcrit hi
a result of this closure, the wastewater, which formerly flowed to WWTP 2, has been redirected out of the watershed of the River Spree, which reduced the load of treated wastewater to the River Spree by two-thirds to about 40,000 m3 d1 today. Measured discharges of treated sewage from WWTP 1 and WWTP 2 are summarized in Table 3. Sampling of the River Spree is taking place at a monitoring station just before the confluence with the River Havel,
Table 2 e Characteristics of the Berlin combined sewer system (city centre only) and the receiving River Spree. Parameter Combined sewer system Total area Drained area Impervious area Inhabitants CSO volume per year Share of wastewater in CSO Number of CSO discharge points River Spree Average flowa River widtha Length of stretch impacted by CSO a At monitoring station (Fig. 1).
Unit
Value
km2 km2 km2 106 106 m3 % e
100 86 64 1.4 7 9 179
m3 s1 m km
26 50 18
downstream of the combined sewer system, WWTP 1 and WWTP 2 (Fig. 1). Fig. 3 exemplifies the typical qualitative reaction of the River Spree to CSO, during the rain event on 2005-09-11, which was studied in detail by Schumacher et al. (2007). During the event w16.7 mm of rain fell over a period of w6 h, which led to a total CSO volume of w510,000 m3 (based on calculations in Schumacher et al. (2007), extrapolated to the total central combined sewer area). A rain event of this extent occurs two to three times in an average year. Shortly after the rain event, a drop in specific conductivity can be observed in continuous measurements in the River Spree (Fig. 3), as a result of dilution with stormwater. Concurrently with the conductivity drop, DO started to decrease and reached a minimum about 9 h later, interrupting the natural dayenight fluctuation of DO. A total of 1.5 days after the rain event, a second drop in conductivity can be seen, again followed by a DO depression. The second signal is also the result of CSO on 2005-09-11, but from CSO discharge points on the River Spree and its side channels further upstream. Since flow speeds in the River Spree are very low between 0.02 and 0.15 m s1, it typically takes several days for the water to pass the area of the combined sewer system (Fig. 1). Finally, three days after the rain event, the natural dayenight DO fluctuation from phytoplankton activity recovers (Fig. 3). While durations, time lags and extent of DO reduction and conductivity drops vary between CSO events, the qualitative phenomena in the River Spree were found to be representative (Riechel, 2009).
4454
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 5 1 e4 4 6 2
Fig. 1 e River network and urban drainage of Berlin, Germany. The monitoring station on the River Spree is impacted by overflows from the central combined sewer area (shaded in grey, only the combined sewer area which influences the monitoring station is displayed) and the two wastewater treatment plants WWTP 1 and, until 2003, WWTP 2. Arrows indicate direction of flow of the rivers.
2.2. Theoretical contribution of CSO to contaminant loads and river concentrations The following methodology allows an estimation of (a) contribution of CSO to annual loads of sewage-based trace substances and (b) types of substances which must be expected to lead to concentration peaks in the receiving river during CSO events. Variables used in the following equations and in the text are explained in Table 1. (a) can be assessed by comparing the untreated sewage that enters a stream via CSO (QCSO) with the inflow of treated sewage from WWTP (QWWTP). However, since treatability
Fig. 2 e Precipitation events at rain gauge in SE Berlin (see Fig. 1 for location). Horizontal line indicates threshold of 4.7 mm for combined sewer overflows in most sensitive areas based on Riechel (2009).
varies for different sewage constituents i, the removal fraction hi in WWTP has to be taken into account to calculate substance-specific load from WWTP, cww,i $ QWWTP $ (1 hi). The share of CSO (xCSO,i) of the total load of sewage-based substances in a receiving river can be calculated based on equation (1), assuming that (1) sewage-based substances only stem from WWTP or CSO, neglecting upstream sources, (2) cww,i is the same for WWTP and CSO and (3) hi is the same during dry and wet weather: xCSO;i ¼
QCSO QWWTP $ð1 hi Þ þ QCSO
(1)
Note that cww,i [mg m3] cancels out on the right-hand side of equation (1). Based on equation (1), relative contribution of CSO can be judged for any sewage-based substance i, for which the removal fraction in WWTP hi is known, without elaborate substance-based mass balances. Although simple, equation (1) requires information on loads from WWTP and CSO, excluding the share of stormwater. Whereas effluents from WWTP are usually measured, total CSO loads and contained raw sewage are typically estimated semi-quantitatively in most larger cities. For (b), the identification of substances which may lead to concentration peaks during CSO, we need to compare expected concentrations in the river during dry weather condition and during CSO, in the following referred to as wet weather condition. Since we are only interested in the ratio between dry and wet weather condition and not dynamics of criv,i, we chose a simplified mixed reactor approach, assuming in addition to assumptions (1e3) above that (4) the affected river stretch is completely mixed, thus neglecting spatial and temporal dynamics in CSO, (5) effluents QCSO and QWWTP, as
4455
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 5 1 e4 4 6 2
Table 3 e Treated and untreated sewage volumes (excluding stormwater) entering the monitored river section for the period before closure of WWTP 2 (before 2003), between 2003 and 2007 and after completion of planned restructuring measures (after 2020). QCSOa
Period
6
3
QWWTP 1b 1
6
3
1
QWWTP 2b 6
3
QCSO/(QWWTP 1 þ QWWTP 2)
1
[10 m yr ]
[10 m yr ]
[10 m yr ]
[%]
0.76 0.61 0.30
16.21c 13.83d 13.83d
33.33c 0 0
1.5 4.4 2.2
2000e2003 2003e2007 2020
a Average simulated volumes of raw sewage flowing to the River Spree, based on 21 years of rainfall data (1961e1981) and change in storage volume. b Treated sewage flowing to the River Spree. c Three-year average (2000e2002). d Five-year average (2003e2007).
well as river flow Qriv,wet and Qriv,dry remain constant over the considered time span and (6) no reactions within the river are considered: dcriv;i Input Output ¼ limDt/0 dt Vriv $Dt cww;i $ð1 hi Þ$QWWTP cww;i $QCSO criv;i $Qriv ¼ þ Vriv Vriv Vriv
cww;i $ð1 hi Þ$QWWTP Qriv;dry
with criv;wet;i ð0Þ ¼ criv;dry;i ; criv;wet;i ðNÞ ¼
(2)
where Input and Output refer to mass fluxes to the receiving river stretch during time interval Dt. Under dry weather conditions QCSO ¼ 0. Moreover, steady state can be assumed, which leads to: criv;dry;i ¼
Qriv;wet ,t criv;wet;i ðtÞ ¼ criv;wet;i ð0Þ criv;wet;i ðNÞ $e Vriv þ criv;wet;i ðNÞ
(3)
During wet weather conditions, substance i enters the river both via WWTP and CSO. Since CSO are of limited duration, steady state assumption is no longer reasonable. Dynamic solution of equation (2) leads to:
cww;i $ð1 hi Þ$QWWTP þ cww;i $QCSO Qriv;wet
ð4Þ
Equation (4) results in exponential change of criv,i from steady state dry weather criv,dry,i towards steady state wet weather condition criv,wet,i(N). Based on equations (2) and (4) a theoretical substance removal fraction hcrit can be calculated, for which criv,dry,i ¼ criv,wet,i: hcrit ¼ 1
Qriv;dry Qcso , QWWTP Qriv;wet Qriv;dry
(5)
For substances with hi > hcrit, concentrations in the river are expected to increase during CSO, for substances with hi < hcrit dilution is expected. In actual measurements, one would expect a clear dilution or peak effect only for substances which are clearly below or above hcrit, respectively, since assumption (4) neglects other aspects that influence
Fig. 3 e Continuous data of dissolved oxygen and conductivity after a typical rain event. Rain data is from rain gauge (Fig. 1), river data is from continuous measurements by the Senate of Berlin near the measuring station.
4456
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 5 1 e4 4 6 2
substance concentrations in the receiving river, such as nonuniform distribution of CSO along the river, substance use patterns (e.g., Heberer, 2002), sediment in the canalization (El Samrani et al., 2004) and changing content of raw sewage in CSO over the time of a CSO event (Bertrand-Krajewski et al., 1998; Stenstrom et al., 2008). hcrit is event-specific, since QCSO, Qriv,dry and Qriv,wet are different for every CSO. Expected uncertainty in the estimation of QCSO directly translates to uncertainty in hcrit. Equation (5) allows simple estimation whether peak concentrations in the river are likely to occur, since it is independent of concentration dynamics in equation (4) during CSO event as well as the affected river volume. In contrast equations (2)e(4) cannot be used to calculate realistic concentration dynamics for a specific CSO event, given the assumptions in deriving equation (2). In summary (a) CSO contribution to annual loads of trace contaminants can be estimated with equation (1), whereas (b) equation (5) provides an idea, whether dilution or increased concentration is expected for substances in a river during CSO. Both (a) and (b) can be calculated without information on concentrations in WWTP, CSO or the receiving river.
2.3.
Sampling and analytics
Sewage-based trace contaminants NTA and EDTA were focused on for the verification of equations (1) and (5), as suggested in Welker (2007). To put findings into perspective, dissolved zinc, which is predominantly expected in stormwater (Welker, 2004) was included in the analysis. Moreover, mixed origin substances AOX and DEHP were considered, for which it is a priori unknown which pathway, sewage or stormwater dominates (Welker, 2004). In addition to trace contaminants, sewage-based hygienic parameters fecal streptococci, Escherichia coli and total coliforms were assessed for further verification regarding particle-related substances. Monthly grab samples were taken for the period 2000e2007 at the monitoring station in a water depth of 0.5 m (Fig. 1). Samples were stored in glass bottles and directly analyzed at the certified state laboratory “Landeslabor Berlin-Brandenburg” for the parameters in Table 4 using international standard ISO methodology.
In addition, continuous measurements of standard water quality parameters, DO and specific conductivity, were performed at the monitoring station via online sensors (WTW TechnoLine Oxi 171 for DO, WTW TechnoLine LF 170 for specific conductivity) on a bypass. Online sensors were calibrated on a weekly interval.
2.4.
Identification of samples during CSO
One particular challenge is the identification of grab samples that are influenced by CSO. The probability to take a grab sample during CSO influence in the River Spree can be estimated at w4%, assuming 37 CSO events per year with average signal duration at the river monitoring station of 10 h. As a result 2e4 samples with CSO influence would be expected for 43e103 available monthly samples (Table 4). It has to be noted that this estimate is rather optimistic, since small CSO events do not necessarily lead to a signal in the river (Riechel, 2009). Although Riechel (2009) showed that drops in conductivity and DO after storm events are a reliable sign for CSO influence (compare Fig. 3), direct identification is not possible, because of natural seasonal and daily variations of these indicator parameters. For instance, only 22% of variance in DO is explained by a multiple regression model based on CSO indicators conductivity, NH4-N and BOD5 for the period 2000e2007. As a result, identification was attempted, based on prior knowledge on CSO occurrence and reactions during CSO in the River Spree. Firstly, rain data were evaluated to find whether a CSO event has taken place before sampling. If a CSO was likely to have occurred, known CSO indicators were assessed in the River Spree at the time of sampling. The approach aims at identifying grab samples with a high probability of CSO influence, accepting the risk of missing samples taken during minor CSO impact. Hourly rain data were obtained from a rain gauge in the south east of Berlin within the combined sewer area (see Fig. 1), which measures precipitation with a 0.1 mm accuracy. These rain data were merged to rain events, separated by at least 6 h without rain (Riechel, 2009). Rain events with a total height above 4.7 mm were considered as critical, i.e., as likely to cause CSO in the most sensitive areas, based on the analysis by Riechel (2009). Since average daily flow velocities of the
Table 4 e Overview of samples and analyzed parameters. Parameter
CAS number
Period of analysis
Number of analyzed samples
Method
AOX Total coliform bacteria
e e
2000e2007 2000e2007
103 103
DEHP
117-81-7
2002e2007
60
EDTA Escherichia coli
60-00-4 e
2001e2006 2000e2007
78 103
FFE GC/MS (laboratory method) DIN EN ISO 16588 DIN EN ISO 9308-3
Fecal streptococci
e
2000e2007
103
DIN EN ISO 7899-1
NTA Dissolved zinc
139-13-9 7440-66-6
2001e2006 2000e2007
78 103
DIN EN ISO 16588 DIN EN ISO 11885
DIN EN ISO 9562 DIN EN ISO 9308-3
Detection range
Number of samples above detection limit
>10.0 mg L1 >30 cfu/100 mL <10000 cfu/100 mL >0.50 mg L1
91 103
>1.0 mg L1 >30 cfu/100 mL <10000 cfu/100 mL >40 cfu/100 mL <8200 cfu/1 mL >5.0 mg L1 >10.0 mg L1
78 103
5
103 50 75
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 5 1 e4 4 6 2
River Spree can be as low as 0.02 m s1, CSO effects can reach the monitoring station with a maximal lag of 18 km/ 0.02 m s1 z 10 days (see also Fig. 3 for comparison). Thus, the data of monthly grab samples were first reduced to samples, which were taken in a 10-day period after a critical rain event. In the next steps it was analyzed, if a CSO influence could be verified for these samples in the River Spree. Since time of CSO impact at the monitoring station is strongly dependent on CSO duration, CSO location and flow velocity in the river (compare Fig. 3), we focused on continuous data to judge whether a grab sample was taken in a CSO “cloud”. As indicated in Fig. 3 and systematically shown by Riechel (2009), negative peaks in DO and conductivity are expected during CSO influence. In addition to the continuous data, grab sample parameters DO, conductivity, NH4-N and BOD5 were evaluated. Both NH4-N and BOD5 are expected to increase as a result of CSO. Since all the parameters can fluctuate strongly as a result of other processes than CSO, we applied four qualitative filters. In the following analysis grab samples were only considered impacted by CSO: 1 if they were taken within a 10-day period after a critical rain event with a total height > 4.7 mm, 2 if continuous DO and conductivity measurements showed a sharp decrease following the critical rain event, 3 if DO and conductivity of grab samples verify that the sample was taken during the observed CSO signal in continuous measurements and 4 if at least one of the two parameters NH4-N or BOD5 in grab sample showed a higher value than grab samples taken in the months before and after the sample (or similarly high, if these samples were also identified as CSO-influenced).
3.
Results and discussion
3.1. Theoretical contribution of CSO to contaminant loads and river concentrations Fig. 4 shows CSO share of annual sewage-based substance loads to the River Spree for varying removal fractions in WWTP, based on equation (1) and values in Table 3. Several substance examples are given in Fig. 4 for the current situation (2003e2007). For instance, CSO currently contribute w4% to the annual load of the chelating agent EDTA, which is practically not removed in WWTP (hEDTA w0%) (Reemtsma et al., 2006). These 4% correspond to the volume ratio of untreated versus total (untreated plus treated) sewage that enters the River Spree (Table 3). For substances i with hi > 0% the CSO share increases (Fig. 4). NTA, with hNTA w97% (Alder et al., 1990), stems predominantly (w60%) from CSO with w40% contribution from WWTP. Two sewage-based substances in between are the synthetic hormone EE2 (hEE2 w90%, (Joss et al., 2004)) and the antibiotic SMX (CAS 723-46-6, hSMx w30%, (LANUV, 2006)). Whereas CSO are almost neglectable for SMX with 6% (94% from WWTP), CSO and WWTP contribute 31% and 69% to EE2 loads, respectively, (Fig. 4). If we compare the three situations in Fig. 4 before 2003, from 2003 to 2007 and the situation as planned for 2020, the CSO share
4457
Fig. 4 e CSO share of sewage-based substance loads as a function of substance removal in WWTP for the River Spree in Berlin. Calculations are based on equation (1) and Table 3. Examples are indicated with dashed lines for nitrilotriacetic acid (NTA), ethylenediaminetetraacetic acid (EDTA), sulfamethoxazole (SMX) and ethinylestradiol (EE2) for the current situation (2003e2007).
decreases both if WWTP effluents increase (before closure of WWTP 2 in 2003) and if CSO decrease (planned measures until 2020). However, it has to be kept in mind that in Fig. 4 only relative shares are shown, not loads. In the Berlin case, total loads are highest before 2003, and lowest in 2020 (Table 3). Even if occurrence of substances is dominated by WWTP on an annual basis in Fig. 4, CSO can lead to concentration peaks during storm events, depending on substance removal fraction. Based on equation (5) we can estimate hcrit, i.e., the substance removal fraction at which dilution from rain water and higher substance loads from CSO are balanced. In the following we estimate hcrit for the River Spree using the 6-h rain event on 2005-09-11, which was discussed earlier (see also Fig. 3). During the event, w510,000 m3 of CSO with an estimated content of raw sewage of 5% (personal communication, E. Pawlowsky-Reusing 2009, based on simulation results) entered the River Spree and its side channels. This corresponds to more than twice the 6-h effluent of treated sewage from WWTP 1 (QWWTP $ 6 h z 11,000 m3). In the example, river flow increased more than 6-fold from Qriv,dry z 12 m3 s1 before the CSO event to Qriv,wet z 75 m3 s1 during the 6 h event. Based on these numbers, equation (5) resulted in hcrit ¼ 56% for 2005-09-11. Thus, for the exemplary storm event with CSO, dilution would occur for substances with hi < 56% and an increase in river concentration for substances with hi > 56%. Therefore, non-removable EDTA and very wellremovable NTA (hNTA w97%) are ideal tracers to test whether the theoretical effect can actually be observed or if other aspects dominate concentrations in the river.
3.2.
Validation with measurements
Fig. 5 shows an example for the identification of grab samples which are impacted by CSO, along the four filter rules described in Section 2.4:
4458
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 5 1 e4 4 6 2
Fig. 5 e Example for the identification of samples impacted by CSO as described in the Materials and methods section. (a) Continuous data for oxygen concentration and conductivity plus rain events yielding at least 4.7 mm; (b) annual course of monthly samples for dissolved oxygen concentration and conductivity in monthly grab samples; (c) ammonium concentration and BOD5 in monthly grab samples. Dotted line indicates time of sampling. Note different time scale in (a).
1 A critical rain event with a volume of 48.8 mm has been recorded from 2001-06-17 until 2001-06-18 (in Fig. 5a rain event period is marked with a grey bar), one day before the sample was taken on 2001-06-18 and thus fulfilling filter rule one. 2 Following the rain event continuously measured DO and conductivity dropped clearly (Fig. 5a), passing filter rule two. 3 The drop in DO and conductivity was verified in the grab sample (Fig. 5b), which satisfies rule three. 4 Finally, a parallel increase in BOD5 was observed, whereas NH4-N does not show a clear signal (Fig. 5c), but is clearly in the upper range compared to three months before and after the sampling. As the four filter rules are satisfied, the grab sample taken on 2001-06-18 was considered as influenced by CSO. Between 2000 and 2007, a total of eight grab samples were preselected, which (i) were within a 10-day period after a critical rain event (filter rule 1) and (ii) showed a decrease in continuous
DO as a consequence of the respective rain event (filter rule 2). These eight samples were further filtered for CSO influence following the filter rules as exemplified in Fig. 5 (Table 5). Based on the filter exercise, we have identified four grab samples with high CSO influence (S-CSO), which can be compared to up to 99 samples without major CSO impact (S-noCSO). In the following a comparison is made for sewagebased tracers NTA and EDTA, as well as the other trace organics and hygienic parameters described in Section 2.3 (Table 6). For the concentration of each monitored parameter, the mean and standard deviation s was calculated (i) of S-CSO, as well as (ii) of S-noCSO. Since WWTP contribution to the contamination of the River Spree by trace substances was significantly changed by the closure of WWTP 2 between November 2002 and March 2003, samples were divided into two time periods, before and after 2003-01-23 (which is the point when effluents from WWTP 2 were reduced by 50%; in the following this date is referred to as the closure of WWTP 2). As S-CSO only contains 1e3 grab samples within each time period, classical statistical tests are not sensible. Instead we used the relative difference between average concentrations as a first mean of comparison. In addition, mean 2s for SnoCSO was calculated to test whether S-CSO are within or outside expected w95% normal distribution of S-noCSO. For hygienic parameters and EDTA, which show highly skewed distributions, expected 95% intervals are also indicated for the assumption of lognormal distribution (using natural logarithm) in the text. Concentrations, rather than loads are considered, since measured data only represent a spot sample. Continuous measurements of concentrations of monitored substances and of flow velocities of the River Spree would be required for reliable load calculations. As expected in the theoretical analysis, sewage-based chelating agents NTA and EDTA show a clear signal as a result of CSO (Fig. 6). The observed increase of NTA in the River Spree during CSO is about tenfold before the closure of WWTP 2 and about fourfold after closure of WWTP 2. EDTA decreased in the River Spree during observed CSO events by 31% before 2003 and by 20% after 2003 (for lognormal distribution: 19% and 7% decrease before and after 2003, respectively). The closure of WWTP 2 is clearly visible in EDTA concentrations in Fig. 6, which underlines the dominance of the pathway via WWTP. On the other hand, no effect of the closure is seen for NTA, verifying the high removal of NTA loads at WWTP. Average NTA concentrations during CSO influence also clearly exceed expected 2s distribution of background samples, further supporting the effect of CSO. In contrast, dilution of EDTA during CSO events is only statistically significant for one of three CSO events (2.4 mg L1 on 2004-08-14) when assuming lognormal distribution (lognormal 95% interval of S-noCSO is 2.5e24.2 mg L1 before 2003 and 2.5e8.3 mg L1 after 2003). Despite a good agreement with theoretical expectations, Fig. 6 shows samples with increased NTA or decreased EDTA concentrations that were not regarded as CSO-influenced. The additional signals could result from an incomplete selection of samples with CSO influence, since filter rules were strictly applied. Hygienic parameters, such as fecal streptococci, E. coli or total coliform bacteria show an increase in concentration during CSO (Table 6). Fecal streptococci and E. coli exceed 2s distribution for two of four samples during CSO (2001-06-18
4459
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 5 1 e4 4 6 2
Table 5 e Application of filter rules for preselected samples, which passed filter rules 1 and 2. Measurement
2000-04-25 2001-06-18a 2003-09-15 2004-08-14a 2005-09-14a 2006-08-14 2007-05-09a 2007-08-16
Filter rule 1
Filter rule 2: Continuous data
Filter rule 3: Grab sample
Filer rule 4: Grab sample
Days since beginning of rain event >4.7 mm
Decreased oxygen?
Decreased c onductivity?
Decreased oxygen?
Decreased conductivity?
Increased NH4-N?
Increased BOD5?
9.7 0.9 4.5 3.4 3.2 1.3 1.0 0.7
U U U U Ub U U U
U U e U Ub e U U
U U U U U U U U
e U U U U e U e
e e e U U U U U
U U U e U e e e
a Samples, which were classified as influenced by CSO. b Continuous measurements were not available at the monitoring station, but 3 km upstream at a temporarily installed set of online sensors.
and 2007-05-09) and for averages after 2003, giving statistical evidence that they differ significantly from other samples. Under the assumption of lognormal distribution one S-CSO sample after 2003 shows significantly higher values than S-noCSO (lognormal 95% interval of S-noCSO after 2003 is 8e336 Fecal streptococci per 100 ml and 8e7477 E. coli per 100 ml). The increase during CSO could be expected, since these specific bacteria are generally removed above 90% by WWTP (Fleischer et al., 2000). The results show that the theoretical approach, based on removal efficiency in WWTP, also holds for other sewage-based parameters than trace organics. As microorganisms tend to attach to suspended matter, the results indicate that a significant share of the particulate fraction of raw sewage enters the River Spree during CSO. As a result, sewage-based particle-bound trace contaminants are also expected to show a similar behavior as
EDTA and NTA. Among hygienic parameters the lowest increase during CSO is found for total coliforms, which are also clearly within 95% boundaries of S-noCSO, both for normal and lognormal distribution. The lower effect of CSO can be explained, since total coliforms are a sum parameter, which includes bacteria that are not sewage-based. The explanation is verified by comparably high average background concentration of 500 2000 total coliform bacteria per 100 ml in the River Spree upstream of Berlin (Berlin Senate, online water quality reports, 2000e2007). AOX, DEHP and dissolved zinc can be contained both in stormwater from impervious surfaces, as well as in raw sewage. For instance AOX are contained in sewage (e.g., X-ray contrast media like diatrizoic acid (CAS 737-31-5) and iopamidol (CAS 60166-93-0) (Seitz et al., 2006)) and in stormwater (e.g., bleaching agents in industrial effluents (Welker and
Table 6 e Comparison of concentrations of monitored substances in samples influenced and in samples not influenced by CSO. Data were separately calculated for samples taken before and after closure of WWTP 2 on 2003-01-23. For measurements below analytical detection limits or above upper bounds for microbiological parameters, measurement limits were used for the calculation of averages. Substances are sorted according to the difference in average values after 2003-01-23. Substance
Before 2003-01-23 Samples not influenced by CSO Mean
Fecal streptococci in 100 ml Escherichia coli in 100 ml NTA (mg L1) Total coliform bacteria in 100 ml DEHP (mg L1) Dissolved zinc (mg L1) AOX (mg L1) EDTA (mg L1)
2s
After 2003-01-23
Samples influenced by CSO N
Mean
N
Diff. of means
Samples not influenced by CSO
Samples influenced by CSO
Diff. of means
Mean
2s
N
Mean
s
N
881
3351
(39)
2000
(1)
127%
165
1426
(60)
4091
6874
(3)
2385%
17,983
96,555
(39)
100,000
(1)
456%
1164
5027
(60)
4120
5963
(3)
254%
1.5 61,743
1.8 326,702
(26) (39)
16.0 200,000
(1) (1)
981% 223%
1.5 2799
1.6 6713
(49) (60)
5.4 5167
3.4 5054
(2) (3)
251% 85%
0.50 16.6
0 16.1
(5) (39)
e 27.5
(0) (1)
e 66%
0.50 13.8
0.16 10.0
(52) (60)
0.87 19.4
0.64 11.0
(3) (3)
74% 40%
16 9.0
9 9.9
(39) (26)
20 6.2
(1) (1)
26% 31%
19 3.2
13 4.1
(60) (49)
23 2.6
6 0.2
(3) (2)
17% 20%
4460
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 5 1 e4 4 6 2
Fig. 6 e Concentrations of NTA and EDTA from samples taken in the years 2000 until 2006. Samples influenced by CSO are marked with a circle. Note that 2007 event is not shown, because of monitoring period of NTA and EDTA (compare Table 3). The horizontal lines indicate average concentrations. Average values were calculated separately for samples taken before and after the shutdown of WWTP 2.
Schmitt, 1997)). Moreover they can stem from natural sources such as algae and plants (Gribble, 2003). In Table 6, AOX does not react strongly to CSO, which is indicated by a slight increase during CSO well within 2s boundaries of S-noCSO. This can be explained by increases during stormwater events below CSO threshold. Similarly to AOX, DEHP, a plasticizer contained in many synthetics, can occur both in sewage and stormwater. Surprisingly, DEHP was almost exclusively found above detection limit during CSO (mean of S-CSO is, therefore, clearly beyond 2s boundaries of S-noCSO), which indicates the important contribution of this pathway. Finally, dissolved zinc, which is mainly contained in stormwater via roofs and rain drains, only shows a minor signal during CSO. The observations for AOX, DEHP and dissolved zinc show that substances of mixed origin can be impacted by CSO, but need to be assessed individually.
3.3.
was detected above EU environmental quality standard for annual averages AA-EQS of 1.3 mg L1 (maximal allowable concentration is not defined for DEHP) only on one instance for the observed 5-year period (1.60 mg L1 during CSO on 200408-16), indicating compliance with AA-EQS. However, there are other, highly removable, sewage-based substances with clearly higher acute toxicities than NTA: e.g., synthetic hormone EE2 (removal in WWTP: w90%, PNEC: 3 106 mg L1 (LANUV, 2007)), human analgesics paracetamol (w99%, 9.2 mg L1 (LANUV, 2007)) or aspirin (CAS 50-78-2, w99%, 8.0 mg L1 (LANUV, 2007)), synthetic fragrance musk xylene (CAS 81-15-2, >92%, 1.1 mg L1 (LANUV, 2006)), disinfectant triclosan (CAS 3380-34-5, 90 to >95%, 0.05 mg L1 (LANUV, 2006)), solvent octylphenol (CAS 140-66-9, 85e95%, 0.1 mg L1 (LANUV, 2006)) or plasticizer triphenyl phosphate (CAS 11586-6, 90 to 95%, 0.1 mg L1 (LANUV, 2006)).
Implications for receiving rivers
4. The analysis shows that annual, sewage-based substance loads in the Berlin River Spree are dominated by CSO for substances with a removal in WWTP > 95%. During CSO events, well-removable substances (with removal above w56% for the analyzed single event) can also lead to peak concentrations in the river, despite dilution with rain water. This behavior could be verified for well-removable (w97%) NTA, which was detected at up to 10-fold background concentrations during CSO in the River Spree. Maximal detected NTA concentration during CSO of 16.0 mg L1 is clearly below PNEC of 930 mg L1 (European Chemicals Bureau, 2005). Similarly, CSO-influenced DEHP
Conclusions
Results imply that CSO may be a significant source of wellremovable trace substances and can lead to potentially problematic acute concentrations in receiving waters during storm events. Measurements in the River Spree show that the phenomenon is not limited to small receiving streams, which were often the focus in earlier CSO studies (e.g., HvitvedJacobsen, 1982; Krejci et al., 1994). In turn, receiving rivers in most historic cities in Europe (e.g., Paris, London or Rome), as well as North America (e.g., most cities in North-Eastern USA) have a combined sewer system and must be expected to show a similar situation as Berlin.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 5 1 e4 4 6 2
The study underlines the potential relevance of substances, which are well-removable in WWTP and, therefore, rarely assessed in detail, in rivers receiving CSO. Regarding the ongoing improvement of WWTP removal fractions by new technologies (Nakada et al., 2007; Zuehlke et al., 2006), the relative contribution of CSO to the total contamination of surface water may become even more important. In the assessment of urban water cycles with CSO influence the following steps are suggested: 1 A simple relative mass balance as exemplified in this paper could be used for a first assessment of the potential contribution of CSO to trace contaminant loads and concentration peaks in the receiving water. 2 If the first assessment indicates possible importance of the CSO pathway, a substance screening on raw sewage is suggested to identify concerned contaminants. 3 Frequent sampling (e.g., via autosampler) on the receiving river during some exemplary rainstorm events is suggested to identify the extent of the problem.
Acknowledgement The study was conducted at Kompetenzzentrum Wasser Berlin in collaboration with Berliner Wasserbetriebe and the Berlin Senate Department of Health, Environment and Consumer Protection. The study was financed by Berliner Wasserbetriebe and Veolia Water. Particular thanks go to Berliner Wasserbetriebe for rain and sewer system data and to the Berlin Senate Department of Health, Environment and Consumer Protection for river water quality data.
references
Alder, A.C., Siegrist, H., Gujer, W., Giger, W., 1990. Behaviour of NTA and EDTA in biological wastewater treatment. Water Research 24 (6), 733e742. Benotti, M.J., Brownawell, B.J., 2007. Distributions of pharmaceuticals in an urban estuary during both dry- and wet-weather conditions. Environmental Science and Technology 41 (16), 5795e5802. Bertrand-Krajewski, J.L., Chebbo, G., Saget, A., 1998. Distribution of pollutant mass vs volume in stormwater discharges and the first flush phenomenon. Water Research 32 (8), 2341e2356. Buerge, I.J., Poiger, T., Mu¨ller, M.D., Buser, H.R., 2006. Combined sewer overflows to surface waters detected by the anthropogenic marker caffeine. Environmental Science and Technology 40 (13), 4096e4102. Caldwell, D.J., Mastrocco, F., Hutchinson, T.H., La¨nge, R., Heijerick, D., Janssen, C., Anderson, P.D., Sumpter, J.P., 2008. Derivation of an aquatic predicted no-effect concentration for the synthetic hormone, 17a-ethinyl estradiol. Environmental Science and Technology 42 (19), 7046e7054. El Samrani, A.G., Lartiges, B.S., Ghanbaja, J., Yvon, J., Kohler, A., 2004. Trace element carriers in combined sewer during dry and wet weather: an electron microscope investigation. Water Research 38 (8), 2063e2076. Eriksson, E., Baun, A., Scholes, L., Ledin, A., Ahlman, S., Revitt, M., Noutsopoulos, C., Mikkelsen, P.S., 2007. Selected stormwater
4461
priority pollutants e a European perspective. Science of the Total Environment 383 (1e3), 41e51. European Chemicals Bureau, 2005. European Union risk assessment report for trisodium nitrilotriacetate, p. 71. Fleischer, J., Schlafmann, K., Otchwemah, R., Botzenhart, K., 2000. Elimination of enteroviruses, other enteric viruses, F-specific coliphages, somatic coliphages and E. coli in four sewage treatment plants of southern Germany. Journal of Water Supply: Research and Technology e AQUA 49 (3), 127e138. Global Water Research Coalition, 2008. Development of an international priority list of pharmaceuticals relevant for the water cycle, London. Grapentine, L., Rochfort, Q., Marsalek, J., 2008. Assessing urban stormwater toxicity: methodology evolution from point observations to longitudinal profiling. Water Science and Technology 57, 1375e1381. Gribble, G.W., 2003. The diversity of naturally produced organohalogens. Chemosphere 52 (2), 289e297. Heberer, T., 2002. Tracking persistent pharmaceutical residues from municipal sewage to drinking water. Journal of Hydrology 266, 175e189. Hvitved-Jacobsen, T., 1982. The impact of combined sewer overflows on the dissolved oxygen concentration of a river. Water Research 16 (7), 1099e1105. Hwang, H.M., Foster, G.D., 2008. Polychlorinated biphenyls in stormwater runoff entering the tidal Anacostia River, Washington, DC, through small urban catchments and combined sewer outfalls. Journal of Environmental Science and Health e Part A Toxic/Hazardous Substances and Environmental Engineering 43 (6), 567e575. Joss, A., Andersen, H., Ternes, T., Richle, P.R., Siegrist, H., 2004. Removal of estrogens in municipal wastewater treatment under aerobic and anaerobic conditions: consequences for plant optimisation. Environmental Science and Technology 38, 3047e3055. Krejci, V., Schilling, W., Gammeter, S., 1994. Receiving water protection during wet weather. Water Science and Technology 29 (1e2), 219e229. LANUV, 2006. Final Report: Effluents and Removal Fractions of Dangerous Contaminants in Communal Wastewater Treatment Plants. Landesamt fu¨r Natur, Umwelt und Verbraucherschutz, Nordrhein-Westfalen (in German), p. 515. LANUV, 2007. Immission, Behaviour and Retention of Pharmaceuticals in the Environment, Literature Study. IWW Rheinisch-Westfa¨lisches Institut fu¨r Wasserforschung Gemeinnu¨tzige GmbH, Mu¨lheim a.d. Ruhr, Recklinghausen (in German), p. 358. Nakada, N., Shinohara, H., Murata, A., Kiri, K., Managaki, S., Sato, N., Takada, H., 2007. Removal of selected pharmaceuticals and personal care products (PPCPs) and endocrine-disrupting chemicals (EDCs) during sand filtration and ozonation at a municipal sewage treatment plant. Water Research 41 (19), 4373e4382. Phillips, P., Chalmers, A., 2009. Wastewater effluent, combined sewer overflows, and other sources of organic compounds to Lake Champlain. Journal of the American Water Resources Association 45 (1), 45e57. Reemtsma, T., Weiss, S., Mueller, J., Petrovic, M., Gonza´lez, S., Barcelo, D., Ventura, F., Knepper, T.P., 2006. Polar pollutants entry into the water cycle by municipal wastewater: a European perspective. Environmental Science and Technology 40 (17), 5451e5458. Riechel, M., 2009. Impact Assessment of Combined Sewer Overflows on the Berlin River Spree. Technical University Berlin, Berlin (in German). Schmitt, T.G, 1993. Detailed combined sewer overflow simulation regarding new German guidelines A-128. In: Proceedings of
4462
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 5 1 e4 4 6 2
the 5th International Conference on Urban Storm Drainage, Niagara Falls, Canada, pp. 1260e1275. Schumacher, F., Gebauer, U., Pawlowsky-Reusing, E., Meier, I., Schroeder, K., Leszinski, M., Heinzmann, B., 2007. Integrated Sewage Management e Water Quality Simulation of River Spree and its Canals (Reach Charlottenburg) Under Consideration of Combined Sewer Overflows for a Storm Event in September 2005. Kompetenzzentrum Wasser Berlin gGmbH, Berlin (in German), p. 84. Seitz, W., Weber, W.H., Jiang, J.Q., Lloyd, B.J., Maier, M., Maier, D., Schulz, W., 2006. Monitoring of iodinated X-ray contrast media in surface water. Chemosphere 64 (8), 1318e1324. Stenstrom, M.K., Barco, J., Papiri, S., 2008. First flush in a combined sewer system. Chemosphere 71 (5), 827e833.
Welker, A., Schmitt, T.G., 1997. A basic investigation in origins of AOX-substances in paper sludges. Water Research 31 (4), 805e815. Welker, A., 2004. Contaminant Fluxes in the Urban Water CycledOccurrence and Distribution in Urban Drainage Systems. Technical University Kaiserslautern, Habilitation (in German) p. 439. Welker, A., 2007. Assessment of organic pollutants with respect to occurrence and fate in combined sewer systems and possible impacts on receiving waters. Water Science and Technology 56 (10), 141e148. Zuehlke, S., Duennbier, U., Lesjean, B., Gnirss, R., Buisson, H., 2006. Long-term comparison of trace organics of removal performances between conventional and membrane activated sludge processes. Water Environment Research 78 (13), 2480e2486.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 1 e4 3 9 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Degradation mechanisms and kinetic studies for the treatment of X-ray contrast media compounds by advanced oxidation/reduction processes Joonseon Jeong a, Jinyoung Jung b,**, William J. Cooper a, Weihua Song a,* a
Urban Water Research Center, Department of Civil and Environmental Engineering, University of California, Irvine, CA 92697-2175, USA b Department of Environmental Engineering, Yeungnam University, 214-1 Dae-Dong, Gyeongsan-Si, Gyeongsangbuk-Do 712-749, Republic of Korea
article info
abstract
Article history:
The presence of iodinated X-ray contrast media compounds (ICM) in surface and ground
Received 30 November 2009
waters has been reported. This is likely due to their biological inertness and incomplete
Received in revised form
removal in wastewater treatment processes. The present study reports partial degradation
9 April 2010
mechanisms based on elucidating the structures of major reaction by-products using
Accepted 31 May 2010
g-irradiation and LC-MS. Studies conducted at concentrations higher than observed in
Available online 9 June 2010
natural waters is necessary to elucidate the reaction by-product structures and to develop destruction mechanisms. To support these mechanistic studies, the bimolecular rate
Keywords:
constants for the reaction of OH and eaq with one ionic ICM (diatrizoate), four non-ionic
X-ray contrast media
ICM (iohexol, iopromide, iopamidol, and iomeprol), and the several analogues of diatrizoate
Advanced oxidation processes
were determined. The absolute bimolecular reaction rate constants for diatrizoate, iohexol,
Hydroxyl radical
iopromide, iopamidol, and iomeprol with OH were (9.58 0.23)108, (3.20 0.13)109,
Hydrated electron
(3.34 0.14)109, (3.42 0.28)109, and (2.03 0.13) 109 M1 s1, and with eaq were (2.13 0.03)1010, (3.35 0.03)1010, (3.25 0.05)1010, (3.37 0.05)1010, and (3.47 0.02) 1010 M1 s1, respectively. Transient spectra for the intermediates formed by the reaction of OH were also measured over the time period of 1e100 ms to better understand the stability of the radicals and for evaluation of reaction rate constants. Degradation efficiencies for the OH and eaq reactions with the five ICM were determined using steady-state g-radiolysis. Collectively, these data will form the basis of kinetic models for application of advanced oxidation/reduction processes for treating water containing these compounds. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Recently, increasing attention has been focused on the presence of pharmaceutical compounds in aquatic environments
due to their widespread use and incomplete removal during wastewater treatment (Ku¨mmerer, 2004; Khetan and Collins, 2007). Although the concentration of pharmaceutical compounds encountered in most aquatic environments
* Corresponding author. Tel.: þ1 949 878 7908; fax: þ1 949 824 3672. ** Corresponding author. Tel.: þ82 53 810 2541; fax: þ82 53 810 4624. E-mail addresses:
[email protected] (J. Jung),
[email protected] (W. Song). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.05.054
4392
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 1 e4 3 9 8
currently are at trace levels, ng L1 to mg L1, their continuous input may constitute a long-term potential risk for aquatic and terrestrial organisms (Sacher et al., 2001). The presence of pharmaceutical compounds in the environment is leading to growing concern because relatively little is known about their impact on human and ecosystem health, especially concerning chronic toxicity from the continuous exposure to multiple compounds at far below medicinal doses (Daughton and Ternes, 1999; Fent et al., 2006; Halling-Sorensen et al., 1998). Iodinated X-ray contrast media compounds (ICMs) are the most widely used pharmaceutical compounds for intravascular administration, which enhances the visualization of the structure or fluids within the body during diagnostic tests. The annual worldwide consumption of ICM is approximately 3.5 106 kg (Pe´rez and Barcelo´, 2007). Conventional wastewater treatment processes do not effectively remove ICMs resulting in their detection at mg L1 levels in effluents of wastewater treatment plants, surface water, and ground water (Pe´rez and Barcelo´, 2007; Putschew et al., 2001, 2000; Sacher et al., 2001; Ternes and Hirsch, 2000). The discovery that these compounds contribute substantially to the adsorbable organic halogen (AOX) fraction in hospital wastewater has raised additional concerns (Ku¨mmerer et al., 1998). In drinking water treatment ICMs were only partially removed by ozonation at 35e55% of non-ionic ICMs and below 20% of ionic ICMs (Seitz et al., 2008). Although there is little if anything known of their impact on health effects, based on precautionary principles, drinking water should be free of these compounds to minimize the potential risk of long-term adverse health effects (Huber et al., 2003). Thus, it is essential that alternative treatment technologies be developed which effectively degrade these compounds. Advanced oxidation/reduction processes (AO/RPs) utilize reactive species involving hydroxyl radical (OH) as the oxidant and either the reducing hydrated electron (eaq) or hydrogen atom (H) to destroy recalcitrant compounds (Rosenfeldt and Linden, 2004; Song et al., 2008b). Results on the treatment of ICMs by advanced oxidation processes such as UV/H2O2, UV/TiO2, O3/H2O2, have been reported (Doll and Frimmel, 2004; Huber et al., 2005; Ternes et al., 2003). However, these studies have been limited to examining the feasibility of such processes on the removal of ICMs, not providing mechanistic and kinetic details necessary to optimize treatment processes utilizing OH, eaq, and/or H. The potential application of free radical processes for the control of ICMs, the destruction mechanisms of five ICM from reaction with OH and eaq in air-equilibrated solution was studied. Reaction by-products were identified with liquid chromatography-mass spectroscopy (LC-MS). These studies were conducted at concentrations higher than observed in natural waters to enable identification of individual reaction by-products. To support this study the bimolecular reaction rate constants for the OH and eaq with one ionic ICM (diatrizoate), four non-ionic ICM (iohexol, iopromide, iopamidol, and iomeprol), and several analogues (model compounds) of diatrizoate were also determined. These data are of importance in eventually developing kinetic models describing the
destruction of the ICMs under advanced oxidation/reduction process conditions.
2.
Materials and methods
2.1.
Materials
All chemicals were of reagent grade and used without further purification. Diatrizoate and its structurally simpler analogues, 2, 3, 5-triiodobenzoic acid, 3-acetamidobenzoic acid, and acetanilide were purchased from SigmaeAldrich Co. One analogue of diatrizoate, 3, 5-diacetamidobenzoic acid was purchased from ChemBridge Co. (San Diego, CA). Iohexol and iopromide were obtained from Schering/Berlin, Germany, and iopamidol and iomeprol were courtesy of Ilsung Pharmaceuticals Co. (Korea). The chemical structures of five ICM selected for this study are shown in Table 1.
2.2.
Pulse radiolysis and g-radiolysis
The experimental details of this portion of the study are provided in the Supplementary Material (Text S1) as they have been discussed in detail elsewhere (Buxton and Stuart, 1995; Whitham et al., 1996). Radiolysis of water generates three highly reactive species (OH, eaq, and H) and, by adjusting the reaction conditions, is the simplest way to obtain quantifiable concentrations of the oxidizing or reducing species for simulating AO/RPs:
(0.28) OH þ (0.06) H þ (0.27) eaq þ (0.05) H2O H2 þ (0.07) H2O2 þ (0.27) Hþ
(1)
where the numbers in parentheses are the concentration of each species (G-values, mmol J1) (Buxton et al., 1988; Spinks and Woods, 1964). The reaction of OH was studied in nitrous oxide (N2O) saturated aqueous solutions where eaq and hydrogen atom (H) are converted into OH (Buxton et al., 1988; Song et al., 2008b). The solutions used to study the reactions of eaq were pre-saturated with nitrogen in the presence of 0.1 M isopropyl alcohol in order to scavenge the OH and H, converting them into relatively inert isopropyl radicals (Buxton et al., 1988; Song et al., 2008a). Steady-state g-irradiation experiments were performed using a Shepherd 109-86, 60Co irradiator with a dose rate of 0.0722 kGy min1 as measured by Fricke dosimetry.
2.3.
Chemical analysis
Iodide released by the reaction of ICM with OH and eaq was quantified by ion chromatography (DX-120, Dionex) with conductivity detection. Separation was performed on an IonPac AS16 anion column (4 250 mm, Dionex) using 35 mM NaOH eluent solution at a flow rate of 1.0 mL min1. Details of the HPLC analysis are provided in the Supplementary Material (Text S2) and are similar to those detailed elsewhere (Song et al., 2008a, 2008b).
Table 1 e Summary of the molecular structures, maximum molar absorptivities of transient spectra, second-order rate constants, and degradation efficiencies for the five ICM in this study. Parameter/units
Diatrizoate
O O
Structure
OH
I N H
Iohexol
I
I
OH
O
O
N H
I
N
O
OH O
I
HO
OH
H N
I N H
O
N
OH I H N
I
O
OH
O
OH
N H
O
I H N I
O
OH
OH
H N
O I
OH I
OH
OH
H N
N I
OH
O
OH
3710 (280 nm) 7540 (360 nm) 4400 (410 nm) 0.958 0.023 2.13 0.03 0.241 0.013
1530 (290 nm) 3500 (400 nm)
1220 (290 nm) 1040 (370 nm)
2060 (280 nm) 1190 (370 nm)
1580 (280 nm) 1260 (370 nm)
3.21 0.13 3.35 0.03 0.385 0.048
3.34 0.14 3.25 0.05 0.380 0.064
3.42 0.28 3.37 0.05 0.361 0.023
2.03 0.13 3.47 0.02 0.354 0.029
0.225 0.035
0.436 0.032
0.396 0.047
0.401 0.058
0.376 0.041
40 7 69 3
79 6 81 16
71 8 87 12
73 11 75 4
68 8 78 4
8
Results and discussion
4 6 Irradiation dose (kGy)
3.
2
8
Deiodination of ICM
0
6
4
Irradiation dose (kGy)
3.1.
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0 2
4393
Solutions of diatrizoate and iohexol, as representative of ionic and non-ionic ICM, were irradiated (60Co), in air-equilibrated, N2O-saturated solution, and N2-saturated 0.1 M isopropyl alcohol solutions, and the iodide ion (I) measured (Fig. 1). The unit of x-axis, irradiation dose (kGy) is defined as the product of dose rate (0.0722 kGy min1 for 60Co irradiator employed in this experiment) and irradiation time (min), which is equivalent to the total yield of radical species during irradiation. Irradiation of both diatrizoate and iohexol resulted in the
A
B
Released Iodide (mM)
0
Fig. 1 e Deiodination of (A) diatrizoate (4.54 3 10L4 M) and (B) iohexol (4.07 3 10L4 M) by g-irradiation in airequilibrated (-C-), N2O-saturated (-B-) solution, and in N2-saturated (-;-) 0.1 M isopropyl alcohol solution (The dashed lines represents stoichiometric IL concentrations in solution).
Released Iodide (mM)
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 1 e4 3 9 8
109 k OH /M1 s1 1010 keaq /M1 s1 Initial degradation rate/mM kGy1 (aerated solution) Initial degradation rate/mM kGy1 (N2O-saturated solution) Degradation efficiency of OH/% Degradation efficiency of eaq/%
OH
NH
I
O
OH
OH
emax (M1 cm1cm1) (lmax)
Iomeprol
OH
OH O
I
Iopamidol
OH
H N
O
Iopromide
4394
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 1 e4 3 9 8
O O
OH
I
I
N H
N H
I
O
Diatrizoate (MW. 614)
O O
OH I
N H
N H
OH
O
O
(MW. 504)
I
I
N H
N H
I
O
O
I
N H
N H
O
O
O
(MW. 474)
N H
I
OH
I
I
N H
N H
OH
OH O
(MW. 504)
OH
I
OH
I N H
(MW. 586)
O O
O
OH
I
O
O
(MW. 476)
O
I
OH O
N H
N H
I
O
(MW. 476)
I
O
N H
N H
I
O
(MW. 474)
Fig. 2 e Proposed degradation pathway of diatrizoate.
release of I in N2-saturated 0.1 M isopropyl alcohol solution. The deiodination for iohexol was more efficient than for diatrizoate, and the total yields of I released from diatrizoate and iohexol at 8.7 kGy was 88% and 99%, respectively. In the air-equilibrated solution, the efficiency of deiodination for both compounds was markedly lower than that observed in N2-saturated 0.1 M isopropyl alcohol solution. This reflects the competition of the eaq with O2 (1.9 1010 M1 s1, Buxton et al., 1988) and suggests that the superoxide radical ion resulting from the reaction of the eaq with O2 does not react to any great extent with the compounds. The release of I from diatrizoate approached a final yield of 30%, while the iohexol achieved 96% mass balance at 8.7 kGy. Deiodination for both compounds observed in N2O-saturated solution was lowest of all, diatrizoate 27% and iohexol 52%. The deiodination for the other three non-ionic ICM (iopromide, iopamidol, iomeprol) under the same experimental conditions employed (Fig. 1) were close to that observed for iohexol (Fig. S1 of Supplementary Material). These data are consistent with the measured degradation efficiencies where a relatively higher efficiency was observed for the four non-ionic ICM relative to diatrizoate (Table 1). In N2O-saturated solution, where the OH should be the sole reactive species responsible for the deiodination, the more efficient deiodination of iohexol compared to that of diatrizoate reflects the faster rate for the reaction of OH with iohexol (3.21 109 M1 s1) than that with diatrizoate (9.58 108 M1 s1) with a resulting efficiency of 79 and 40%, at
O
8.7 kGy. One possible explanation for the leveling off of I concentration with increasing irradiation dose is the competing reaction of OH with I (k ¼ 1.1 1010 M1 s1, Buxton et al., 1988) as the concentration of I increased. However, from the LC-MS data, there was no indication of iodination of the ICMs which might have occurred with the oxidized iodine. More details on the degradation efficiencies and rate constants for OH and eaq with those compounds will be discussed later.
3.2.
Destruction pathways of ICM
The identification of products for degradation of ICMs by g-irradiation was based on the analysis of the total ion chromatograms (TIC) and the corresponding mass spectra that were obtained by negative ion electrospray LC-MS. The masses of the different products were determined from the (M H)/z peaks corresponding to the molecular ion, referred to as molecular weight (MW). Typical peak analysis data for the degradation products of iopromide at different radiation doses and the retention times and relative peak intensities for each product of all ICMs detected in this study are provided in Figs. S2eS7, Table S1, and Figs. S8eS12 of the Supplementary Material, respectively. Postulated structures for the degradation products identified from g-irradiation of diatrizoate (MW 614) are illustrated in Fig. 2. The product with MW 586 was characterized, corresponding to the loss of 28 mass units to the parent compound.
O
O
O
OH
HO I
I
O
I O
N H
N H I
OH
I
O
I O
N H
N H I
- CO2
I
O
O N H
N H I
Scheme 1 e The ipso attack of the OH at the formate moiety in the parent compound resulted in the phenolic product with a MW 586.
4395
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 1 e4 3 9 8
O
OH
H N
O I
OH I
OH
H N
N OH
I
O OH
O
O
O OH
OH I H N
OH
OH
OH
O OH
OH OH
H N
OH
OH
OH
OH
H N
OH
HO O
OH OH
O N OH
OH
H N
OH
OH
H N
N
O
OH
OH
OH
OH
O
OH OH
OH OH
O
OH
OH OH
O
(MW. 555)
H N OH
H N
OH
O
OH
N
OH I
OH
HO O
OH
OH
O
N
O
H N
O
O
O
H N
HO O
(MW. 557)
OH
H N
N
(MW. 663)
H N
OH
OH I
OH
I
OH
O
OH
I
OH
N
O
H N
O
O
O
H N
O
(MW. 541)
O
OH
O
H N
O
(MW. 665)
H N
OH
H N
OH
N
O
I
OH
I
O OH
I
OH
N
O
H N
O
O
H N
O
(MW. 667)
I
OH
I
OH
N
OH
H N
N
(MW. 651)
O
OH
O
O
O
H N
OH
(MW. 775)
OH
H N
O
I
OH
N OH
H N
OH
I
OH I
OH
H N
I
I N
Iomeprol (MW. 777)
O
O
H N
O
O
H N
OH
HO O
OH OH
H N
N
OH
OH
O
OH
(MW. 431)
(MW. 399)
OH OH H N
N
OH
O
HO O
OH
O
H N
O
OH
OH OH
O
(MW. 445)
Fig. 3 e Proposed degradation pathway of iomeprol (Red arrow: the addition of OH at the iodo site, Blue arrow: dissociative electron attachment yielding a deiodinated product, Black arrow: the addition of OH to the deiodinated product, Green arrow: H abstraction from the side chains forming a ketone product). For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.
The plausible reaction is the ipso addition of OH at the formate moiety with subsequent decarboxylation (eHCOOH), giving a phenolic product with MW 586 as shown in Scheme 1. Structural assignments made in Scheme 1 were confirmed by the mass spectral fragmentation patterns as shown in the Supplementary Material, Fig. S13. The possibility of such oxidative decarboxylation via OH has been reported by monitoring 14CO2 production from [7-14C] benzoate with Fenton’s reagents for producing OH (Gary and Arthur, 1982). Fig. 3 summarizes the postulated structures for the degradation products identified from g-irradiation of iomeprol (MW 777) as one of the four non-ionic ICM employed in this study. In total, there are four parallel reaction pathways represented by different colored arrows illustrated in Fig. 3: (a) ipso addition of OH at the iodo site which resulted in the loss of the iodine atom, forming a phenolic product (red arrow). This pathway has been described for diatrizoate in the Scheme 2, (b) the dissociative electron attachment yielding a deiodinated product (blue arrow) as described in Scheme 3 Scheme 4 Scheme 5. For interpretation of the references to
R I
OH transient spectra
To better understand the stability of the intermediate radical species from reaction of OH with the five ICM in neutral solution (pH 7), transient spectra were measured in the wavelength range of 250e500 nm (Fig. S17 of Supplementary
OH N H
R
R
OH I
O
I
3.3.
R I
O N H
colour in this figure legend, the reader is referred to the web version of this article. Postulated structures for the degradation products of other three ICM (iohexol, iopromide, and iopamidol) are illustrated in Figs. S14eS16 of Supplementary Material, which is in agreement with the result of previous study showing very similar spectrum of degradation products with deiodination being a prominent route in the photodegradation of iopromide (Pe´rez et al., 2009). Overall, the degradation pathways proposed based on the products identified by the analyses of LC-MS were similar to that of iomeprol shown in Fig. 3 despite the minor differences in the degradation are found according to those side chains.
I
O
I O
- HI
I
O
O
O
+ H2O
OH
O
O
- OH N H
N H I
N H
N H I
N H
N H I
Scheme 2 e The product at MW 476 resulting from the combination of Scheme 1, and ipso attack at one of the iodine sights, with further oxidation by OH, leading to the formation of quinone products with MW 474.
4396
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 1 e4 3 9 8
Scheme 3 e (c) The addition of OH to the deiodinated product resulting in a phenolic compound (black arrow), as shown in Scheme 4.
Material). The absorption spectrum recorded in the reaction of OH with diatrizoate exhibited three peaks centered at 280, 360, and 410 nm, respectively. The time-dependence of transient spectra suggests an initial increase from 5 to 10 ms, with subsequent decay with no apparent intra-molecular reaction occurring (Fig. S18 of Supplementary Material). The spectrum for iohexol had two absorption peaks at 285 and 400 nm. The other three had small peaks at 265 nm with a broad absorption band between 360 and 450 nm. The molar absorptivity at lmax (emax) for the five ICMs were obtained, using a G-value of 0.59 mmol J1 for the OH at the time of the respective maximum absorbances, and listed in Table 1 (Laverne and Pimblott, 1993).
3.4.
Kinetic measurements
To support the data on destruction mechanisms, the absolute bimolecular reaction rate constants for OH reaction with the five ICM and several analogues of diatrizoate were determined from the buildup of the maximum transient absorption at the lmax. Typical kinetic data for diatrizoate are shown in Fig. S19 of Supplementary Material. The formation of intermediates exhibited first order kinetics and the rate (kobs) increased linearly with the concentration, 0.28e1.0 mM (Fig. S19(B) of Supplementary Material). The absolute OH rate constants (kOH ) were obtained from the slope of the linear plot for the pseudo-first-order rate constants (kobs) as a function of concentration of the compound employed. The kOH values obtained for the five ICMs and several analogues of diatrizoate used in this study are listed in Tables 1 and 2, respectively. The rate constant for OH reaction with diatrizoate, the simplest of the five ICM tested was (9.58 0.23) 108 M1 s1, which is slightly higher than that reported previously (7 108 M1 s1) (Quintiliani et al., 1979). To access the site of reaction of OH with diatrizoate, the rate constant for OH reaction with diatrizoate was compared with that for several model compounds, which were determined in this study or were reported elsewhere (Table 2). The reaction rate of OH with 3, 5-bis-acetamino-benzoic acid (6.0 109 M1 s1) was
approximately six times faster than that with diatrizoate. This indicates that the presence of iodide depressed the OH reaction, which is supported by the fact that the reaction of OH with iodopentafluoro benzene (1.1 109 M1 s1) has been reported to be slower than that with pentafluoro benzene (4 109 M1 s1) (Koester and Asmus, 1973; Mohan and Mittal, 1995). This is consistent with our assignment that the initial product was formed by ipso attack of OH at the iodide site in the structure of diatrizoate. The effect of the acetamide moiety on the reaction of OH with diatrizoate was evaluated by comparing the rate constants for three model compounds with a different number of acetamide group in the structure, 3,5-bis-acetamino-benzoic acid, 3-acetaminobenzoic acid, and benzoic acid. No significant difference in the rate constants for these compounds (5.4e6.0 109 M1 s1) suggests that the site of acetamide in diatrizoate is not involved in the reaction with OH and the substitution of acetamide group has no significant effect on the electron density in the ring system. The rate constants for OH reaction with four non-ionic ICMs ranged from 2.3 to 3.4 109 M1 s1, consistent with the value reported for iopromide ((3.3 0.6) 109 M1 s1), as measured by competition kinetics under steady-state g-radiolysis (Huber et al., 2003). The faster reaction of OH with four non-ionic ICMs, relative to diatrizoate is presumably attributed to the presence of side chains providing additional sites for OH attack. The rate constants for eaq reaction with five ICMs and several analogues of diatrizoate selected in this study were measured by directly monitoring the change in the absorption of eaq at 700 nm in N2-saturated solution at pH 7.0 as shown in Fig. S20 of Supplementary Material for diatrizoate. The decay curves were fitted to pseudo-first-order exponential kinetics, giving the second-order linear plot shown in Fig. S20 (B) of Supplementary Material. The slope of such a plot is the second-order rate constant for eaq reduction of diatrizoate. The bimolecular reaction rate constants for eaq with the five ICM and several analogues of diatrizoate selected in this study are summarized in Tables 1 and 2, respectively. The absolute reaction rate constant for eaq with diatrizoate was (2.13 0.03) 1010 M1 s1, consistent with the value reported previously (2.1 1010 M1 s1) (Quintiliani et al., 1979). The average values for the reaction of eaq with the four non-ionic ICM were very similar and ranged from 3.3 to 3.8 1010 M1 s1. The reaction rate of eaq with diatrizoate determined in this study was considerably faster than that with the non-iodinated analogues of diatrizoate listed in Table 2 such as 3,5-bis-acetamino-benzoic acid (4.9 109 M1 s1), 3-acetaminobenzoic acid (3.8 109 M1 s1), benzoic acid
Scheme 4 e (d) Hydrogen abstraction from the side chains with OH forming a carbon centered radical, with subsequent addition of O2 to produce peroxyl radical, which undergoes intra-molecular rearrangement to eliminate OL 2 /HO2 to a ketone product as shown in Scheme 5 (green arrow).For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.
4397
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 1 e4 3 9 8
OH
OH
H N
O
OH
OH
H N
O
R
OH
O2
H N
O
R
H
O
O
O OH
- HO2
R
O H N
O
OH
R
Scheme 5 e The degradation products of iomeprol identified by molecular weight in Fig. 3 may result from sequential combinations of the four main reaction pathways suggested above.
(3.5 109 M1 s1), and acetanilide (5.0 108 M1 s1), and comparable with the rate constant determined for 2,3,5triiodobenzoic acid (2.5 1010 M1 s1) which contains the same number of iodide as diatrizoate, but at different positions on the aromatic ring. These results suggest dissociative electron attachment of the iodide, which is further supported by the fact that the rate constants for the single iodosubstituted aromatic rings such as 4-iodobenzoic acid
Table 2 e Rate constants for OH and eLaq reaction with the analogues of diatrizoate. Compounds
Chemical structure
Reaction rate constant (M1s1) OH
eaq
6.0109 (this study)
4.9109 (this study)
9.7109 (this study)
2.51010 (this study)
5.4109 (this study)
3.8109 (this study)
5.7 109 (pH 7)a
3.5 109 (pH 7)a
5.2 109 (pH 9)a
5.0108 (this study)
2.5 109 (pH 9)a
9.1 109 (pH 11)a
5.0 109 (pH 9)a
1.2 1010 (pH 11)a
O
3,5-bis-acetaminobenzoic acid
OH
O
O N H
N H
O
2,3,5-triiodobenzoic acid
OH I
I
O
I
OH O
3-acetamino-benzoic acid
N H
OH
O N H
Acetanilide
HO I
4-iodobenzoic acid
O
I
Iodobenzene
a Buxton et al., 1988.
3.5.
Degradation efficiency
In order to better understand the treatment of ICMs by OH and eaq, estimates of the reaction (destruction) efficiency for both reactive species were obtained using g-irradiation, as described elsewhere (Mezyk et al., 2007; Song et al., 2009). The concentration of each reactive species produced during g-irradiation was calculated based on the G-values in Eq. (1) and the kinetic rate constants for each reactive species obtained and listed in Table 1. The details on the calculation of the individual degradation efficiencies for the reaction of OH and eaq with the ICM are described in Text S3 and Fig. S21 of Supplementary Material. Table 1 summarizes the degradation efficiencies of OH and eaq for the five ICM. The efficiencies for OH reaction ranged from 68 to 79% except for diatrizoate which was significantly lower (40%) than the four non-ionic ICM. This is consistent with the results of our kinetic measurements, where the reaction rate constants for OH with the four non-ionic ICMs are approximately 2e3 times faster than that with diatrizoate, attributed to the presence of bulky side chains in non-ionic ICMs that are the additional site of OH attack. The efficiencies of eaq reaction with the five ICM ranged from 69 to 87%. The higher efficiency of eaq reaction with the four non-ionic ICMs relative to that of diatrizoate suggests the presence of side chains serves to stabilize the transition state and accelerated the degradation of non-ionic ICMs.
4.
O
Benzoic acid
(9.1 109 M1 s1) and iodobenzene (1.2 1010 M1 s1) are slower than those for the triply iodo-substituted compounds determined in this study.
Conclusions
In this study, it appears that both oxidative and reductive processes result in the removal (destruction) of these ICMs. However, the eventual application of advanced oxidation/ reduction processes in ‘real-world’ waters which can contain high levels of dissolved organic matter (DOM) and carbonate alkalinity must be evaluated. The degradation mechanisms of the ICM are extremely complex and additional studies will have to be conducted to complete the mechanisms reported here and to evaluate the potential of these by-products to elicit biological activity. However, the loss of iodine from the parent compounds should result in more biodegradable products during treatment. Studies should also be conducted on these mixtures to see if they are either biologically degraded or if sunlight induced photolysis (direct or indirect) are important in their environmental fate.
4398
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 9 1 e4 3 9 8
Acknowledgements This work was performed at the Radiation Laboratory, University of Notre Dame, which is supported by the Office of Basic Energy Science, U.S. Department of Energy. Financial support for J. Jeong is also acknowledged from the Korea Research Foundation Grant funded by the Korean Government (MOEHRD) (KRF-2007-357-D00146) and J. Jung thanks the Korean Institute of Science and Technology for their support. This is contribution 53 from the Urban Water Research Center, University of California, Irvine.
Appendix. Supplementary data Supplementary data associated with this article can be found in the online version at doi:10.1016/j.watres.2010.05.054.
references
Buxton, G.V., Greenstock, C.L., Helman, W.P., Ross, A.B., 1988. Critical review of rate constants for reactions of hydrated electrons, hydrogen atom, and hydroxyl radicals (OH/O) in aqueous solutions. J. Phys. Chem. Ref. Data 17, 513e886. Buxton, G.V., Stuart, C.R., 1995. Re-evaluation of the thiocyanate dosimeter for pulse radiolysis. J. Chem. Soc. Faraday, Trans. 91, 279e281. Daughton, C.G., Ternes, T., 1999. Pharmaceuticals and personal care products in the environment: agents of subtle change? Environ. Health Perspect. 107, 907e938. Doll, T.E., Frimmel, F.H., 2004. Kinetic study of photocatalytic degradation of carbamazepine, clofibric acid, iomeprol and iopromide assisted by different TiO2 materialsedetermination of intermediates and reaction pathways. Water Res. 38 (4), 955e964. Fent, K., Weston, A., Caminada, D., 2006. Ecotoxicology of human pharmaceuticals. Aquat. Toxicol. 76, 122e159. Gary, W.W., Arthur, I.C., 1982. Oxidative decarboxylation of benzoate to carbon dioxide by rat liver microsomes: a probe for oxygen radical production during microsomal electron transfer. Biochemistry 21, 4265e4270. Halling-Sorensen, B., Nielsen, S.N., Lanzky, P.F., Ingerslev, F., Lutzhft, H.C.H., Jorgensen, S.E., 1998. Occurrence, fate and effects of pharmaceutical substances in the environmenteA review. Chemosphere 36 (2), 357e393. Huber, M.M., Canonica, S., Park, G., von Gunten, U., 2003. Oxidation of pharmaceuticals during ozonation and advanced oxidation processes. Environ. Sci. Technol. 37, 1016e1024. Huber, M.M., Go¨bel, A., Joss, A., Hermann, N., Lo¨ffler, D., McArdell, C.S., Ried, A., Siegrist, H., Ternes, T., von Gunten, U., 2005. Oxidation of pharmaceuticals during ozonation of municipal wastewater effluent: a pilot study. Environ. Sci. Technol. 39, 4290e4299. Ku¨mmerer, K., 2004. Pharmaceuticals in the Environment: Sources, Fate, Effects and Risks. Springer, Berlin. Ku¨mmerer, K., Erbe, T., Gartiser, S., Brinker, L., 1998. AOXeemissions from hospitals into municipal waste water. Chemosphere 36, 2437e2445. Khetan, S.K., Collins, T.J., 2007. Human pharmaceuticals in the aquatic environment: a challenge to green chemistry. Chem. Rev. 107, 2319e2364.
Koester, R., Asmus, K.D., 1973. Reaction of fluorinated benzenes with hydrated electrons and hydroxyl radicals in aqueous solutions. J. Phys. Chem. 77, 749e755. Laverne, J.A., Pimblott, S.M., 1993. Yields of hydroxyl radical and hydrated electron scavenging reactions in aqueous solutions of biological interest. Radiat. Res. 135, 16e23. Mezyk, S.P., Neubauer, T.J., Cooper, W.J., Peller, J.R., 2007. Freeradical-induced oxidative and reductive degradation of sulfa drugs in water: absolute kinetics and efficiencies of hydroxyl radical and hydrated electron reactions. J. Phys. Chem. A 111, 9019e9024. Mohan, H., Mittal, J.P., 1995. Formation and redox properties of radical ions of iodopentafluorobenzene in aqueous solution: a pulse radiolysis study. J. Phys. Chem. 99, 12559e12564. Pe´rez, S., Barcelo´, D., 2007. Fate and occurrence of X-ray contrast media in the environment. Anal. Bioanal. Chem. 387, 1235e1246. Pe´rez, S., Eichhorn, P., Ceballos, V., Barcelo´, D., 2009. Elucidation of phototransformation reactions of the X-ray contrast medium iopromide under simulated solar radiation using UPLC-ESI-QqTOF-MS. J. Mass. Spectrom. 44, 1308e1317. Putschew, A., Schittko, S., Jekel, M., 2001. Quantification of triiodinated benzene derivatives and X-ray contrast media in water samples by liquid chromatography-electronspray tandem mass spectrometry. J. Chromatogr. A930, 127e134. Putschew, A., Wischnack, S., Jekel, M., 2000. Occurrence of triiodinated X-ray contrast agents in the aquatic environment. Sci. Total Environ. 255, 129e134. Quintiliani, M., Betto, P., Davies, J.V., Ebert, M., 1979. Radiation studies of iodinated benzoic acids. Radiat. Biol. Chem. Res. Dev. 6, 49e57. Rosenfeldt, E.J., Linden, K.G., 2004. Degradation of endocrine disrupting chemicals bisphenol A, ethinyl estradiol, and estradiol during UV photolysis and advanced oxidation processes. Environ. Sci. Technol. 38, 5476e5483. Sacher, F., Lange, F.T., Brauch, H.J., Blankenhorn, I., 2001. Pharmaceuticals in groundwaterseanalytical methods and results of a monitoring program in Baden-Wu¨rttemberg. Germany. J. Chromatogr. A938, 199e210. Seitz, W., Jiang, J., Schulz, W., Weber, W.H., Maier, D., Maier, M., 2008. Formation of oxidation by-products of the iodinated X-ray contrast medium iomeprol during ozonation. Chemosphere 70, 1238e1246. Song, W., Chen, W., Cooper, W.J., Greaves, J., Miller, G.E., 2008a. Free-radical destruction of b-lactam antibiotics in aqueous solution. J. Phys. Chem. A 112, 7411e7417. Song, W., Cooper, W.J., Mezyk, S.P., Greaves, J., Peake, B.M., 2008b. Free radical destruction of b-blockers in aqueous solution. Environ. Sci. Technol. 42 (4), 1256e1261. Song, W., Cooper, W.J., Peake, B.M., Mezyk, S.P., Nickelsen, M.G., O’Shea, K.E., 2009. Free-radical-induced oxidative and reductive degradation of N, N’-diethyl-m-toluamide (DEET): kinetic studies and degradation pathway. Water Res. 43 (3), 635e642. Spinks, J.W.T., Woods, R.J., 1964. An Introduction to Radiation Chemistry. John Wiley & Sons, NY. Ternes, T., Hirsch, R., 2000. Occurrence and behaviour of X-ray contrast media in sewage facilities and the aquatic environment. Environ. Sci. Technol. 34, 2741e2748. Ternes, T.A., Stu¨ber, J., Hermann, N., McDowell, D., Ried, A., Kampmann, M., Teiser, B., 2003. Ozonation: a tool for removal of pharmaceuticals, contrast media and musk fragrances from wastewater? Water Res. 37 (8), 1976e1982. Whitham, K., Lyone, S., Miller, R., Nett, D., Treas, P., Zante, A., Fessenden, R.W., Thomas, M.D., Wang, Y., 1996. IEEE Proceedings Particle Accelerator Conference and International Conference on High Energy Accelerators. Laurie Gennari Publishers, Dallas, TX (EEE Operations Center).
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 6 3 e4 4 7 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Development and application of a method for quantifying factors affecting chloramine decay in service reservoirs Arumugam Sathasivan a,*, K.C. Bal Krishna a, Ian Fisher b a
Department of Civil and Construction Engineering, Curtin University of Technology, GPO Box U1987, Perth, Western Australia 6845, Australia b Watervale Systems Pty. Ltd. PO Box 318, Potts Point NSW 1335, Australia
article info
abstract
Article history:
Service reservoirs play an important role in maintaining water quality in distribution
Received 3 July 2009
systems. Several factors affect the reservoir water quality, including bulk water reactions,
Received in revised form
stratification, sediment accumulation and wall reactions. It is generally thought that biofilm
21 May 2010
and sediments can harbour microorganisms, especially in chloraminated reservoirs, but
Accepted 6 June 2010
their impact on disinfectant loss on disinfectant loss has not been quantified. Hence, debate
Available online 12 June 2010
exists as to the extent of the problem. To quantify the impact, the reservoir acceleration factor (FRa) is defined. This factor represents the acceleration of chloramine decay arising
Keywords:
from all causes, including changes in retention time, assuming that the reservoir is
Biofilm
completely mixed. Such an approach quantifies the impact of factors, other than chemical
Chloramine decay
reactions, in the bulk water. Data from three full-scale chloraminated service reservoirs in
Nitrification
distribution systems of Sydney, Australia, were analysed to demonstrate the generality of the
Microbial decay factor
method. Results showed that in two large service reservoirs (404 103 m3 and 82 103 m3)
Reservoir
there was minimal impact from biofilm/sediment. However, in a small reservoir (3 103 m3),
Stratification
the biofilm/sediment had significant impact. In both small and large reservoirs, the effect of stratification was significant. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Chloramine (NH2Cl) is often used in water distribution systems where a longer lasting residual is desired or a reduction in the formation of chlorinated disinfection by-products (Brodtmann and Russo, 1979; Cotruvo, 1981). However, the stability of chloramine presents some additional challenges for water utilities. In addition to chemical reactions with waterborne constituents, microbes (including nitrifiers) accelerate chloramine decay and promote bacterial regrowth. Microbial decay is due to nitrification or other microbial processes that can accelerate chloramine decay by an order of magnitude compared to chemical decay. Nitrification is a two-
step microbiological process: ammonia (NH3) is initially oxidised to nitrite (NO 2 ) by ammonia-oxidising bacteria (AOB), and NO2 is then oxidised to nitrate (NO3 ) by nitrite-oxidising bacteria (NOB). In order to understand and quantify the roles of microbes and chemical reactions separately in a chloraminated distribution system, Sathasivan et al. (2005) proposed a simple method, the microbial decay factor, Fm. The Fm method is more quantitative, sensitive and general than the traditional indicators (e.g. NO 2 , NH3, total chlorine), in quantifying the microbial contribution to chloramine decay. The Fm method was shown to be helpful in maintaining an adequate chloramine residual by minimizing microbial acceleration of chloramine decay (Sathasivan et al., 2010).
* Corresponding author. E-mail address:
[email protected] (A. Sathasivan). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.009
4464
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 6 3 e4 4 7 2
Nomenclature kc kc,T kbc,T km km,T Fm Fm,T FRa
chemical decay coefficient at 20 C, h1 chemical decay coefficient at T C, h1 “base” chemical decay coefficient at T C, h1 microbial decay coefficient at 20 C, h1 microbial decay coefficient at T C, h1 microbial decay factor at 20 C, (km/kc) microbial decay factor at T C (km,T/kc,T) reservoir acceleration factor
Service reservoirs (tanks) play a critical role in water quality management, because water spends most of its time in reservoirs, rather than in pipes. In reservoirs, many factors contribute to such problems: water retention time, mixing regime, stratification, biofilm bacterial activity, bulk water bacterial activity and sediment presence. If a reservoir is not properly mixed, stratification is one of many factors contributing to nitrification in summer. Such stratification ensures a longer retention time for the near-surface water and enhances the chemical and microbial chloramine decay, thereby decreasing the chloramine residuals substantially in surface layers. Ike et al. (1988) showed that, in summer, AOB concentrations were 10e20 times higher in the surface layer (0.3 m below surface) compared with 5 m below the surface. In addition, Barrios (1989) and Burlingame and Brock (1985) reported the presence of microbial stratification during summer in reservoirs. Using the Fm method, Fisher et al. (2009) found microbial stratification in two large reservoirs (capacity 404 103 m3 and 82 103 m3) in Sydney’s water distribution system in winter, despite their chemical and thermal homogeneity. In summer and autumn, the reservoirs were found to be thermally, chemically and microbialy stratified. Biofilm bacterial activity also depends on the geometry of the reservoir; i.e. wall surface (summation of vertical walls and bottom area) to bulk water volume ratio (S/V). Many researchers have reported higher numbers of AOB present in biofilm and sediment rather than in bulk water. Biofilms and sediments enhance the growth of AOB, as well as provides protection from disinfectant. Ike et al. (1988) detected higher numbers of AOB per unit area in sediment than in biofilm. Stewart and Lieu (1997) sampled from a 10.45 103 m3 reservoir in Southern California and observed that, although AOB were not detected in the water column, significant levels of AOB were present in the wall biofilm layer. The AOB concentrations on biofilm surfaces ranged from a low 11 at the middle level to a high 860 MPN/cm2 at the bottom level. The sludge sample also had a high level of AOB of 4000 MPN/mg. Baribeau et al. (2001) determined the concentrations of AOB in the bulk water samples, which typically varied between 0 and 10 MPN/ml, whereas in the biofilm in the bottom layer of the reservoir, AOB levels were more than 20e200 MPN/cm2. A recent study (Srinivasan and Harrington, 2007) quantified the relative abundance of bacteria in both biofilm and bulk water in chlorinated pilot scale pipe loops. They concluded that the relative bacterial abundance in the bulk water (compared with biofilm) increased when residual decreased.
kRt kt kt,T E R S/V q TCl
total decay coefficient at T C, h1 summation of kc and km, h1 summation of kc,T and km,T, h1 activation energy, J/Mol universal gas constant, 8.314 J/Mol/K wall surface (summation of vertical walls and bottom area) to bulk water volume ratio (m1) reservoir retention time, day total chlorine, mg/L
Despite such detailed studies, none has quantified the impact of biofilm and sediments on chloramine decay. It is necessary to understand the contribution of sediment presence, biofilm bacterial activity and stratification for modelling and management purposes. In this paper, a simple approach to quantifying the effects of these niches of microorganisms on chloramine residual is developed. To demonstrate the generality of the method, three full-scale system reservoirs were studied in detail.
2.
Concept development
In many delivery systems, water spends the majority of its travel time in service reservoirs. Therefore, reservoirs are critical for maintaining chloramine residual. The chloramine decay rate in reservoirs can be influenced by several factors; temperature, mixing regime, stratification, biofilm bacterial activity, bulk water bacterial activity and sediment presence. Fig. 1 illustrates the ways that stratification, sediment and biofilm bacterial activities can affect chloramine residual in a reservoir. Details of the processes occurring in different niches (bulk water, sediment and biofilm) are also provided in the Figure. If retention time, and inlet and outlet chloramine residuals are known, the total chloramine decay coefficient of reservoir water can be estimated from the following Equation, assuming that the reservoir is completely mixed. TClout ¼
TClin ð1 þ kRt qÞ
(1)
where, TClout and TClin are outlet and inlet chloramine residual (as total chlorine) respectively, kRt is the total chloramine decay coefficient and q is the retention time in the reservoir. Average retention time can be estimated by various means and inlet and outlet residual can be easily measured using either online chlorine analysers or grab sample measurements. Then, the only unknown value, kRt, can be estimated. Because all the parameters are measured at reservoir water temperature, kRt defines the total decay coefficient of the reservoir contents at that temperature. Sathasivan et al. (2008) reported that, in most mildly nitrifying (NO2eN < 0.010 mg/L) Sydney’s water samples, the average chemical decay coefficient (kc) was 0.0015 h1 at 20 C. The variation was only 0.0002 h1, when measured at 20 C.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 6 3 e4 4 7 2
4465
Fig. 1 e Conceptual representation of factors affecting chloramine residual. X is the bacterial number and subscripts b, w, s and se represented bulk water, biofilm (wall), stratification and sediment, respectively.
Consequently, in this case, kc at 20 C is taken as 0.0015 h1 and termed the “base” chemical decay coefficient, kbc,20. Temperature variation of chemical (kc) and microbial decay coefficients (km) is assumed to follow the Arrhenius equation shown in Equation (2) (Sathasivan et al., 2009). E 1 1 kT ¼ k20 exp R 273 þ T 273 þ 20
(2)
where T C is temperature, and E/R values are 3551 and 6924 K1 for kc and km respectively (Sathasivan et al., 2009). If Fm of a sample (thus km and kc) is known from in-reservoir or outlet samples, it is possible to estimate the chemical and microbial decay coefficients at reservoir water temperature using Equation (2). If these are termed kc,T and km,T , then the total bulk water decay coefficient (kt,T) at reservoir water temperature can be estimated by adding both microbial and chemical decay coefficients. Within a reservoir several factors can accelerate chloramine decay, sometimes well above the inlet decay rates. Tracking the acceleration present within a reservoir can provide system managers with the dynamic status of each reservoir, on which they can base and assess remedial action. As total inlet decay rates can vary greatly, comparison between reservoirs to prioritize remedial actions is not straightforward. Reservoir status can be standardised against a particular decay coefficient by comparing the kRt of a reservoir with the “base” chemical decay coefficient (kbc) converted to reservoir water temperature using Equation (2) to yield kbc,T. Then, FRa, the reservoir acceleration factor, is defined with respect to kbc,T as shown in Equation (3). FRa ¼
kRt kbc;T kbc;T
(3)
FRa represents the degree of acceleration present in the reservoir with respect to kbc,T (0.0015 h1 at a temperature of 20 C) measured in mildly nitrifying samples. This value is valid for Sydney water samples (Sathasivan et al., 2008), but it is likely to be different in different systems. Contributions from factors other than bulk water contents to acceleration of chloramine decay in a reservoir can be
understood by comparing the total decay coefficient in the bulk water (kt,T) with that in the reservoir as a whole (kRt) at temperature T. If the reservoir is clean (i.e. no biofilm or sediment present) and there is no stratification, the difference between kt,T and kRt would be theoretically zero and chloramine decay would only be due to bulk water reactions. In other words, if chemical decay coefficients measured in bulk water samples are the same as the “base” chemical decay coefficient, then FRa and Fm,T should be directly compararable.
3.
Materials and methods
3.1.
General description of water distribution system
Sydney Water Corporation supplies water to about 4 million customers. Raw water is taken mainly from large surface water storages. Water is treated by coagulation/flocculation/ direct filtration, followed by chlorination and ammoniation. The finished water quality before entering the distribution systems typically meets the following criteria: turbidity less than 0.1 NTU, chloramine (as total chlorine) 1.5e1.7 mg/L, dissolved organic carbon 3.3 0.4 mg/L, pH 8.0 0.2 and a chlorine to NH3 ratio about 4:1. Water temperature in these systems varies between 12 and 25 C, seasonally. Nitrification is usually noticed during the summer months (December to March) and later in April and May (autumn).
3.1.1.
Description of reservoirs
This study analyses data from three different reservoirs in Sydney’s distribution systems. These reservoirs are named Reservoir A, B and C. All three reservoirs contained common inlet and outlet structures. Their size varied from 404 103 to 3 103 m3. None of the reservoirs had mechanical mixers installed. Reservoir A (Fig. 2) is the first reservoir in one of the major distribution systems in Sydney, Australia. It is located about 12 km downstream of the initial chloramination point and is a buffer tank for many reservoirs downstream. Its maximum
4466
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 6 3 e4 4 7 2
of the lower long wall, as shown in Fig. 2. Using the same procedure described for Reservoir A, the S/V ratio was calculated to be 0.2 m1. Reservoir C, located in one of Sydney’s many distribution subsystems, is an elevated circular tank of diameter 18 m, height 13 m and capacity of 3 103 m3. This reservoir receives water from a surface water source and treatment plant different from those supplying Reservoirs A and B. As the reservoir is situated in a bushfire prone area, the highest possible water level must be maintained in it during summer. Water is received directly from a relatively clean upstream reservoir. Retention time in the reservoir during the sampling period was 2.4e2.9 days. The reservoir has a common inlet/ outlet pipe of 0.45 m diameter, opening 1.0 m above the bottom and at the reservoir centre. Its S/V ratio was 0.31 m1. Mixing in this reservoir is achieved only by inflow.
176 m
Reservoir A Common Inlet/outlet
176 m
Sampling point
Common inlet/out
61.0 m
48.75 m
48.75 m
S1
S3
97.5 m
3.2.
Reservoir B S2
N
Common Inlets/outlets
Fig. 2 e Detailed top view of Reservoir A and B with sampling points. These two figures were adopted from Fisher et al. (2009).
depth is 13 m, its capacity is 404 103 m3 and it is completely covered by a metal roof. Although it has separate inlet and outlet structures, both were operating as common inlet/ outlets for the duration of the study. About two thirds of the flows passed through the inlet structure comprised of two horizontal pipes entering at the base of the reservoir wall. The remaining one third passed through the outlet, a smaller pipe that discharges vertically upwards from the bottom. The outlet opening was close to the middle of the reservoir (Fig. 2). Water temperature in the reservoir was just above 11 C in winter and up to 23 C near the surface in summer. An average depth of water over the period of testing was calculated by dividing the average reservoir water volume by the bottom area. Total surface area was calculated as the sum of side (wall) and bottom surfaces. The surface to volume ratio (S/V) could then be calculated by dividing total surface area by average operational volume. For example, for 70% of the maximum depth, the S/V value was 0.08 m1. Reservoir B is supplied from the same treatment plant as Reservoir A, but is fed from another reservoir along the route. Reservoir B has a capacity of 82 103 m3, with a multiple-arch roof covered by a layer of turf. At full supply, the water level is 6 m above the bottom and the minimum operating depth is equivalent to 65% of capacity. Common inlet/outlet structures are located in the reservoir floor, on either side of the midpoint
Sampling
Inlet/outlet chorine residuals, temperature and reservoir water level were monitored online in Reservoir B, but in-reservoir vertical sampling was conducted manually. In-reservoir samples were collected from the surface and mid-depth. Middepth samples were from 6 m below the water surface in Reservoir A and 3 m above the bottom in Reservoir B. Sampling from Reservoir A was collected from the sampling point shown in Fig. 2. Sampling from Reservoir B was carried out at three different sites (S1, S2 and S3); one site (S2) was located almost at the centre of the reservoir and the other two symmetrically as far away from the inlet as possible (Fig. 2). The common inlet/ outlet was located at the bottom but through the long wall opposite the sampling points S1 and S3. Water samples were taken by lowering a Niskin bottle (midpoint) to each designated depth. When taking surface samples, the sampler was just submerged so that the nominal (midpoint) depth was 0.2 m. Temperature and chloramine were measured at the time of sampling. Samples (except those for Fm analysis) were put on ice for transport to the laboratory. Decay tests were conducted at 20 C, when samples arrived at the laboratory. All other samples were refrigerated until scheduled for analysis of NO2eN, NO3eN and NH3eN. The details of measurements are given in Fisher et al. (2009). Similarly, water samples were collected manually from the inlet/outlet of Reservoir C and respective concentrations of total chlorine, NO2eN, NO3eN and NH3eN were analysed.
3.3.
Analytical methods
NO2eN, NO3eN and total NH3eN concentrations were measured by using the flow injection analysis method 4500 (Clesceri et al., 1998). NO2eN was measured by the sulphanilamide method (4500-NO 2 B). NO3eN was first reduced by the cadmium reduction method (4500-NO 3 E) to NO2eN and then NO2eN was measured by the sulphanilamide method. NO2eN had the lowest detection level of 0.002 mg/L. NH3eN was measured by the phenate method (4500-NH3 F). Total chlorine residuals were measured by the DPD colorimetric method using a HACH pocket colorimeter. Averages of two readings from duplicate samples were reported. Total chlorine measurement had an experimental error of 0.03 mg/L.
4467
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 6 3 e4 4 7 2
3.5.
20
1.5 o
16 1.2 0.9
12
0.6 Inlet total chlorine 0.3
8
In-reservoir total chlorine Temperature 28/05/05
17/02/05
09/11/04
01/08/04
23/04/04
14/01/04
06/10/03
4 28/06/03
0.0 20/03/03
Determining kc and km involved four major steps: sample preparation, incubation, monitoring chloramine decay, and estimating decay rate from the resulting data. Sample preparation involved splitting the sample into two sub-samples. The first sub-sample was not processed at all and the second sub-sample was processed to remove microbial activity by either aseptic filtration or addition of silver nitrate. Both subsamples were then subjected to incubation at a constant temperature of 20 C. For each sub-sample, total chlorine was monitored over time, from which the decay rate was estimated. Fm is the ratio of km to kc for a given water sample measured at 20 C. Full details of the method are given in Sathasivan et al. (2005). The km and kc values were converted to those at desired temperatures using Equation (2). The corresponding Fm parameter is termed Fm,T .
1.8
Temperature ( C)
Microbial decay factor (Fm) method Total Chlorine (mg/L)
3.4.
Date
Fig. 3 e Profiles of In-reservoir temperature and inlet/inreservoir residuals in Reservoir A.
Error estimation for FRa and Fm,T
The measurement variation was 0.0002 h1 for both kc, and km, at the 95% confidence level. The maximum Fm,T value was calculated by adding 0.0002 h1 to km,T and subtracting 0.0002 h1 from kc,T, whereas minimum Fm,T was calculated by subtracting 0.0002 h1 from km,T and adding 0.0002 h1 to kc,T. The difference between the average value and the maximum or minimum value of Fm,T was considered to be the Fm,T measurement error. Assuming that reservoir retention time did not change much and chlorine measurement error was 0.03 mg/L, FRa measurement error was determined. The maximum kRt value was calculated by adding 0.03 mg/L to inlet and subtracting 0.03 mg/L from outlet chlorine residual, in Equation (1). The minimum kRt value was calculated by subtracting 0.03 mg/L from inlet and adding the same value to outlet residual. The maximum and minimum FRa values were determined using the maximum and minimum values of kRt respectively, in Equation (3), keeping the kc,T as a converted value from 0.0015 h1 at 20 C. The difference between average FRa and maximum or minimum FRa was considered to be the FRa measurement error.
4.
Results and discussion
4.1.
Nitrification surrogate parameters in Reservoir A
Reservoir A is the second largest reservoir (404 103 m3) in the Sydney’s distribution system. It had been operated with a retention time between 2.8 and 4.5 days. The reservoir received water with consistent inlet total chlorine residuals of 1.48 0.03 mg/L. The inlet total chlorine to NH3eN ratio was maintained at about 4.5:1 to minimize the free NH3eN available for AOB growth, as well as to reduce the possibility of dichloramine formation. Inlet NO2eN concentration was below the detection limit (less than 0.002 mg/L), whereas NO3eN was 0.105 0.005 mg/L. There was no change in NOxeN (NO2eN þ NO3eN) observed between dosing point and inlet to the reservoir, indicating that there was no nitrifier source upstream of Reservoir A. The Fm value of the inlet
sample was consistently 0.2 0.2, indicating minimal microbial acceleration. It could be one of the reasons for consistent in-reservoir total chlorine residuals of 1.04e1.26 mg/L (Fig. 3), even over summer. Water temperature varied from 11.3 C (winter) to 18 C (summer). Low in-reservoir NO2eN concentrations (maximum about 0.003 mg/L in summer) showed that there was minimal nitrifier activity within the reservoir. Every week, there was one filling cycle from about 45 to 95% of total capacity. In addition, there were smaller filling cycles. Such weekly cycling was maintained throughout the study period. It provided adequate mixing of the bulk water and avoided the possibility of stratification. Repeated total chlorine, Fm and temperature measurements of in-reservoir samples taken from the surface and mid-depth confirmed minimal stratification in the reservoir (data not reported).
4.2. Reservoir acceleration factor (FRa) and microbial decay factor at temperature T (Fm,T) in Reservoir A In order to quantify the aggregate effect of bulk water, biofilm and sediment bacterial activities on chloramine loss, Reservoir A was selected as the first case, because the reservoir was shown to be well mixed (surface and mid samples showed similar chloramine and Fm for the duration of the study). The FRa and Fm,T values were determined as described in the concept development section. The FRa value was calculated using in-reservoir residuals collected at the mid-depth (6 m below surface) of the reservoir, and Fm was determined for the same samples. The inlet kc was within the experimental error for assumed kbc, 0.0015 h1 (20 C). Therefore, all calculations were made using 0.0015 h1 as the “base” chemical decay coefficient at 20 C. Calculated Fm,T and FRa values varied from 0.14 to 0.6 and from 0.2 to 0.8, respectively (Fig. 4). Usually higher FRa and Fm,T values than those shown in Fig. 4 were observed in summer, due to favourable conditions for microbial activity and higher decay rates at higher water temperatures. Calculated experimental FRa and Fm,T errors were in the range of 0.1 to 0.2 and 0.2 to 0.3, respectively. After considering experimental errors, measured FRa and Fm,T values fell within the same band of maximum and minimum
4468
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 6 3 e4 4 7 2
Fm,T
FRa
20
16
0.6 12 0.4
o
0.8
Temperature ( C)
1.0
8
0.2
08/04/05
29/12/04
20/09/04
12/06/04
04/03/04
25/11/03
17/08/03
4 09/05/03
0.0
Date
Fig. 4 e Fm,T, FRa and temperature profiles in Reservoir A. Calculated maximum and minimum Fm,T, errors were ±0.3 and ±0.2, respectively whereas maximum and minimum FRa errors were ±0.2 and ±0.1, respectively.
values. For example, on October 2004, at temperature of 18.0 C, the average Fm,T value was 0.6 0.2 (average error), i. e. maximum and minimum Fm,T values were 0.8 (0.6 þ 0.2) and 0.4 (0.6e0.2), respectively. On the same day, the estimated FRa value was 0.5 0.1 which falls between the maximum and minimum Fm,T values. Consequently, it was concluded that FRa and Fm,T values were equal, after considering experimental error. The reservoir was completely mixed due to its hydraulic operating conditions, as confirmed by repeated in-reservoir depth sampling. The chemical decay coefficients of in-reservoir samples were the same as the “base” chemical decay coefficient. Therefore, the difference between FRa and Fm,T or that between kRt and kt,T can be used to understand the additional role of biofilm and sediment bacterial activities. As found for FRa and Fm,T values, the differences between kRt and kt,T were within experimental/calculation errors. These results indicated that the majority of the decay in the reservoir was due to bulk water chemical and microbial activity. In other words, there was an insignificant contribution to accelerated chloramine decay from biofilm and sediments in Reservoir A in this two-year period.
Water D epth (m)
0.0
Total Chlorine (mg/L) 0.3 0.6 0.9
The FRa concept and related parameters were applied to Reservoir B, to show the effect of stratification phenomena on chloramine decay. This reservoir is also in Sydney’s water distribution system, receiving water from the same treatment plant and dosing station as Reservoir A. Water samples were collected from inlet, outlet and inreservoir and measurements of Fm related parameters, total chlorine and temperature were carried out on each. Inlet total chlorine, NH3eN, NO2eN were 1.0e1.5, 0.25 to 0.32 and 0.002 to 0.01 mg/L, respectively. Water retention time was 5e6 days during the sampling. In-reservoir samples were collected in late autumn (6th May) and early winter (8th June) across the depth from designated points (S1, S2 and S3) as shown in Fig. 2. Total chlorine and temperature measured vertically through the depth of the reservoir (Fig. 5) showed the existence of thermal and chemical stratification in autumn, whereas these values were reasonably uniform in winter. However, there was stratification in terms of Fm related parameters in winter (Table 1), as higher Fm and higher kt values were obtained in surface than in mid-depth samples. Fisher et al. (2009) previously reported similar behaviour, i.e. there was vertical stratification in terms of Fm related parameters despite uniformity of other parameters in winter. This phenomenon might be due to insufficient mixing strength of the inflow/thermal mixing in winter, even though it is sufficient to mix dissolved substances. Further, higher kc values were obtained in autumn samples compared to winter in-reservoir samples. Variation of in-reservoir kc values was mainly due to nitrification, as evidenced by high nitrite (NO2eN) concentrations (higher than 0.050 mg/L) in autumn samples and low nitrite concentration (lower than 0.010 mgN/L) at the beginning of winter (8th June 2004). In autumn, there was a large vertical variation in residual within the reservoir. It is difficult therefore to choose a single representative value. FRa values were initially calculated for Reservoir B using outlet residual and later using the average in-reservoir residual. Calculated FRa values using outlet residual varied between 1.1 and 5.7 (Fig. 6) between winter and autumn. By following FRa values, the operator would know that microbial activity was accelerating chloramine decay by almost six times the “base” chemical decay coefficient in o
1.2
14
0
0
1
1
2 Average total chlorine
3 4 6th May
8th June
Temperature ( C) 15 16
17
18
2 3 4 5
5 6
Water D epth (m)
Fm,T, FRa
4.3. Reservoir acceleration factor (FRa) and microbial decay factor at temperature, T (Fm,T) in Reservoir B
Temperature
8th June
6th May
6
Fig. 5 e Total chlorine and temperature profiles along the depth in Reservoir B. Data is presented only for sampling point S1 because data from S2 and S3 are similar (Note: Filled triangle with dotted line is the averaged in-reservoir total chlorine residual).
4469
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 6 3 e4 4 7 2
Table 1 e Inlet and in-reservoir Fm related parameter results in Reservoir B. kt (h1)
Date
Sampling point
6th May 2004
Inlet 0.2 m BSa 3.0 m ABa Outflow
0.0027 0.0208 0.0107 0.0110
8th June 2004
Inlet 0.2 m BS 3.0 m AB
0.0023 0.0001 0.0038 0.0002 0.0029 0.0001
kc (h1)
0.00016 0.0008 0.0005 0.0005
0.0015 0.0059 0.0046 0.0045
0.00015 0.0005 0.0003 0.0004
0.0015 0.0001 0.0016 0.0001 0.0015 0.0001
km (h1)
Fm
0.0003 0.0013 0.0008 0.0009
0.8 2.7 1.3 1.4
0.0008 0.0002 0.0022 0.0003 0.0014 0.0002
0.5 1.4 0.9
0.0012 0.0157 0.0061 0.0065
a BS refers to below surface and AB refers to above bottom (these results are average of all S1, S2 and S3 samples).
Fm,T (0.2m BS)
Fm,T (3.0m AB)
Temperature
6.0
16
5.0
15
4.0
14
Stratified
3.0
De-stratified
13
S2
2.0
12
S1
S1 S1, S3
1.0
S1, S2, S3
S2, S3
inlet kc value was 0.0015 h1 (Table 1) in autumn and winter, which is the same as that in Reservoir A: i.e. kbc. However, kc values for in-reservoir samples were almost triple to quadruple kbc (Table 1). The Fm value is determined by dividing by kc of the sample. Because these two (kbc and kc)are different for severely nitrifying samples, such as those from Reservoir B in autumn, other parameters need to be compared. The direct comparison of kRt and kt,T is useful to determine the contribution from different niches for these cases. On 6th May 2004, variable kt,T values were found in mid-depth and surface water samples (Table 2). To account for the variability within the reservoir, the kt,T values of mid-depth samples were averaged. The average value (0.0078 0.0007 h1) calculated from Table 2 was compared with the kRt value of 0.0125 0.0016 h1. Clearly, these two values were significantly different. However, it should be noted that there was a dilemma as to which residual to use for kRt and FRa calculations, particularly as there was a difference between age of water in different layers and because mixing can only occur upwards from the bottom inflow. Therefore, it is difficult to conclude that there was significant impact from biofilm and/ or sediment using autumn results. However, since there was better uniformity within the reservoir in winter, the analysis is more definitive. On 8th June 2004, kRt (0.0017 0.0005 h1) and average kt,T (0.0019 0.0002 h1) for samples collected from mid-depth at various sites in the reservoir were not statistically different. Therefore, there was an insignificant contribution from biofilm/sediment in this reservoir, which had an S/V ratio of 0.2 m1, when samples were analysed in winter. As there was an insignificant difference (between kt and kRtwhen water temperature was 14.2 C, it can be expected that there was an insignificant difference at 15 C too. Therefore, the difference between kRt and kt,T observed in autumn can be mainly attributed to stratification.
11
4.4.
Analysis of Reservoir C
17/06/04
12/06/04
07/06/04
02/06/04
28/05/04
23/05/04
18/05/04
13/05/04
08/05/04
03/05/04
28/04/04
10 23/04/04
0.0
o
FRa
Temperature ( C)
Fm,T, FRa
autumn, but is barely noticeable in winter, and appropriate action can be taken when necessary. It is notable that outlet residuals (0.68 mg/L) were higher than the average in-reservoir residuals (0.48 mg/L shown in Fig. 5). If FRa is calculated using in-reservoir residuals, the value is 9.3 1.3, which is much higher than 5.7, the FRa calculated using outlet residual. Therefore, in a reservoir, which undergoes strong stratification in summer, the outlet residuals only give part of the story, and it is important to use in-reservoir samples to obtain the full picture. In Reservoir B, the Fm,T values were determined for samples collected at three different sampling points (S1, S2 and S3 shown in Fig. 2). At each site, samples were collected at two different depths: 0.2 m below surface (BS) and 3.0 m above bottom (AB). In winter (8th June), the difference in Fm,T was noted between different depths, similar to the Fm values reported by Fisher et al. (2009). The Fm,T values converted from measured Fm values at both depths for each sampling site differed in autumn (6th May) due to thermal stratification (Figs. 5 and 6). As for Reservoir A, FRa (using outlet residuals) and Fm,T are plotted in Fig. 6 for comparison. The difference between Fm,T and FRa was less than experimental error only in winter. However, in autumn, Fm,T and FRa values did not match each other even after considering the errors for both. The measured
Date
Fig. 6 e FRa, Fm,T and temperature profiles in Reservoir B. Calculated maximum and minimum Fm,T errors were ±0.2 and ±0.1 in May and ±0.4 and ±0.3 in June, respectively whereas maximum and minimum FRa errors were ±0.6 and ±0.3, respectively (Note: Stratified and de-stratified conditions were separated based on water temperature measured across the reservoir depth).
A similar analysis was conducted for Reservoir C (3 103 m3), which is much smaller than the other two reservoirs and nitrified mainly in summer and autumn. The reservoir is an elevated tank with maximum water level at 13 m above bottom. The inflow/outflow regime was controlled to maintain reservoir levels between 80 and 90% until September 2005. Thereafter, it was changed to 60e90% in order to increase mixing intensity. The small volume variation (80e90%) before September was unlikely to mix the reservoir contents properly, resulting in thermal stratification
4470
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 6 3 e4 4 7 2
Table 2 e Details of kRt and kt,T values in Reservoir B. kRta (h1) (using outlet residuals)
kRtb(h1) (using in-reservoir residuals)
6th May 2004
0.008 0.001
0.0125 0.0016
8th June 2004
0.0028 0.0006
0.0017 0.0005
Date
Sampling depth 0.2 3.0 0.2 3.0
m m m m
BS AB BS AB
kt,T (h1) at different sampling points S1 0.0099 0.0063 0.003 0.002
S2 0.0006 0.0005 0.0003 0.0002
0.017 0.011 0.0023 0.0019
S3 0.0005 0.0007 0.0002 0.0002
0.016 0.0063 0.0024 0.0018
0.0009 0.0005 0.0003 0.0002
Temperature ( C) 15.0c 14.2
a kRt values were calculated using outlet/inlet total chlorine residuals and retention time in Equation (1). b kRt values were calculated assuming the average in-reservoir residuals as outlet residual. c Autumn water temperature.
(Grayman et al., 2004). However, due to the depth of the reservoir (13 m), the bottom half may have been in a completely mixed condition (Fisher et al., 2009), making it difficult to see the effect of stratification at the outlet. In winter, better uniformity in terms of chemical parameters was noted in all the reservoirs studied. Therefore, the outlet residual in winter can be safely used for calculation. Inreservoir samples were not collected from this reservoir for occupational health and safety reasons. Therefore, all the analyses reported here are based on inlet and outlet chloramine residual measurements. The inlet chlorine and NH3eN varied from 1.3 to 1.6 mg/L and 0.3 to 0.35 mg/L, respectively. The inlet NO2eN concentrations were in the range of 0.002e0.005 mg/L, seasonally, which is higher than that of the other two reservoirs. Reservoir C was nitrifying each summer (temperature 20e24 C) and the total chlorine and NH3eN dropped to 0.5 and 0.18 mg/L, respectively while NO2eN increased up to 0.027 mg/L. For most of the winter (temperature 10e15 C), outlet chlorine residual was stable around 1.0 mg/L and the maximum NO2eN concentration was less than 0.005 mg/L. As for the other two reservoirs, FRa values were calculated using inlet kbc (0.0015 h1 at 20 C). The FRa, Fm,T and temperature profiles for 2005 are shown in Fig. 7. The kc values in FRa
Fm,T
Temperature
5
24
Water level variation 80-90 % 60-90%
22
Fm,T, FRa
3
18
2
16 14
o
20
Temperature ( C)
4
1 12
4.5. Summary of analysis for contribution from different niches
02/02/06
14/12/05
25/10/05
05/09/05
17/07/05
10 28/05/05
0
outlet samples were always about the same as the kbc values, possibly due to better mixing in the bottom half of the reservoir. This makes the Fm,T and FRa values comparable as mentioned in the concept development section. The calculated FRa values were in the range of 2.3e4.2. A high FRa value (4.2) was observed even in winter (13.8 C) as shown in Fig. 7. It could be due to the existence of an earlier summer effect. After that, the FRa value decreased, but remained above 2.3 through winter into spring. This could be due to low bacterial activity and no thermal and chemical stratification at lower ambient temperature (Fisher et al., 2009). Similarly, a low Fm,T value (about 0.6) was observed in winter and it increased up to 1.2 in summer (Fig. 7). The calculated FRa errors were in the range of 0.7 to 0.9, following a linear relationship with outlet chlorine residual or FRa values; i.e. if the outlet residual is low or FRa value is high, the magnitude of error in FRa estimation was high. There were significant differences between Fm,T and FRa values, after considering measurement errors for both. For example, on August 2005 (at 12.2 C), FRa was 2.7 0.6 whereas Fm,T was 0.7 0.3. As discussed above, analysis in winter can indicate the relative contribution of other niches to chloramine decay, although consideration should be given to possible microbial stratification (Fisher et al., 2009). This microbial stratification in winter, however, was limited to the surface layer. Hence, it is possible to neglect the microbial stratification to estimate the contribution from different niches, if mid-depth samples are considered. The difference in FRa and Fm,T is 2 indicating that the actual difference in decay is twice the value of kc,T at 12.2 C (0.0012 h1), i.e. 0.0024 h1. From this analysis, one can conclude that there is additional decay of about 0.0024 h1 at 12.2 C due to biofilm and sediment, which is substantial compared to the expected kc,T of 0.0012 h1.
Date
Fig. 7 e Fm,T, FRa and temperature profiles in Reservoir C. Calculated maximum and minimum Fm,T errors were ±0.4 and ±0.2, respectively, whereas maximum and minimum FRa errors were ±0.9 and ±0.7, respectively. Maximum and minimum error calculated in August 2005 for both Fm,T and FRa are presented to show significance of their difference.
In order to avoid the influence of stratification (thermal and chemical) on outlet residual and bulk water Fm, kRt and kt,T values were calculated in winter (12e13 C) for all three reservoirs. Both kRt and kt,T values were standardised to 20 C using Equation (2), assuming all the decay other than kbc is due to microorganisms. The relations between S/V ratio and kRt,20 and kt,20 values are presented in Fig. 8. Calculated kRt,20 and kt,20 values in Reservoir A, which has a low S/V ratio (0.08 m1), were 0.0026 0.0006 and 0.0022 0.0004 h1 respectively, whereas in Reservoir C, which has a relatively high S/V ratio
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 6 3 e4 4 7 2
addition, by measuring parameters in winter, chloramine residual can be predicted for the forthcoming summer using Equations (1) and (3). At low temperatures when inlet and outlet residuals are similar, FRa determination can present problems. However, performing the Fm test for samples collected in winter by raising temperature to 20 C can provide meaningful results. Therefore, a combination of FRa and Fm provides a better basis than traditional indicators (such as NO 2 and NH3) for reservoir chloramine residual management.
0.011 kt,20
Reservoir C
kRt,20
-1
kRt,20 and kt,20 (hr )
0.008
0.006
0.004
Reservoir B
Reservoir A
0.002
0.000 0.00
0.05
0.10
0.15
0.20
0.25
0.30
4471
0.35
5.
Conclusions
-1
Surface area to bulk water volume ratio, S/V (m )
Fig. 8 e Relation between surface area to bulk water volume ratio (S/V), kRt,20 and kt,20 values.
(0.31 m1), kRt,20 and kt,20 values were 0.0067 0.0016 and 0.0028 0.0004 h1, respectively. The difference between kRt,20 and kt,20 was insignificant at S/V ratio lower than 0.2 m1. However, when S/V ratio was 0.31 m1, the kRt,20 was found to be significantly different. These results demonstrated that in smaller reservoirs, there is a significant role of microorganisms associated with biofilm/sediments in accelerating chloramine loss (Fig. 8).
4.6. Implications of FRa and Fm for reservoir residual management It is necessary to understand the major factors contributing to accelerated chloramine decay before developing any control mechanisms for water distribution systems. Sathasivan et al. (2005) developed the Fm method for separating microbially assisted choramine decay from the chemical decay. Even though this method is simple and gives precise results, it measures only bulk water chemical and microbial activity. The FRa method developed in this paper integrates the major parameters contributing to chloramine decay in service reservoirs. Nevertheless, it is difficult to quantify the role of biofilm bacterial activity, sediment presence and stratification in chloramine decay, but comparing Fm,T and FRa can give the aggregate effects of these factors in accelerating this decay. The FRa concept can be a conceptual benchmark for modelling chloramine residual and for further investigation of individual roles of major decay factors in reservoirs. For example, determination of the biofilm effect in a well-controlled environment can provide a better insight into the role of biofilm in accelerating chloramine decay in a distribution system, especially in pipes. Furthermore, FRa can be used as an indicator to control microbial decay. The increasing trend of FRa values before the start of summer gives an early warning of potential loss of chloramine that may occur in the following summer. There will be then sufficient time to take remedial measures, such as decreasing retention time in the reservoir, and monitoring their effect using the Fm method (to be more accurate for bulk water) to ensure that the measures have been effective (Sathasivan et al., 2010). In
There are numerous factors responsible for loss of chloramine residual in reservoirs. To understand these and model them, it is necessary to quantify the contributions of chemical and microbial activity in bulk water and other factors separately. This paper attempts to separate these factors from the aggregate, to facilitate further investigation of processes that accelerate chloramine in reservoirs. A factor (FRa) is proposed that defines the degree of acceleration present in the reservoir over and above the chemical decay in the bulk water. Moreover, because of the advantages over traditional indicators, FRa and Fm related parameters can be effective indicators to control rapid chloramine loss progressively, especially in summer. The following conclusions are drawn from the results presented: i) The reservoir acceleration factor (FRa) can be determined using temperature, retention time, inlet and oultlet chloramine residuals. These values can be easily obtained by an operator. ii) Ongoing FRa monitoring can offer information on processes occurring in the reservoir to accelerate chloramine decay. In well-mixed reservoirs, the information will be more complete than in stratified reservoirs, as FRa is determined using outlet residuals. iii) Comparison of reservoir decay characteristics (FRa and kRt) and bulk water decay characteristics (Fm, km and kc), can assist in separating the role of chemical and microbial activity in bulk water from other major factors (biofilm bacterial activity, sediment presence, stratification and severe nitrification) which would be beneficial to understanding and modelling chloramine residual in reservoirs. iv) A significant effect of biofilm and sediment on chloramine decay was identified in a small reservoir (surface to volume ratio 0.31 m1), although impact was minimal in two large reservoirs (surface area to bulk water volume ratio 0.08 and 0.2 m1). In all reservoirs however, the effect of vertical thermal stratification was much greater than biofilm/sediment impact.
Acknowledgment Sydney Water Corporation approved the use of data from their distribution systems and funded the collection of samples as part of a routine monitoring program and
4472
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 6 3 e4 4 7 2
reservoir improvement project. The assistance of George Kastl, Paul Chuo, Clive Copelin and Mark Angles is gratefully acknowledged. The analysis was conducted as part of ARC linkage project LP0776766.
references
Baribeau, H., Prevost, M., Desjarddins, R., Lafrance, P., Gates, D.J., 2001. Chlorite and chlorate ion variability in distribution systems. J. Am. Water Works Assoc. 94 (7), 96e105. Barrios, M.R., 1989. Chloramine depletion in covered reservoirs. In: Proceedings of AWWA WQTC, Philadelphia, Pennsylvania, November 12e16. Brodtmann, N.V., Russo, P.J., 1979. The use of chloramine for reduction of trihalomethanes and disinfection of drinking water. J. Am. Water Works Assoc. 71, 40e42. Burlingame, G.A., Brock, G.L., 1985. Water quality deterioration in treated-water storage tanks. In: Proceedings of the AWWA Annual Conference, Washington, D.C., June 23e27. Clesceri, L.S., Greenberg, A.E., Eaton, A.D., 1998. Standard Methods for the Examination of Water and Wastewater. American Public Health Association; American Water Works Association and Water Environment Federation, Washington, DC. Cotruvo, J.A., 1981. THM’s in drinking water. Environ. Sci. Technol. 15, 268e274.
Fisher, I., Sathasivan, A., Chuo, P., Kastl, G., 2009. Effects of stratification on chloramine decay in distribution system service reservoirs. Water Res. 43 (5), 1403e1413. Grayman, W.M., Rossman, L.A., Deininger, R., Smith, C., Arnold, C., Smith, J., 2004. Mixing and aging of water in distribution system storage facilities. J. Am. Water Works Assoc. 96 (9), 70e80. Ike, N.R., Wolfe, R.L., Means, E.G., 1988. Nitrifying bacteria in a chloraminated drinking-water system. Wat. Sci. Technol. 20 (11e12), 441e444. Sathasivan, A., Fisher, I., Kastl, G., 2005. A simple method for measuring microbiologically assisted chloramine decay in drinking water. Environ. Sci. Technol. 39 (14), 5407e5413. Sathasivan, A., Fisher, I., Tam, T., 2008. Onset of severe nitrification in mildly nitrifying chloraminated bulk waters and its relation to biostability. Water Res. 42 (14), 3623e3632. Sathasivan, A., Chiang, J., Nolan, P., 2009. Temperature dependence of chemical and microbiological chloramine decay in bulk waters of distribution system. Wat. Sci. Technol. Water Supply vol. 9 (5), 493e499. Sathasivan, A., Fisher, I., Kastl, G., 2010. Application of the microbial decay factor to maintain chloramine in large tanks. J. Am. Water Works Assoc. 102 (4), 94e103. Srinivasan, S., Harrington, G.W., 2007. Biostability analysis for drinking water distribution systems. Water Res. 41 (10), 2127e2138. Stewart M.H., Lieu, N.I., 1997. Nitrification in chloraminated drinking water and its association with biofilms. In: Proceedings of the AWWA Water Quality Technology Conference, Denver Colorado.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 4 1 e4 3 5 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Development of granular sludge for textile wastewater treatment Khalida Muda a,*, Azmi Aris a, Mohd Razman Salim a, Zaharah Ibrahim b, Adibah Yahya b, Mark C.M. van Loosdrecht c, Azlan Ahmad a, Mohd Zaini Nawahwi b a
Department of Environmental Engineering, Faculty of Civil Engineering, Universiti Teknologi Malaysia, 81310, Skudai, Johor, Malaysia Department of Biological Sciences, Faculty of Biosciences and Bioengineering, Universiti Teknologi Malaysia, 81310, Skudai, Johor, Malaysia c Department of Biotechnology, Delft University of Technology, Julianalaan 67, 2628BC Delft, The Netherlands b
article info
abstract
Article history:
Microbial granular sludge that is capable to treat textile wastewater in a single reactor
Received 18 January 2010
under intermittent anaerobic and aerobic conditions was developed in this study. The
Received in revised form
granules were cultivated using mixed sewage and textile mill sludge in combination with
22 April 2010
anaerobic granules collected from an anaerobic sludge blanket reactor as seed. The gran-
Accepted 2 May 2010
ules were developed in a single sequential batch reactor (SBR) system under alternating
Available online 25 May 2010
anaerobic and aerobic condition fed with synthetic textile wastewater. The characteristics of the microbial granular sludge were monitored throughout the study period. During this
Keywords:
period, the average size of the granules increased from 0.02 0.01 mm to 2.3 1.0 mm and
Granulation
the average settling velocity increased from 9.9 0.7 m h1 to 80 8 m h1. This resulted in
Granule characterization
an increased biomass concentration (from 2.9 0.8 g L1 to 7.3 0.9 g L1) and mean cell
Textile wastewater
residence time (from 1.4 days to 8.3 days). The strength of the granules, expressed as the
Sequencing batch reactor
integrity coefficient also improved. The sequential batch reactor system demonstrated
Color removal
good removal of COD and ammonia of 94% and 95%, respectively, at the end of the study. However, only 62% of color removal was observed. The findings of this study show that granular sludge could be developed in a single reactor with an intermittent anaerobiceaerobic reaction phase and is capable in treating the textile wastewater. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
The development of aerobic granules as a novel treatment technology for wastewater has been extensively reported using sequencing batch reactor (SBR) systems. The system has been used to treat different types of wastewater and pollutants such as dairy effluent (Arrojo et al., 2004), soybean-processing wastewater (Su and Yu, 2005), nitrogen and phosphorus-rich effluent (Cassidy and Belia, 2005), phenol effluent (Carucci et al., 2009) and also municipal wastewater (de Kreuk and van Loosdrecht, 2006). The use of anaerobic
granules as seeding material for aerobic granules development has recently been reported by Linlin et al. (2005). In recent years, the ability of biodegradation for textile dyeing wastewater and dyestuffs involving both anaerobic and aerobic processes has been widely reported in the literature (Ong et al., 2005; Isik and Sponza, 2008; Franciscon et al., 2009). Color removal and complete mineralization of the dyes have been achieved through the combination of both processes. For azo dyes, the cleavage of N]N bond, which results in the removal of color, occurs during anaerobic stage along with the generation of aromatic amines, a toxic
* Corresponding author. Tel.: þ60 607 5531522/81; fax: þ60 607 5566157. E-mail address:
[email protected] (K. Muda). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.05.023
4342
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 4 1 e4 3 5 0
compound which is detrimental to human health (van der Zee and Villaverde, 2005). Under aerobic condition, mineralization of the amines takes place to complete the treatment process. Several studies have been carried out using anaerobic granular sludge but this doesn’t lead to complete removal of the dyes. Study conducted using granular sludge grown in anaerobic/aerobic system for complete dye removal is apparently missing. Since complete dye degradation requires both anaerobic and aerobic conditions, several studies have been focused on treatment systems which utilized two different reactors to fulfill both conditions (Ong et al., 2005; Moosvi and Madamwar, 2007; Isik and Sponza, 2008). The operation of such a system is rather complicated as the anaerobic microorganisms in the anaerobic tank need to be separated before the wastewater can be pumped to the aerobic tank. To simplify the system, a study was conducted to develop microbial granular sludge that can survive and function in both anaerobic and aerobic conditions and hence requires only one reactor. Potential strict anaerobic microorganisms can survive easily since oxygen only penetrates partially in the granules during aerated phase of the process. The study focused on the development of this type of granular sludge and the effectiveness of the system in treating synthetic textile dyeing wastewater.
2.
Materials and methods
2.1.
Wastewater composition
DO probe
2.2.
Reactor set-up
The schematic representation of the reactor set-up is given in Fig. 1. A column reactor was designed based on Wang et al. (2004) and Zheng et al. (2005) with several modifications. The column was designed for a working volume of 4 L with internal diameter of 8 cm and total height of 100 cm. The wastewater was fed into the reactor from the bottom of the column. Air was supplied into the reactor by a fine air bubble diffuser also from the bottom of the column. The decanting of the wastewater took place via an outlet port located at 40 cm
Peristaltic pumps controlled by timers
pH probe
Effluent
Sampling point
Influent tank
Effluent tank Influent
Fig. 1 e Schematic layout of the SBR reactor system.
height from the bottom of the reactor. The mean cell residence time (SRT) was set by the discharge of suspended solids with the effluent.
2.3.
Synthetic wastewater with the following composition was used: NH4Cl 0.16 g L1, KH2PO4 0.23 g L1, K2HPO4 0.58 g L1, CaCl2$2H20 0.07 g L1, MgSO4$7H2O 0.09 g L1, EDTA 0.02 g L1and trace solution 1 mL L1. The carbon sources used in this experiment were glucose (0.5 g L1), ethanol (0.125 g L1) and sodium acetate (0.5 g L1). The trace elements used were based on the composition recommended by Smolders et al. (1995). The composition of the trace element was H3BO3 (0.15 g L1), FeCl3$4H2O (1.5 g L1), ZnCl2 (0.12 g L1), MnCl2$4H2O (0.12 g L1), CuCl2$2H2O (0.03 g L1), NaMoO4$2H2O (0.06 g L1), CoCl2$6H2O (0.15 g L1), and KI 0.03 g L1. Mixed dyes consisted of Sumifix Black EXA, Sumifix Navy Blue EXF and Synozol Red K-4B with total concentration of 50 mg L1 was used in this study. The mixture gave an initial COD of 1270 mg L1; 1020 ADMI (American Dye Manufacturing Index) and average ammonia concentration of 38 mg L1. The pH of the synthetic wastewater was adjusted to 7.0 0.5 before feeding.
Air compressor controlled by a timer
Mass-flow controller
Analytical methods
The morphological and structural observations of the granules were carried out by using a stereo microscope equipped with digital image processing and analyzer ((PAXITv6, ARC PAX-CAM)). The microbial compositions within the granules were observed qualitatively with scanning electronic microscope (FESEM-Zeiss Supra 35 VPFESEM). The granules were left to dry at room temperature prior to gold sputter coating (Bio Rad Polaron Division SEM Coating System) with coating current of 20 mM for 45 s. Other parameters such as mixed liquor suspended solid (MLSS), mixed liquor volatile suspended solid (MLVSS), COD, color and NH3 were analyzed according to the Standard Methods (APHA, 2005). The granules developed in the SBR column were analyzed for their physical, chemical and biological characteristics. Physical characteristics include settling velocity, sludge volume index (SVI) and granular strength. The settling velocity was determined by averaging the time taken for an individual granule to settle at a certain height in a glass column filled with tap water. The SVI assessment was carried out according to the procedure described by Beun et al. (1999). Determination of the granular strength was based on Ghangrekar et al. (2005). Shear force on the granules was introduced through agitation using an orbital shaker at 200 rpm for 5 min. At certain amplitude of the shear force, parts of the granules that are not strongly attached within the granules were detached. The quantity of the ruptured granules was separated by allowing the fractions to settle for 1 min in a 150 ml measuring cylinder. The dry weight of the settled granules and the residual granules in the supernatant were measured. The ratio of the solid in the supernatant to the
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 4 1 e4 3 5 0
total weight of the granular sludge used for granular strength measurement was expressed in percentage as an integrity coefficient (IC). This percentage indirectly represents the strength of the granules. Smaller IC value indicates stronger granule and vice versa. The granules were analyzed chemically for their mineral content which includes Ca2þ, Mg2þ, Naþ, Kþ and Fe2þ. The mineral content was determined using Perkin Elmer Analyst 400 Flame Atomic Absorption Spectrophotometer (FLAA). The microbial activity of the microbial granular sludge was conducted by measuring the oxygen utilization rate (OUR), specific oxygen utilization rate (SOUR) and specific methanogenic activity (SMA). The OUR (mg.O2 L1 h1) measurement was performed according to the Standard Methods (APHA, 2005). The SMA measurements were conducted according to Erguder and Demirer (2008) with several modifications where a 500 mL BOD bottle seeded with FSG with final concentration of 1e2 g VSS L1 and basal medium (250 mL effective volume). The bottle was flushed with N2 gas mixture for 5 min to obtain an anaerobic condition. The bottle was then sealed with rubber septum. Acetic acid (HAc) was fed into the serum bottle at a concentration of 3000 mg L1. The experiments were conducted in a temperature controlled condition of 30 2 C. The production of methane gas (CH4) was determined according to Erguder et al. (2001). The production of methane gas (CH4) was determined daily for 5e7 days using liquid displacement methods containing concentrated KOH stock solution (20 g L1). After each gas measurement, the bottle was manually shaken. At the end of the SMA assay, the VSS content in the bottle was measured. The SMA was calculated as the maximum CH4 produced per gram of VSS per hour (mL CH4/g1 VSS h1) (Zitomer and Shrout, 1998).
2.4.
Experimental procedures
A mixture containing an equal volume of sludge from a municipal sewage treatment plant and a textile mill wastewater treatment plant that gave a total volume of 2 L was used in this experiment. The sludge inoculums were sieved with a mesh of 1.0 mm to remove large debris and inert impurities. The sludge mixture was acclimatized for two months with 2 L of synthetic wastewater containing dye degrading microbes. The dye degrading microbes used in this study was based on the study conducted by Nawahwi (2009) and Ibrahim et al. (2009). Together with the sludge mixture, about 100 mL of anaerobic granules with size less than 1 mm diameter were used as seed for the granulation process. The anaerobic granules were collected from an anaerobic sludge blanket reactor system treating paper mill industrial effluent. The MLSS of the anaerobic granules were 3.3 g L1. During the start-up period, 2 L of mixed sludge and 2 L of synthetic textile wastewater were added into the reactor system making the final volume of 4 L with total sludge concentration after inoculation of 5.5 g L1. The system was supplied with external carbon sources consist of glucose, sodium acetate and ethanol which gave a substrate loading rate of 2.4 kg COD m3 d1. The reactor was operated in successive cycles of 6 h. Each cycle comprised of 5 min filling, 340 min reaction, 5 min settling, 5 min decanting and 5 min idle. The reaction phase started with an anaerobic phase of
4343
40 min, followed by aerobic phase for 130 min, another 40 min of second anaerobic phase and 130 min for second aerobic phase. The dissolved oxygen (DO) concentration remained low during the anaerobic condition (0.2 mg L1) and reached saturation concentration during the aerobic phase. The superficial air velocity during the aerobic phase was 1.6 cm s1. The pH during the reaction process varied in the range of 6.0e7.8 and the temperature of the reactor system was set at 30 2 C. The reactor system was operated for a period of 66 days. Two liters of the wastewater remained in the reactor after the decanting stage yielding a volumetric exchange rate (VER) of 50%. 20 mL of sample from the influent and effluent (wastewater released after decanting stage) of the reactor system were collected for the measurement of COD, ammonia and color removal (Fig. 1).
3.
Results and discussions
3.1.
Biomass profile
The change in biomass concentration (i.e. MLSS) from the start-up until the end of the study is shown in Fig. 2. During the first few days of the experiment, almost half of the sludge was washed-out from the reactor causing a rapid decrease in the biomass concentration. The MLSS reduced from initial concentration of 5.5 g L1 to 2.9 g L1mainly due to the short settling time used in the cycle (i.e. 5 min). During this initial stage, the anaerobic granules were also observed to disintegrate into smaller fragmented granules and small debris resulted from shear force caused by the aeration during the aerobic stage. These small fragments have poor settling ability and were washed out from the reactor. This caused an increase of the suspended solids concentration in the effluent as shown in Fig. 2. As the experiment continued and granules with adapted biomass were formed, the concentration of the biomass in the reactor increased and finally reached 7.3 0.9 g MLSS L1 when the experiment was discontinued on the 66th day. The MLVSS followed the same trend as MLSS,
Fig. 2 e Change in biomass concentration during the formation of granular sludge in the SBR. (C) MLSS, (,) MLVSS, (A) Suspended solid in the effluent.
4344
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 4 1 e4 3 5 0
Fig. 3 e The changes of dissolved oxygen and oxygen uptake rate in one complete cycle of the SBR reactor system (A) DO, (,) OUR at day 66th of the experiment. (PI) First aerobic phase, (PII) First anaerobic phase, (PIII) Second aerobic phase, (PIV) Second anaerobic phase.
ranging from 1.9 0.5 g L1 to 5.6 0.8 g L1. The SRT also increased from 1.4 days at the initial stage to 8.3 days on the 66th day, indicating the accumulation of the biomass in the reactor.
3.2.
Bioactivities of the granules
A typical DO concentration profile for one complete cycle and the OUR profile during both of the aerobic reaction phases are shown in Fig. 3. Stage PI and PIII show the first and second stage of anaerobic reaction phase, respectively. At these anaerobic stages, most of the dye degradation is expected to occur where amines, as the byproduct, were released (Sponza and Isik, 2005). Stage PII and PIV represent the first and second stage of aerobic reaction phase, respectively. Most of the substrates provided to the reactor system were anticipated to be consumed within a few minutes of the first aerobic reaction phase (PII), known as the feast period. During the feast period, the DO concentration in the reactor was low (about 4 mg L1). The high utilization of DO during the feast period was also indicated by the high OUR which was 281 mg L1 h1. The amines, which were produced during anaerobic reaction phase (PI), were mineralized under this aerobic condition (PII) as they cannot be further degraded under anaerobic phases (Sponza and Isik, 2005). When all the carbon sources (substrate and amines) in the wastewater were utilized, endogenous respiration process took place, which is referred as the famine period. The transition from the feast to famine phase was clearly observed with the drastic increase of the dissolved oxygen and the extreme drop of the OUR within few minutes of the aerobic reaction phase (PII). The DO concentration immediately increased to around 7.0 mg L1 which was closed to the DO
saturation level. The OUR also reduced to 14 mg L1 h1 indicating low utilization of DO. Since there was no addition of substrate during the second aerobic reaction phase (PIV), the consumption of DO during this phase was also low. This is shown by high DO level reaching saturation value of 7.6 mg L1. A sharp increase in the OUR was observed at the beginning of this phase but at a lower value than the one observed in Stage PII. Apparently, the residual dyes which were not degraded in Stage PI and PII were transformed into smaller molecules (e.g. amines) during the second stage of the anaerobic phase (PIII). These smaller molecules were further mineralized in Stage PIV which resulted in a sharp increase in the OUR. As the concentration of these molecules were reduced, the OUR also became lower until it reached a minimum of 11 mg L1 h1 Table 1 shows the OUR value during both of the aerobic reaction phases. The SOUR of the microbial granular sludge was determined before the termination of the experiment. The SOUR was 51.1 6.8 mg DO g1 VSS h1. This value was slightly lower than those of the aerobic granules reported by Tay et al. (2001) which ranged from 55.9 to 69.4 mg DO g1 VSS h1and higher than the coupled granules reported by Erguder and Demirer
Table 1 e The OUR levels during the aerobic reaction phase of one complete cycle. Aerobic reaction phase Begin react End react
OUR (mg L1 h1) 1st stage (PII)
2nd stage (PIV)
281 39 14 2
167 51 11 2
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 4 1 e4 3 5 0
(2005) (6e47 mg DO g1 VSS h1). The specific methanogenic activity (SMA) of the microbial granular sludge is lower (10.3 mL CH4 g1 VSS h1) than the one reported by Erguder and Demirer (2005) (14e42 mL CH4 g1 VSS h1). However, despite the low SMA emission, it provides the evidence of the existence of methanogens within the microbial granular sludge. Obviously, the granular sludge offered the methanogens sufficient protection from the toxic oxygen concentration in the bulk liquid.
3.3.
Morphology of granular sludge
A week after inoculating the reactor, visual and microscopic observations of granule formation were made. At this stage, the developed granules were composed mostly of loosely clumped sludge which could easily break up into pieces if placed under vigorous shaking. Within a week, the anaerobic seed granules had undergone morphological changes from spherical in shape and black in color with average diameter of 1 mm into smaller grey granules. It is likely that the sulfides in the anaerobic sludge were oxidized due to exposure to the aerobic condition. On day 30, two different types of granules were clearly observed in the reactor as shown in Fig. 4. Fig. 4a shows mainly irregular shaped, yellow colored granules (Type A) that are solely developed from the activated sludge. In Fig. 4b, the anaerobic granules that have fragmented into smaller pieces have formed different sizes of granules (Type B) that contained pieces of anaerobic granules. The outer layer of the latter were yellow in color indicating the
4345
domination of aerobic or facultative microorganisms while the darker spots within the granules indicate the presence of anaerobic fragments originated from anaerobic granules. The formation of Type A granules could be elucidated by the mechanisms explained by Beun et al. (1999). The development was initiated from the mycelial pellets that were retained in the reactor due to high settling velocity. These mycelial pellets eventually become the support matrix for the bacteria growth. Bacteria that were able to attach to this matrix were retained and suppressed the filamentous growth and became the dominant species in the reactor. The formation of Type B granules has been discussed by Linlin et al. (2005). These granules were formed through a series of physical and morphological changes. The anaerobic granules initially disintegrated into smaller size flocs and debris when exposed to aeration forces in the SBR column. Some of the granules and debris that were too small were washed-out with the effluent while the heavier ones were retained in the column and acted as nuclei for the formation of the aerobic granules. This type of granules that consisted of combination of aerobic and anaerobic portions within the granules could increase the possibility of degradation process that requires both aerobic and anaerobic conditions for complete degradation particularly for textile wastewater that contains azo-dyes. Fig. 4c shows the sludge particles during the initial stage of the experiment with an average size of 0.02 0.01 mm while Fig. 4d shows the granules at the final stage (day 66) of the experiment with the average particle diameter size of 2.3 1.0 mm with maximum size reaching up to 4 mm.
Fig. 4 e Morphological development of granular sludge from anaerobic granular sludge and aerobic activated sludge. Pictures were taken using stereo microscope with magnification of 6.33, Scale bar equals to 1 mm. (a) Granules developed from the activated sludge. (b) Granules developed from anaerobic granules patches. (c) Sludge particles during early stage of experiment. (d) Microbial granular sludge at the 66 days of the experiment.
4346
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 4 1 e4 3 5 0
Fig. 5 e SEM microstructure observation on mature microbial granular sludge at a magnification of 10,000 K. (a) Coccoid bacteria tightly linked to one another. (b) Cavities that appear between bacteria clumped inside the granules.
The microstructure of the microbial granular sludge was examined using SEM and shown in Fig. 5. The SEM observation of the mature granules shows the domination of non-filamentous coccoid bacteria that are tightly linked and embedded to one another and form a rounded shape on the surface of the granule, covered with extrapolysaccarides (EPS) (Fig. 5a). The absence of filamentous bacteria in the developed granules may be due to the experimental conditions that did not favor the growth of filamentous bacteria at such high concentration of DO during aerobic phase (i.e. 7.0 0.5 mg L1) and considerably high organic loading rate (2.4 kg COD m3 d1) (Chudoba, 1985; Zheng et al., 2006). Fig. 5b shows the presence of cavities between the clumped bacteria. These cavities are anticipated to be responsible in allowing smooth mass transfer of substrates or metabolite products in and out of the granules.
3.4.
Physical characteristics of granular sludge
3.4.1.
Size
The shear force imposed in the development of granules in this experiment, in terms of superficial upflow air velocity which was 1.6 cm s1, resulted in the development of microbial granular sludge with average diameter of 2.3 1.0 mm. According to Peng et al. (1999), the diameter of the developed aerobic granules is in the range of 0.3e0.5 mm which is much smaller as compared to anaerobic and anoxic granules that could develop up to 2 to 3 mm. The strong shearing force produced by mixing and aeration during the reaction phase could prevent the development of bigger granules which can be achieved in an anaerobic system (van Benthum et al., 1996; Kwok et al., 1998).
3.4.2.
settling velocity of the aerobic granules reported by Zheng et al. (2005) (i.e. 18e31 m h1). The increase in settling velocity has given significant impact on the biomass concentration in the reactor. The relationship between settling velocity of the granules and the concentration of the MLSS is shown in Fig. 6. Despite the short settling time (5 min), the high settling velocity possessed by the developed microbial granular sludge enabled the granules to escape from being flushed out during the decanting phase. Such conditions have caused more microbial granular sludge to be retained in the system and resulted in the increase of biomass concentration in the reactor. In this experiment, the SRT value was 1.4 days during the start-up (partly low due to wash out of inoculated sludge) and rose up to 8.3 days at the end of experiment. As less biomass was washed-out during the decanting period, the increase in SRT is another manifestation of good settling characteristics resulting from the high settling velocity. The SVI value has improved from 276.6 mL g1 at the initial stage of the experiment to 69 mL g1 at the end of the experiment indicating the good settling properties of the
Settling velocity
The average settling velocity of the seed sludge and seed anaerobic granular sludge was 9.9 0.7 m h1 and 42 8 m h1 respectively. The average settling velocity of the anaerobic granular seed is in accordance with those reported by Schmidt and Ahring (1996) which is in the range of 18e100 m h1. The average settling velocity of the granular sludge developed in this study increased from 17.8 2.6 m h1 to 80 8 m h1 at the end of experiment. The settling velocity obtained in this study is almost three times greater than the
Fig. 6 e The relationship between the settling velocity of the microbial granular sludge and the biomass concentration retained in the reactor. (,) Settling velocity (C) Biomass concentration.
4347
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 4 1 e4 3 5 0
Fig. 7 e The SVI and mean cell residence time (SRT) profile. (B) SVI, (-) SRT.
granules which is favorable in wastewater treatment plant operation. The change of the SVI value and the SRT as a function of time are given in Fig. 7. The SVI value achieved in this experiment is in agreement with the result reported by McSwain et al. (2004) with SVI values of 115 36 mL g1 (settling time 10 min) and 47 6 mL g1 (settling time 2 min). The higher settling velocity and lower SVI value of the mature microbial granular sludge as compared to previous reports by other researchers indicate that the formation of granules seeded with anaerobic granules would develop better settling properties of the granules. It may also be due to the specific reaction condition of anaerobic/aerobic setting in the experiment that induced the well settling of the granule.
3.4.3.
Granular strength
The granular strength of the granules was measured based on the integrity coefficient (IC) as described by Ghangrekar et al. (2005). The smaller the IC value, the higher the strength and ability of the granules to remain as high structural integrity granules during aeration phase that caused the shear force.
Fig. 8 shows the IC profile of the developed granular sludge as a function of time. The IC reduces as the granules developed. With an initial value of 30 0.3, the IC was reduced to about 9.4 0.5 at the termination of the experiment. A sharp reduction of IC was observed after 40 days of the experimental run. According to Ghangrekar et al. (2005) granules with integrity coefficient of less than 20 were considered high strength granules. The reduction in IC value indicates the increase in the strength of the bond that holds the microorganisms together within the developed granules. During the early stage of the granule development, the microbes within the granules were loosely bounded to each other. When the microbes were loosely linked together, the granules may contain more cavities which make the granules less dense, as shown by low settling velocity value. As more microbes were linked together, the granules increased in size. Under the applied selective pressures (i.e. short settling time, hydrodynamic shear force, feast-famine regime) within the reactor, microbes may produce more EPS (Qin et al., 2004). As reported by Adav et al. (2008), the EPS could contribute greatly to the strength and the stability of anaerobic granules. When more EPS are being produced by the microbial cells, they form a cross-linked network and further strengthen the structural integrity of the granules. The cavities within the granules may be filled with the EPS as it is a major component of the biogranule matrix material in both anaerobic and aerobic granules. This caused the granules to become denser and stronger as shown by their high settling velocity and low IC value at the end of the experiment.
3.4.4.
Mineral content
The concentration of minerals in granular sludge, newly developed and matured granular sludge were determined in mg/g of dry sludge and presented in Table 2. The concentration of Naþ and Kþ are not much different in the sludge, newly developed and matured granules. However, there is an obvious increase in the concentration of Ca2þ and Mg2þ within the matured granules. The concentration of Fe2þ was slightly reduced in the matured and newly developed granules as compared to the sludge. Basically, an unchanged concentration of Naþ and slightly decreasing Kþ concentration in the sludge and matured granules may indicate that these monovalent cations may not be involved in the granulation process. It has been reported that high concentration of the Naþ and Kþ may cause adverse
Table 2 e Comparison of mineral content at different stages during the development of microbial granular sludge. Mineral contents (mg g1 of dry sludge) Mineral
Fig. 8 e The change in integrity coefficient (representing the granular strength) during the formation of granular sludge in the anaerobic/aerobic SBR.
Ca2þ Mg2þ Naþ Kþ Fe2þ
Sludge
1.53 0.13 0.22 1.31 2.32
0.02 0.01 0.07 0.07 0.02
Newly developed Matured microbial microbial granular granular sludge sludge (1 week) (10 weeks) 2.0 0.6 0.322 0.003 0.25 0.04 1.15 0.03 1.90 0.04
4.65 1.75 0.24 0.93 1.98
0.04 0.08 0.05 0.05 0.08
4348
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 4 1 e4 3 5 0
effect on the granules formation. It could cause reduction in sludge concentration, settling velocity of the sludge, granular strength and treatment efficiency (Ghangrekar et al., 2005). The monovalent cation, Naþ has also been reported causing detrimental impact on the flocculation system (Sobeck and Higgins, 2002). Nevertheless, there are contradictory reports related to the effect of these monovalent cations in granulation development processes. Fernandez et al. (2008) reported that the concentration of granular biomass has improved significantly in the reactor system fed with high concentration of inorganic salt influent. A quick decrease of solids concentration in the effluent was observed after the addition of NaCl. The development of the granular sludge in this study showed higher accumulation of Ca2þ and Mg2þ towards the end of the experiment. This may indicate the involvement of the inorganic elements in the granulation process. Based on the divalent cation bridging theory, the presence of Ca2þ and Mg2þ promotes equivalent floc properties (Sobeck and Higgins, 2002). According to Ren et al. (2008), granule-rich Ca2þ showed more rigid granular structure and higher strength as compared to granule without Ca2þ accumulation.
3.5.
After 340 min of intermittent anaerobic and aerobic modes, about 93%, 95% and 62% of COD, ammonia, and color respectively were removed. During the first anaerobic phase (PI) (0e40 min), approximately 15% and 4% of COD and ammonia respectively, were removed. In the first aerobic phase (PII) (40e170 min), about 68% of the COD was removed while 80% of the ammonia was oxidized. Most of the organic compounds and nitrification of ammonia were achieved during this stage. The supply of oxygen at this stage enabled good oxidation of these compounds (Brauer and Henning, 1986). The nitrate produced
Removal performance
The performance of the reactor system from start-up until the end of granules development period based on the removal of COD, color and ammonia is given in Fig. 9. At the initial stage of the operation, the percentage removal for COD and ammonia was 71% and 67% respectively (Fig. 9a and b). The removal efficiency increased to 94% for COD and 95% for ammonia at the end of the experiment. The increase in the removal efficiency indicates the occurrence of high biological activity in the reactor system. During the first month, the removal efficiency for COD and ammonia fluctuated but the removal became stable for the remaining period. The removal efficiency for color was fluctuating almost throughout the study period (Fig. 9c). The percentage of color removal was about 25% during the start up and increased to 62% at the end of the experiment. The average of color removal was 55%. This low percentage of the color removal may be due to insufficient adaptation time. As dye substances are recalcitrant and difficult to be degraded, more time is required to accumulate sufficient organisms which degrade the dyes in the reactor. The inconsistent percentage for color removal may also be contributed by the unstable condition of the aromatic amines, the byproduct of dye degradation which easily oxidized and recolor when exposed to oxygen during the aerobic phase. The increase of color during autoxidation of aromatic amines was confirmed by several researchers (Cruz and Buitron, 2001; Libra et al., 2004; Sponza and Isik, 2005). The inconsistency of color removal may be also influenced by the absorption of color into sludge biomass throughout the experiment. The absorption of color into the sludge biomass has been reported by other researchers (Otero et al., 2003; Wang et al., 2006; Sirianuntapiboon and Srisornsak, 2007). Fig. 10 shows the percentage removal of COD, ammonia and color in a complete 340-min reaction phase of the SBR system recorded on the 66th days of experiment. The profile and the percentage removal for COD and ammonia were almost the same while the removal of color was much lower.
Fig. 9 e Percentage removal for (a) COD, (b) ammonia and (c) color of the SBR system. (A) Influent, (B) Percentage removal, (-) Effluent. (ADMI) American Dye Manufacturing Units.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 4 1 e4 3 5 0
4349
sludge since the average size of microbial granular sludge developed in this study was 2.3 1.0 mm. Aerobic microorganisms may be found at the outer layer of the granules which can easily access the oxygen molecule and mainly responsible for the COD removal. The facultative microorganisms may be found in any part of the microbial granular sludge due to its capability to live both under anaerobic and aerobic condition.
4.
Fig. 10 e The removal for COD, ammonia and color in one complete cycle of the SBR system. (-) Color, (B) COD, (:) Ammonia. (PI) First aerobic phase, (PII) First anaerobic phase, (PIII) Second aerobic phase, (PIV) Second anaerobic phase.
will be removed through the denitrification process that will occur in the second stage of anaerobic phase. The second anaerobic phase (PIII) (170e210 min) showed only about 5% of COD and ammonia being removed while the remaining (about 4%) was removed in the second aerobic phase (PIV) (210e340 min). As for color, about 45% and 16% were removed during anaerobic and aerobic phases respectively. The result shows the ability of anaerobic microbes within the granule to degrade the dye. The high percentage for color removal indicated active cleavage of dye compound took place during both of the anaerobic phases (PI and PIII). The degradation and decolorization of dye during anaerobic condition has been widely reported in the literatures (dos Santos et al., 2007). In anaerobic condition, the electrons from the electron donor are transferred to the N]N bond of the azo dye causing the cleavage of the bond forming aromatic amines. The amines are then degraded under aerobic condition reducing the COD value of the wastewater. In addition to the degradation mechanism, dye removal may also occur via adsorption onto the biomass (Aksu, 2001 and Crini, 2006). Amines, the colorless byproduct of anaerobic degradation of dye compound are unstable compound that could easily be oxidized during the presence of oxygen. These autoxidation of the amines may form different intensity colored compound (Cruz and Buitron, 2001; Libra et al., 2004; Sponza and Isik, 2005). This reaction may cause the reduction on the overall percentage of color removal during aerobic condition (PII and PIV). Based on the removal performance of the system, it has been proven that the developed microbial granular sludge is capable to perform the degradation process during anaerobic and aerobic phases. This indicates the presence of aerobic, facultative and anaerobic microorganisms in the microbial granular sludge. According to Li and Liu (2005), when the granules grew to a size larger than 0.5 mm, the diffusion of oxygen into the inner part of the granules became a limitation. This may give an indication of the presence of anaerobic microorganisms within centre part of the microbial granular
Conclusion
Stable microbial granular sludge could be cultivated in a single SBR system with the application of intermittent anaerobic and aerobic reaction mode during the reaction phase. The matured granules showed the domination of non-filamentous bacteria that were tightly linked and embedded to one another and covered with EPS. The SVI value of the biomass has decreased from 276.6 mL g1 to 69 mL g1at the end the 66 days, also indicating the excellent settling properties of the granules. The development of the granular sludge is positively correlated with the accumulation of divalent cationic Ca2þ and Mg2þ in the granules suggesting the role played by the cations in the granulation process. The results indicate the viability of the single reactor system for treating textile wastewater under intermittent anaerobic and aerobic phase strategy. The OUR/SOUR and SMA analyses indicate the presence of anaerobic and aerobic microorganisms activities in the granular sludge which is capable to perform degradation process both in anaerobic and aerobic conditions.
Acknowledgements The authors wish to thank the Ministry of Science, Technology and Innovation (MOSTI), Ministry of High Education (MOHE) and Universiti Teknologi Malaysia for the financial supports of this research (Grants No.: 79137, 78211 and 75221).
references
Adav, S.S., Lee, D.J., Tay, J.H., 2008. Extracellular polymeric substances and structural stability of aerobic granule. Water Research 42, 1644e1650. Aksu, Z., 2001. Biosorption of reactive dyes by dried activated sludge: equilibrium and kinetic modeling. Biochemical Engineering Journal 7, 79e84. Arrojo, B., Mosquera-Coral, A., Garrido, J.M., Mendez, R., 2004. Aerobic granulation with industrial wastewater in sequencing batch reactors. Water Research 38, 3389e3399. Beun, J.J., Hendriks, A., van Loosdrecht, M.C.M., Morgenroth, E., Wilderer, P.A., Heijnen, J.J., 1999. Aerobic granulation in a sequencing batch reactor. Water Research 33 (10), 2283e2290. Brauer, H., Henning, R., 1986. Aerobic two-stage biological degradation of organic carbon compounds and ammonium. Bioprocess Engineering 1, 13e22. Carucci, A., Milia, S., De Gioannis, G., Piredda, M., 2009. Acetatefed aerobic granular sludge for the degradation of 4chlorophenol. Journal of Hazardous Materials 166, 483e490.
4350
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 4 1 e4 3 5 0
Chudoba, J., 1985. Control of activated sludge filamentous bulking VI formulation of basic principle. Water Research 19, 471e479. Crini, G., 2006. Non-conventional low-cost adsorbents for dye removal: a review. Bioresource Technology 97, 1061e1085. de Kreuk, M.K., van Loosdrecht, M.C.M., 2006. Formation of aerobic granules with domestic sewage. Journal of Environmental Engineering 132, 694e697. Erguder, T.H., Demirer, G.N., 2005. Investigation of granulation of a mixture of suspended anaerobic and aerobic cultures under alternating anaerobic/microaerobic/aerobic conditions. Process Biochemistry 40, 3732e3741. Erguder, T.H., Demirer, G.N., 2008. Low-strength wastewater treatment with combined granular anaerobic and suspended aerobic cultures in upflow sludge blanket reactors. Journal of Environmental Engineering 134, 295e303. Erguder, T.H., Tezel, U., Guven, E., Demirer, G.N., 2001. Anaerobic biotransformation and methane generation potential of cheese whey in batch and UASB reactors. Waste Management 21, 643e650. Franciscon, E., Zille, A., Garboggini, F.F., Silva, I.S., Paulo, A.C., Durrant, L.R., 2009. Microaerophiliceaerobic sequential decolourization/biodegradation of textile azo dyes by a facultative Klebsiella sp. strain VN-31. Process Biochemistry 44, 446e452. Ghangrekar, M.M., Asolekar, S.R., Joshi, S.G., 2005. Characteristics of sludge developed under different loading conditions during UASB reactor start-up and granulation. Water Research 39, 1123e1133. Ibrahim, Z., Amin, M.F.M., Yahya, A., Aris, A., Umor, N.A., Muda, K., 2009. Characterisation of microbial flocs formed from raw textile wastewater in aerobic biofilm reactor (ABR). Water Science and Technology 60.3, 683e688. Isik, M., Sponza, D.T., 2008. Anaerobic/aerobic treatment of a simulated textile wastewater. Separation and Purification Technology 60, 64e72. Linlin, H., Jianlong, W., Xianghua, W., Yi, Q., 2005. The formation and characteristics of aerobic granules in sequencing batch reactor (SBR) by seeding anaerobic granules. Process Biochemistry 40, 5e11. McSwain, B.S., Irvine, R.L., Wilderer, P.A., 2004. The influence of settling time on the formation of aerobic granules. Water Science and Technology 50 (10), 195e202. Moosvi, S., Madamwar, D., 2007. An integrated process for the treatment of CETP wastewater using coagulation, anaerobic and aerobic process. Bioresource Technology 98, 3384e3392. Nawahwi, M.Z., 2009. Degradation of Reactive Red 195 by Selected Bacteria from Textile Wastewater, Master Thesis. Ong, S.A., Toorisaka, E., Hirata, M., Hano, T., 2005. Treatment of azo dye orange II in aerobic and anaerobic-SBR systems. Process Biochemistry 40, 2907e2914. Peng, D., Bernet, N., Delgenes, J.P., Moletta, R., 1999. Aerobic granular sludge-a case study. Water Research 33, 890e893. Qin, L., Tay, J.H., Liu, Y., 2004. Selection pressure is a driving force of aerobic granulation in sequencing batch reactors. Process Biochemistry 39, 579e584. Ren, T.T., Liu, L., Sheng, G.P., Liu, X.W., Yu, H.Q., Zhang, M.C., et al., 2008. Calcium spatial distribution in aerobic granules and its effects on granule structure, strength and bioactivity. Water Research 42, 3343e3352. Schmidt, E., Ahring, B.R., 1996. Granular sludge formation in upflow anaerobic sludge blanket (UASB) reactors. Biotechnology Bioengineering 49, 229e246. Sirianuntapiboon, S., Srisornsak, P., 2007. Removal of disperse dyes from textile wastewater using bio-sludge. Bioresource Technology 98, 1057e1066. Sobeck, D.C., Higgins, M.J., 2002. Examination of three theories for mechanisms of cation-induced bioflocculation. Water Research 36, 527e538.
Sponza, D.T., Isik, M., 2005. Reactor performances and fate of aromatic amines through decolorization of Direct Black 38 dye under anaerobic/aerobic sequentials. Process Biochemistry 40, 35e44. Tay, J.H., Liu, Q.S., Liu, Y., 2001. Microscopic observation of aerobic granulation in sequential aerobic sludge reactor. Journal of Applied Microbiology 91, 168e175. van Benthum, W.A.J., Garrido, J.M., Tijhuis, L., Van Loosdrecht, M. C.M., Heijnen, J.J., 1996. Formation and detachment of biofilms and granules in a nitrifying biofilm airlift suspension reactor. Biotechnology Progress 12, 764e772. Zheng, Y.M., Yu, H.Q., Sheng, G.P., 2005. Physical and chemical characteristics of granular activated sludge from a sequencing batch airlift reactor. Process Biochemistry 40, 645e650. Zheng, Y.M., Yu, H.Q., Liu, S.J., Liu, X.Z., 2006. Formation and instability of aerobic granules under high organic loading conditions. Chemosphere 63, 1791e1800. Zitomer, D.H., Shrout, J.D., 1998. Feasibility and benefits of methanogenesis under oxygen-limited conditions. Waste Management 18 (2), 107e116. APHA, 2005. Standard Methods for the Examination of Water and Wastewater. American Public Health Association, Washington, DC. Cassidy, D.P., Belia, E., 2005. Nitrogen and phosphorus removal from an Abattoir wastewater in a SBR with aerobic granular sludge. Water Research 39 (19), 4817e4823. Cruz, A., Buitron, G., 2001. Biodegration of disperse blue 79 using sequenced anaerobic/aerobic biofilters. Water Science and Technology 44, 159e166. Fernandez, I., Vazquez-Padin, J.R., Mosquera-Corral, A., Campos, J.L., Mendez, R., 2008. Biofilm and granular systems to improve Anammox biomass retention. Biochemical Engineering Journal 42 (3), 308e313. Kwok, W.K., Picioreanu, C., Ong, S.L., Loosdrecht, M.C.M., Ng, W.J., Heijinen, J.J., 1998. Influence of biomass production and detachment forces on biofilm structures in a biofilm airlift suspension reactor. Biotechnology and Bioengineeing 58 (4), 400e407. Li, Y., Liu, Y., 2005. Diffusion of substrate and oxygen in aerobic granule. Biochemical Engineering Journal 27, 45e52. Libra, J.A., Maren Borchert, M., Vigelahn, L., Storm, T., 2004. Two stage biological treatment of a diazo reactive textile dye and the fate of the dye metabolites. Chemosphere 56, 167e180. Otero, M., Rozada, F., Calvo, L.F., Garcı’a, A.I., Mora’n, A., 2003. Kinetic and equilibrium modelling of the methylene blue removal from solution by adsorbent materials produced from sewage sludges. Biochemical Engineering Journal 15, 59e68. dos Santos, A.B., Cervanes, F.J., van Lier, J.B., 2007. Review paper on current technologies for decolourisation of textile wastewaters: perspective for anaerobic biotechnology. Bioresource Technology 98, 2369e2385. Smolders, G.J.F., Klop, J., van Loosdrecht, M.C.M., Heijnen, J.J., 1995. A metabolic model of the biological phosphorus removal process. Effect of the sludge retention time. Biotechnology and Bioengineering 48, 222e233. Su, K.Z., Yu, H.Q., 2005. Formation and characterization of aerobic granules in a sequencing batch reactor treating soybeenprocessing wastewater. Environmental Science Technology 39, 2818e2827. Wang, Q., Dua, G., Chena, J., 2004. Aerobic granular sludge cultivated under the selective pressure as a driving force. Process Biochemistry 39, 557e563. Wang, Y., Mu, Y., Zhao, Q.B., Yu, H.Q., 2006. Isotherms, kinetics and thermodynamics of dye biosorption by anaerobic sludge. Separation and Purification Technology 50, 1e7. van der Zee, F.P., Villaverde, S., 2005. Combined anaerobiceaerobic treatment of azo dyes-a short review of bioreactor studies. Water Research 39, 1425e1440.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 8 1 e4 5 8 9
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Disinfection by-products in filter backwash water: Implications to water quality in recycle designs N.J. McCormick, M. Porter, M.E. Walsh* Department of Civil and Resource Engineering, Dalhousie University, Halifax, Nova Scotia, Canada B3J 1Z1
article info
abstract
Article history:
The overall purpose of this research was to investigate disinfection by-product (DBP)
Received 24 September 2009
concentrations and formation potential in filter backwash water (FBWW) and evaluate at
Received in revised form
bench-scale the potential impact of untreated FBWW recycle on water quality in conven-
25 May 2010
tional drinking water treatment. Two chlorinated organic compound groups of DBPs
Accepted 26 May 2010
currently regulated in North America were evaluated, specifically trihalomethanes (THMs)
Available online 2 June 2010
and haloacetic acids (HAAs). FBWW samples were collected from four conventional filtration water treatment plants (WTP) in Nova Scotia, Canada, in three separate sampling
Keywords:
and plant audit campaigns. THM and HAA formation potential tests demonstrated that the
Filter backwash water (FBWW)
particulate organic material contained within FBWW is available for reaction with chlorine
Disinfection by-products (DBPs)
to form DBPs. The results of the study found higher concentrations of TTHMs and HAA9s in
FBWW recycle
FBWW samples from two of the plants that target a higher free chlorine residual in the wash water used to clean the filters (e.g., clearwell) compared to the other two plants that target a lower clear well free chlorine residual concentration. Bench-scale experiments showed that FBWW storage time and conditions can impact TTHM concentrations in these waste streams, suggesting that optimization opportunities exist to reduce TTHM concentrations in FBWW recycle streams prior to blending with raw water. However, mass balance calculations demonstrated that FBWW recycle practice by blending 10% untreated FBWW with raw water prior to coagulation did not impact DBP concentrations introduced to the rapid mix stage of a plant’s treatment train. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Conventional drinking water treatment trains have a filtration process which is considered a polishing step. Many modifications and additions have been made to water treatment processes, but filtration remains one of the fundamental technologies (Baruth, 2005). Over a filtration cycle in granular media filtration units, the filters become loaded with particles, organic material and bacteria. To maintain performance and treatment efficacy, backwashing with water is conducted to periodically remove contaminant material that is captured in
the filtration units. For individual plants, backwash cycles are scheduled anywhere from 1 to 7 days, depending on raw water quality and main treatment train design and operations (e.g., conventional versus direct filtration, coagulant dose). Waste filter backwash water (FBWW) is the largest primary liquid waste residual volume generated in a conventional or direct filtration plant. It is estimated that 3e10% of a plant’s total water production is consumed by this process (AWWA 1999). FBWW streams from surface water plants have been shown to be highly concentrated in total suspended solids (TSS), turbidity, and inorganics such as aluminum and/or iron that reflect
* Corresponding author. Tel.: þ1 902 494 8430; fax: þ1 902 494 3105. E-mail address:
[email protected] (M.E. Walsh). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.05.042
4582
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 8 1 e4 5 8 9
naturally occurring contaminants in the raw water as well as precipitates from the inorganic coagulants used during treatment (Cornwell et al., 2001; Edzwald et al., 2001; Arora et al., 2001; Bourgeois et al., 2004; Walsh and Gagnon, 2006; Walsh et al., 2008). The concentrated organic content of FBWW streams has been shown to be primarily in the particulate size fraction (i.e., >0.45 mm), with total organic carbon (TOC) concentrations consistently found to be significantly higher than DOC concentrations in several studies (Arora et al., 2001; Cornwell et al., 2001; Edzwald et al., 2001; Gottfried et al., 2008). Most FBWW streams from drinking water treatment plants will contain many of the above constituents, but the quality of FBWW will vary from plant to plant due to the differences in raw water quality and differences in treatment train design. Chlorine is the most commonly used disinfectant in North America. Chlorine reacts with biogenic organic matter, such as humic and fulvic acids which are present in all natural surface waters (Health Canada, 1996), and the substitution reactions that occur form halogenated organic compounds referred to as disinfection by-products (DBPs) (Rook, 1974; Bellar et al., 1974). Over the past 25 years, two prevalent classes of DBPs, trihalomethanes (THMs) and haloacetic acids (HAAs) have been reported due to the identification of potential adverse health risks, such as increased incidence of cancer and adverse human reproductive outcomes (Bull and Kopfler, 1991; Bull et al., 2001; Dodds et al., 2004). In the United States, Stage 1 of the Disinfectants-Disinfection ByProducts Rule (D/DBP Rule) has established maximum contaminant levels (MCLs) of 80 mg/L for total trihalomethanes (TTHMs) and 60 mg/L for five species of the haloacetic acids group (HAA5) (U.S. EPA, 1998), based on running annual averages of plant quarterly samples. The recent promulgation of the Stage 2 Disinfectants and Disinfection Byproducts Rule by the U.S. Environmental Protection Agency (U.S. EPA, 2006) builds on existing regulations by requiring water systems to comply with the MCLs for TTHM and HAA5 at specific monitoring sites in the distribution system, with monitoring frequency established by the size of the system. In Canada, the Federal-Provincial-Territorial Committee on drinking water quality has an established maximum acceptable concentration (MAC) for TTHMs of 100 mg/L and 80 mg/L for HAA5 (Health Canada, 2009). The European Union has no limit value for HAAs, but has a maximum contaminant level of 100 mg/L for the four THMs (Council Directive 98/83/EC, 1998). Results of water quality analysis conducted on waste FBWW samples collected from 25 water systems over a one year period showed TTHM concentrations ranged from non-detect (ND) to 198 mg/L and HAA6 concentrations ranged from ND to 211 mg/L (Cornwell et al., 2001). As highlighted in that study, elevated DBP concentrations in waste FBWW samples may be attributed to the use of chlorinated finished water for the purpose of filter backwashing. Chlorinated water from a plant’s clearwell is often used to backwash filters in drinking water treatment, although the exact contribution of the NOM trapped within filters and reactions with chlorinated wash water is complex and not well documented in literature. Work conducted by Edzwald et al. (2001) in this area examined the contribution to DBP formation of particulate versus adsorbed organic carbon contained within a filter related to differences in treatment train design (e.g., direct versus conventional filtration). The
results of that study determined that filter associated organic carbon contributes to DBP formation in waste FBWW streams. Other work has demonstrated that the type of disinfectant used in the main treatment train can impact THM and HAA concentrations in waste FBWW. A study by Walsh et al. (2008) found that waste FBWW from a conventional filtration plant that uses chloramines for disinfection had TTHM and HAA5 concentrations that were 76 and 96% lower, respectively, than those measured in waste FBWW samples from another conventional filtration plant that uses chlorine for disinfection. However, the exact contribution of chlorinated wash water on DBP formation in waste FBWW was not identified as part of that study, as the plant employing chlorine for disinfection also has a chlorine injection before filtration for manganese removal. The overall objective of this study was to investigate TTHM and HAA9 concentrations and formation potential in FBWW sampled from four different conventional filtration plant designs. For the purposes of this study, FBWW is defined as waste filter backwash water generated in dual-media filters that is not treated prior to recycle (e.g., no solid separation). Because DBPs are of potential concern when recycling FBWW streams within a water treatment plant, surrogate water quality parameters that can be easily measured in FBWW samples were evaluated as to the capability to present predictive tools for DBP concentrations in FBWW streams. In addition, as the formation of DBPs has been shown to be a function of chlorine dose, pH, temperature and contact time (Health Canada, 1996), this study also evaluated, at benchscale, TTHM and HAA9 concentrations in the FBWW samples under different storage conditions.
2.
Materials and methods
2.1.
Source waters and WTP descriptions
Raw source and FBWW samples were obtained from four surface water treatment plants in Nova Scotia, Canada, at three different sampling times throughout the year. Representative FBWW samples from all of the plants were obtained by evaluating backwash flow rates and selecting sample times and volumes that would provide a composite sample of the entire volume over the period of one complete backwash cycle. The FBWW samples were obtained directly from the overflow catchments of the filters during the backwash cycle. A summary of the process design of the four WTPs in Windsor, Bridgewater, Truro and New Glasgow, N.S. is presented in Table 1. All four plants are conventional treatment trains with either sedimentation (SED) or dissolved air flotation (DAF) clarification followed by dual-media filters comprised of anthracite and sand. All four plants use aluminum sulphate (alum) as the primary coagulant. Two of the plants studied have chlorine injection upstream of filtration for oxidation of iron and manganese. The New Glasgow WTP adds chlorine at a dose of 0.4 mg/L at the raw water intake to the plant, and the Truro WTP adds chlorine at a dose of 1.2 mg/L to the clarified water before filtration. All of the four plants in this study add chlorine to the clearwell for disinfection and the chlorinated water from the clearwell is used to backwash the filters.
4583
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 8 1 e4 5 8 9
Table 1 e Water treatment plant process design. WTP Plant
Windsor Bridgewater Truro New Glasgow
Design Capacity (MLD)
Treatment Train
Alum Dose (mg/L)
Chlorine Addition (Location) (mg/L)
Chlorine Addition to Clearwell (mg/L)
Clearwell Free Chlorine Residual Target (mg/L)
Backwash Frequency & (FBWW Volume Generated)
3.3 7.8 10.6 11.4
Coag/Floc/DAF/Filtr. Coag/Floc/SED/Filtr. Coag/Floc/SED/Filtr. Coag/Floc/DAF/Filtr.
40 mg/L 30 mg/L 48 mg/L 26 mg/L
e e 1.24 (clarified water) 0.44 (raw water)
2.84 2.25 1.70 1.88
1.0e1.5 1.2 1.4e1.6 1.5e1.7
2e4 days (60,500 L) 2e4 days (110,600 L) 4 days (190,000 L) 7 days (285,850 L)
2.2. FBWW DBP concentrations under variable storage conditions Samples of FBWW collected from each of the plants were analyzed for TTHM and HAA9 concentrations. Initial measurements (i.e., time ¼ 0 h) were taken from samples collected immediately following a backwash cycle at each plant. The FBWW samples were then transported back to the laboratory, and held in 1L flasks at room temperature for a maximum storage time of 30 h. To evaluate changes in DBP concentrations over different storage times, FBWW samples were collected from the storage bottles at timed intervals and TTHM, HAA9 and free chlorine residual measurements were performed in duplicate. For the two plants that add chlorine upstream of the filters in the main treatment train (e.g., New Glasgow and Truro WTPs), a second series of experiments were conducted to evaluate DBP concentrations over the same storage time, but under different storage conditions. Specifically, one FBWW sample was stored at room temperature in a 1-L Erlenmeyer flask under stagnant conditions (e.g., not mixed). A second FBWW sample was stored in five 250-mL Erlenmeyer flasks on a shaker table (Barnstead E-Class, Model, 2000; Fischer Scientific) constantly operated at a speed of 150 rpm. FBWW was sampled from a 250-mL flask at timed intervals and TTHM, HAA9 and free chlorine residuals were measured in duplicate.
2.3.
Analytical methods
TOC samples were collected headspace free in 40-mL precleaned glass vials and preserved with concentrated phosphoric acid to a pH < 2 and measurements were performed using a TOC-V CHP analyzer (Shimadzu Corporation, Kyoto, Japan). DOC samples were filtered through 0.45 mm pore size filters (ColeeParmer Nylon Membranes) that had been rinsed with deionised water as described in Standard methods for the Examination of Water and Wastewater (APHA, AWWA, and WEF, 1998). DOC measurements were taken using the same procedure and equipment as for the TOC samples. True color and UV254 were measured on the 0.45 mm filtrate samples using a HACH DR/4000 spectrophotometer (Hach Company, Loveland, CO) according to methodology as described in Standard Methods for the Examination of Water and Wastewater (APHA, AWWA, and WEF, 1998). Field and laboratory temperature and pH values were measured using an Orion model 210A pH meter with a combination electrode. Turbidity was measured using a HACH 2100AN turbidimeter (Hach Company, Loveland, CO) according to the instructions in the manufacturer’s
operating manual (Hach Company, Loveland, CO). Samples for total suspended solids (TSS) were filtered though a 1.5 mm glass microfibre filter (Fisher Scientific) and measured according to method 2540D in Standard Methods for the Examination of Water and Wastewater (APHA, AWWA, and WEF, 1998). Free chlorine residuals were measured using a HACH DR/4000 spectrophotometer (Hach Company, Loveland, CO) according to method 8021, Free Chlorine DPD Method, as described in the HACH DR/2010 spectrophotometer handbook (Hach Company, Loveland, CO). DBP formation potential tests were conducted on the raw water and FBWW samples collected from each plant using the Uniform Formation Conditions (UFC) test proposed by Summers et al. (1996) under defined conditions of 24 1 h incubation time, 20.0 1.0 C incubation temperature, 8.0 0.2 incubation pH and 1.0 0.4 mg/L 24-h free chlorine residual. These tests were conducted within 1e2 days after sampling. THM samples were collected headspace-free in 20-mL precleaned glass vials and preserved with ammonium chloride, sodium thiosulphate and acidified to pH 4.5 with hydrochloric acid. THM analysis was conducted as per U.S. EPA Method 551. Gas chromatographic analyses of THMs were performed using a Hewlett Packard 5890 Series II-Plus GC, equipped with a DB-5 column for primary analysis, and a DB-1701 column for confirmation. A Fisons Mass Spectrophotometer (Trio 1000) was periodically used for compound identification. The THMs measured were chloroform, bromoform, dichlorobromomethane, and dibromochloromethane, and collectively referred to as total trihalomethanes (TTHMs). HAA samples were collected headspace-free in 20-mL precleaned glass vials and preserved with ammonium chloride. HAA concentrations were measured according to EPA Method 552.2. Gas chromatographic analyses of HAAs were performed using a Hewlett Packard 5890 Series II-Plus GC, equipped with a DB-5 column for primary analysis, and a DB-1701 column for confirmation. The HAAs measured were chloroacetic acid, bromoacetic acid, dichloroacetic acid, trichloroacetic acid, bromochloroacetic acid, dibromoacetic acid, bromodichloracetic acid, chlorodibromoacetic acid and tribromoacetic acid, and collectively referred to as HAA9.
3.
Results and discussion
3.1.
Water quality analysis
A summary of the water quality results for the raw water and FBWW samples collected from the four water treatment
4584
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 8 1 e4 5 8 9
facilities is presented in Table 2. The four source waters were found to be low in turbidity (0.8e2.5 NTU) and exhibited neutral pH (6.1e6.8). The source waters showed variations in true color, ranging from 16 to 52 TCU. TOC concentrations of the source water for the four facilities ranged from 4.0 to 5.6 mg/L. DOC concentrations ranged from 4.0 to 5.6 mg/L, and UV254 values were observed from 0.118 to 0.353 cm-1, suggesting that most of the organics in the source waters are in the dissolved form. All of the raw waters sampled had specific UV absorbance (SUVA) at 254 nm values ranging from 3.02 to 4.06 L/mg m, indicating that the raw water samples were comprised of a varied mixture of hydrophobic and hydrophilic organic compounds. The FBWW samples from all four plants were highly concentrated in particulate matter compared to the raw water samples, with turbidity ranging from 26 to 73 NTU, and TSS concentrations ranging from 129 to 414 mg/L. In comparison to the raw source water of each of the plants, analysis of the FBWW samples showed high TOC concentrations (20e85 mg/ L), medium DOC concentrations (3.0e4.8 mg/L) and low UV254 values (0.049e0.062 cm-1). These results agree with the observation of previous work that most of the organic material in FBWW is contained in the particulate phase (Cornwell et al., 2001; Edzwald et al., 2001; Tobiason et al., 2003; Gottfried et al., 2008). All of the FBWW samples were found to have SUVA values ranging from 1.41 to 1.80 L/mg$m, indicating that the FBWW contained mostly non-humic and low hydrophobicity organic material. Collectively, the FBWW characterization results of this study are in agreement with previous studies (Edzwald et al., 2001; Cornwell et al., 2001; Gottfried et al., 2008), and demonstrate that the organic material present in FBWW is primarily enmeshed in alum flocs.
3.2.
DBP formation potential study
3.2.1.
DBP formation potential in raw and FBWW samples
THM and HAA formation potential measurements (THMfp, HAAfp) were conducted on the raw and FBWW water samples from the four water treatment facilities in this study (Fig. 1). As presented in Table 3, initial DBP concentrations in the FBWW samples were low compared to the DBP formation potentials measured in these samples. Although true formation potential measurement should take into account both initial and final DBP concentrations, the initial TTHM and HAA9 concentrations measured in the FBWW samples accounted for less than 10% of the DBP concentrations formed
Fig. 1 e DBP Formation Potential of raw water and FBWW samples (N [ 4).
under UFC conditions. As such, the formation potential data presented in Fig. 1 for the FBWW samples was not adjusted for initial DBP concentrations present in these samples. All of the water samples tested showed the potential to form very high concentrations of DBPs. Overall, the THMfp exceeded the HAAfp at all plants for both raw water and FBWW samples. In addition, the FBWW samples showed a higher potential to form both THMs and HAAs than the raw water samples. The study found that FBWW samples from two of the plants (Bridgewater and Windsor) required much higher doses of chlorine (45 and 46 mg/L, respectively) than Truro and New Glasgow (14 and 15 mg/L, respectively) to achieve a free chlorine residual of 1.0 0.4 mg/L after 24 h incubation. As outlined in Table 1, the lower chlorine demand of the Truro and New Glasgow FBWW samples compared to the Windsor and Bridgewater FBWW samples to reach uniform formation conditions may be due to differences in clearwell free chlorine residual targets in the WTPs. Both the Truro and New Glasgow plants target a free chlorine residual of 1.4e1.7 mg/L in the clearwell, while Bridgewater and Windsor have a lower 1.0e1.5 mg/L free chlorine residual target in their finished water. However, as a result of the uniform formation conditions of the DBP formation potential test used in this study, there was no distinction found between THM or HAA formation potential observed in the FBWW samples collected from the four plants.
3.2.2. Correlation of DBP formation potential with organic water quality parameters In drinking water treatment, it has been shown that UV254 is a good surrogate parameter for predicting THMfp in source or
Table 2 e Raw water and filter backwash water characteristics (average ± standard deviation). Analyte
Units
Bridgewater WTP Raw
pH TSS Turbidity True Color UV254 TOC DOC SUVA Aluminum
6.1 0.3 mg/L NTU TCU cm-1 mg/L mg/L mg/L$m mg/L
1.1 29.8 0.187 4.2 4.4 3.85 0.1
0.4 12.2 0.056 0.8 0.9 1.22 0.1
FBWW 6.3 186.7 57.1 2.0 0.049 44.8 3.0 1.60 18.4
0.3 26.3 17.9 1.0 0.023 17.9 0.5 0.48 17.6
Windsor WTP Raw 6.2 0.3 0.8 0.4 51.9 9.9 0.210 0.094 5.6 1.0 5.6 0.9 3.91 1.87 0.1 0.1
FBWW 6.1 414.0 57.9 4.4 0.062 85.4 4.8 1.41 24.3
0.2 367.6 45.3 3.9 0.061 73.2 4.3 0.83 5.7
Truro WTP Raw 6.4 0.1 1.6 0.5 33.4 9.7 0.205 0.063 5.0 1.9 5.1 1.7 4.06 0.37 0.1 0.1
New Glasgow WTP
FBWW 6.7 129.0 25.8 2.3 0.056 20.3 3.2 1.80 20.3
0.1 30.4 5.8 1.4 0.027 5.4 0.8 0.91 5.3
Raw 6.8 0.1 2.5 1.7 15.9 7.6 0.118 0.007 4.0 0.7 4.0 0.7 3.02 0.61 N/D
FBWW 6.7 227.0 73.2 2.6 0.053 23.2 3.2 1.67 32.0
0.1 120.6 41.8 2.1 0.005 9.3 0.5 0.12 4.2
4585
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 8 1 e4 5 8 9
Table 3 e TTHM and HAA9 concentrations measured in FBWW samples over time. WTP Plant
Elapsed Time (h)
Plants with Chlorine Injection in Clearwell Bridgewater 0 0e8 8e30 Windsor 0 0e8 8e30
TTHM Average (mg/L)
TTHM Range (mg/L)
17.9 17.7 16.8 19.2 15.9 23.3
11.7e24.5 11.8e23.4 8.7e27.8 5.8e28.9 6.0e21.8 18.0e31.2
Plants with Chlorine Injection in Clearwell and Upstream of Filters Truro 0 101.4 0e8 145.7 8e30 174.6 New Glasgow 0 107.6 0e8 143.2 8e30 158.1
treated water (Edzwald et al., 1985). This relationship between DBP formation potential and UV254 measurements was found to exist for the raw water samples evaluated in this study (Fig. 2a and b). In general, as measured UV254 values increased, the DBP formation potential values increased as well. As demonstrated in Fig. 2a, there was a strong linear correlation between UV254 and THMfp (R2 ¼ 0.88) in all four of the source waters sampled. A linear correlation was also observed between HAAfp and UV254 (R2 ¼ 0.77) in all four of the source waters, as demonstrated in Fig. 2b. Increasing UV254 measurements suggest there is more dissolved organic
Fig. 2 e Raw water THM formation potential (a) and HAA formation potential (b) versus UV254.
45.6e153.6 63.5e206.6 112.6e228.5 71.5e151.9 91.9e188.7 81.8e216.8
HAA9 Average (mg/L) 3.7 4.7 7.5 3.6 4.0 2.3
115.4 124.1 110.7 62.2 64.4 57.0
HAA9 Range (mg/L) 0.1e8.0 1.7e8.7 5.0e13.6 1.3e5.4 1.2e6.7 1.1e4.5
37.8e193.6 57.7e209.3 43.5e225.0 2.4e127.9 20.8e115.4 17.4e142.5
material available for reaction with residual chlorine, and therefore a higher potential to form DBPs. As expected, such a correlation was not found between THM and HAA formation potential and UV254 in the FBWW samples evaluated in this study (R2 < 0.1) due to the fact that most of the organic material present in FBWW is contained in the particulate phase, with dissolved organic material primarily enmeshed in alum floc. As presented in Fig. 3 a & b, a moderate linear correlation was found between the FBWW TOC concentrations and THMfp (R2 ¼ 0.52) and HAAfp (R2 ¼ 0.48). Although a weaker correlation was found to exist with using TOC as a predictive water quality parameter for DBP formation potential in the FBWW samples compared to the correlation between UV254 measurements and DBP formation potential in the raw water samples, it demonstrates that the organic material enmeshed in alum floc is available for reaction with chlorine to form THMs and HAAs.
3.3.
FBWW storage conditions bench-scale study
3.3.1.
Impact of FBWW storage time on DBP concentrations
The TTHM and HAA9 concentrations in collected FBWW samples were measured over time for the four water treatment facilities used in the study. The TTHM and HAA9 concentrations were measured in duplicate on two separate FBWW samples from each plant. The results are presented in Table 3. The plants that do not add chlorine upstream of the filters (Bridgewater and Windsor) are listed first, and the plants that do add chlorine upstream of the filters (Truro and New Glasgow) are listed second. FBWW samples from New Glasgow and Truro were found to have significantly higher initial concentrations of TTHMs and HAA9s than the FBWW samples from the Windsor and Bridgewater plants. This could be due to differences between free chlorine residuals and/or differences between DBP concentrations in the wash water used to clean the filters at each of the plants. Measurements taken on both the New Glasgow and Truro FBWW samples at time of collection showed 0.7e1.0 mg/L free chlorine residual, respectively. In contrast, free chlorine residuals measured on FBWW samples at time of collection from the Windsor and Bridgewater WTPs were lower, at 0.1 and 0.4 mg/L, respectively. Although the
4586
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 8 1 e4 5 8 9
Fig. 3 e FBWW THM formation potential (a) and HAA formation potential (b) versus TOC.
chlorine addition to each plant’s clearwells were recorded when samples were collected for analysis (Table 1), the corresponding free chlorine residual in the respective plant clearwells were not monitored during the study. However, data obtained from the plants with regards to free chlorine residual targets in the clearwell at the time of this study (Table 1) showed that both the New Glasgow and Truro plants target a free chlorine residual of 1.4e1.7 mg/L, while in contrast the Windsor and Bridgewater plants target a lower free chlorine residual of 1.0e1.5 mg/L. The higher free chlorine residual in the wash water of the New Glasgow and Truro plants compared to that of the Windsor and Bridgewater plants would possibly explain the differences in initial DBP concentrations measured in the FBWW samples collected in this study. In light of the free chlorine targets of each of the plants, it is also possible that at the time of the study, the clean, backwash water (i.e., clearwell water) used at the Truro and New Glasgow plants contained higher TTHM and HAA9 concentrations compared to DBP concentrations in the wash water used to clean the filters at the Bridgewater and Windsor WTPs. DBP analysis of the clearwell water was not conducted as part of this study. However, DBP data of quarterly clearwell samples collected during the same time period of this study was obtained from each of the plants. This dataset showed that TTHM and HAA9 concentrations in clearwell samples taken from the Windsor, Bridgewater and Truro WTPs ranged from 21.3 to 52.8 mg/L and 34.3 to 58.6 mg/L, respectively. Data from the New Glasgow plant showed TTHM and HAA9 concentrations in clearwell samples averaged 118 60.2 mg/L and 47.7 14.6 mg/L, respectively. The higher TTHM
concentrations in the New Glasgow clearwell water compared to the other plants would offer possible explanation for the elevated initial TTHM concentrations found in the FBWW samples from this plant. However, a similar correlation between wash water and FBWW DBP concentrations and was not seen for the Truro plant, that also showed elevated initial TTHM and HAA9 concentrations in FBWW samples compared to the Windsor and Bridgewater plants. Further investigation would also be warranted to determine if the additional chlorine injection points upstream of the filters at the Truro or New Glasgow plants significantly impact DBP concentrations in the FBWW generated at these plants. Within 4e8 h after sampling, the TTHM concentrations increased to a maximum of 150e250 mg/L for the New Glasgow and Truro FBWW samples. A much lower TTHM concentration of less than 30 mg/L was found after the same retention time for the Bridgewater and Windsor FBWW samples. The retention time of 4e8 h was observed to correspond to when the free chlorine residual in the FBWW dropped to below 0.1 mg/L. This demonstrates that most of the chlorine that will react with the organics in the water to form DBPs will happen in the first few hours. For both the New Glasgow and Truro FBWW samples, TTHM concentrations were found to remain higher than 80 mg/L after a 26 to 30 h retention time. For both the Windsor and Bridgewater FBWW samples, the TTHM concentrations did not increase beyond 30 mg/L once the chlorine residual decreased to less than 0.1 mg/L after a 4e 8 h retention time. These results show that once TTHMs are formed in FBWW, they remain relatively stable after chlorine residuals are diminished. The HAA9 concentrations were also measured over time for each of the four water treatment facilities. The results are presented in Table 3. Similar results to the TTHM measurements were found. Specifically, the FBWW samples from the New Glasgow and Truro plants showed higher initial HAA9 concentrations compared to the FBWW samples collected from the Windsor and Bridgewater plants. Within 4e6 h after sampling, the HAA9 concentrations increased to a maximum of 50e200 mg/L for the New Glasgow and Truro FBWW samples. In contrast, a maximum HAA9 concentration of 14 mg/L for the Bridgewater and Windsor FBWW samples was found. HAA9 concentrations in the New Glasgow and Truro FBWW samples remained higher than 30 mg/L throughout the 26e30 h retention time of this study. In contrast, HAA9 concentrations of less than 15 mg/L were observed after the same retention time in the Windsor and Bridewater FBWW samples. These results show that, like TTHM concentrations, HAA9 concentrations in FBWW samples remain relatively stable after chlorine residuals are diminished.
3.3.2. Impact of FBWW mixing conditions on DBP concentrations A second series of experiments were run with the FBWW samples collected from the New Glasgow and Truro WTPs to evaluate potential impact of storage conditions on DBP concentrations. The results of the experiments that evaluated TTHM concentrations over retention time under both agitated and stagnant storage conditions for the Truro FBWW samples are presented in Fig. 4. Samples that were held stagnant (i.e., not mixed) are shown as shaded points and agitated (i.e.,
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 8 1 e4 5 8 9
Fig. 4 e TTHM concentrations over time in stagnant and agitated FBWW samples from Truro WTP. mixed) samples are shown as outlined points. Each different marker type on the graphs represents a specific trial. In general, the stagnant and agitated FBWW samples showed a slight increase in TTHM concentrations in the first 4e6 h as the chlorine residual in the FBWW samples continued to react with the organics. While the TTHM concentrations in the stagnant FBWW samples varied slightly over time, the TTHM concentrations in the agitated samples were found to decrease significantly with increased retention time. Similar experimental results on TTHM concentrations were found with the New Glasgow FBWW samples, however the data is not presented. HAA9 concentrations in the stagnant and agitated FBWW samples taken from both New Glasgow and Truro WTPs showed varying behaviour in each trial, with HAA9 concentrations increasing or decreasing over time. In general, no clear trend was observed with either the New Glasgow or Truro FBWW samples with regards to HAA9 concentrations measured over time under the stagnant or mixing conditions evaluated. Results of the New Glasgow experiments are presented in Fig. 5. The results of this set of experiments suggest that TTHM concentrations in stored FBWW streams are more greatly impacted by constant mixing than HAA9 concentrations. Under constant mixing conditions (e.g., agitated FBWW samples), TTHM concentrations were reduced in the liquid phase (e.g., FBWW) while HAA9 concentrations remained relatively constant in the liquid phase. The volatility of the
4587
DBP compounds measured in this study in terms of Henry’s law constant (HA) may offer explanation for these results. TTHM concentrations in the FBWW samples were primarily comprised of chloroform, which is very volatile as represented by a high Henry’s Law Constant of 3.67 10-3 atm m3/mol (Gossett, 1987). In contrast, the HAA9 concentrations in the FBWW streams were primarily comprised of dichloroacetic acid (DCAA) and trichloroacetic acid (TCAA), which have Henry’s Law Constants of 3.52 y 10-7 and 1.35 10-8 atm m3/ mol, respectively (Gossett, 1987). The reduction in TTHM concentrations could have been due to transfer from the liquid phase to the gas phase due to the established higher volatility of chloroform compared to DCAA and TCAA. For plants that have elevated TTHM concentrations in FBWW streams, these results suggest that storage conditions could impact ultimate TTHM concentrations that get returned into the main treatment train in FBWW recycle designs. Further study should be conducted on full-scale plants to confirm these bench-scale experimental results, as recycling FBWW that has been continuously mixed in an open vessel could result in significantly lower TTHM concentrations that are introduced into the main treatment train with FBWW recycle streams.
3.3.3.
Mass balance calculations
To model DBP concentrations that would be expected in plant designs employing FBWW recycle, mass balance calculations were performed using equations presented by Cornwell and MacPhee (2001) that investigated the impact of FBWW recycle on sedimentation basin and plant performance. Specifically, a mass balance around the raw and FBWW recycle feedstreams was used to calculate the coagulant rapid mix and flocculation basin influent DBP concentrations using the following equation: ðCraw ÞðQraw Þ þ ðCbw ÞðQbw Þ ¼ ðQÞðCÞ
(1)
where: Craw¼ plant influent DBP concentration Qraw ¼ plant influent raw water flow Cbw ¼ FBWW DBP concentration Qbw ¼ FBWW flow Q ¼ Rapid Mix Tank influent flow C ¼ Rapid Mix Tank influent DBP concentration Assuming negligible DBP concentrations exist in the source water to a plant (e.g., Craw ¼ 0 mg/L for both TTHM and HAA9), Equation (1) can be simplified as presented in Equation (2): C ¼ ðCbw ÞðRÞ
Fig. 5 e HAA9 concentrations over time in stagnant and agitated FBWW samples from New Glasgow WTP.
(2)
where: R ¼ FBWW recycle rate Equation (2) was used to calculate DBP concentrations (e.g., TTHM and HAA9) that would be introduced to the rapid mix stage of a plant’s treatment train employing 10% FBWW recycle (i.e., untreated FBWW mixed with raw water prior to coagulation) through the introduction of TTHM and HAA9 concentrations already formed in FBWW streams. These calculations were performed using FBWW TTHM and HAA9 maximum concentrations found in the batch retention experiments for each of the plant source waters, or worse-
4588
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 8 1 e4 5 8 9
case scenario. It is recognized that both TTHM and HAA9 concentrations would be expected to be variable, depending upon storage time before recycling and mixing conditions, as demonstrated in the experimental work presented earlier. For the Bridgewater and Windsor WTPs, blending 10% untreated FBWW with the raw water would result in negligible TTHM and HAA9 concentrations less than 2 mg/L introduced at the rapid mix stage of the treatment train. These low concentrations reflect the lower preformed DBP concentrations in the FBWW recycle stream found in this study. For both of these plant process designs, the introduction of additional organic material contained within an untreated FBWW stream would not be expected to contribute to additional DBP formation, due to the absence of chlorine injection until after filtration, assuming that organic removal efficiencies are maintained within the main treatment train. For the Truro and New Glasgow WTPs, blending 10% untreated FBWW with the raw water would result in TTHM and HAA9 concentrations less than 20 mg/L introduced at the rapid mix stage of the treatment train. With the Truro plant design, the mixture of raw and FBWW recycle water would not be expected to contribute significant additional increases to DBP formation, assuming that flocculated organic removal efficiency is maintained in clarifier operations prior to the chlorine injection point before filtration. However, for the New Glasgow plant design, the mixture of raw and FBWW recycle water could also potentially contribute to additional DBP formation above this baseline of 20 mg/L, due the presence of organic material contained within the FBWW that would potentially react with the chlorine added to the plant raw water intake.
4.
Conclusions
This study evaluated FBWW samples from four surface water treatment plants in terms of DBP concentrations and formation potential to examine the potential impacts to water quality in plants that recycle untreated FBWW to the main treatment train. The following outlines the main findings of the study: The higher THMfp and HAAfp measurements found with all of the FBWW samples in comparison to those measured in the raw water samples demonstrate that FBWW particulate organic material, as indicated with significantly higher TOC versus DOC concentrations in the FBWW samples evaluated in this study, is available for reaction with chlorine to form THMs and HAAs. FBWW samples collected from two of the four plants in the study showed higher initial TTHM and HAA9 concentrations compared to the DBP concentrations measured in the FBWW samples collected from the other two plants. Differences in free chlorine residual targets in the clearwell, resulting in higher concentrations of chlorine in the washwater used to clean the filters at each plant, may account for the variability observed in the initial DBP concentrations measured in the FBWW samples in this study. The results of bench-scale experiments to investigate the impact of storage conditions on DBP concentrations in
FBWW demonstrated that FBWW storage conditions prior to recycle can impact TTHM concentrations, and that opportunities may exist to reduce TTHM concentrations at fullscale plants by constantly mixing FBWW in an open vessel prior to the recycle of this waste residual stream. Mass balance calculations using maximum TTHM and HAA9 concentrations measured in FBWW samples evaluated in this study demonstrated that the recycle of FBWW (by blending 10% untreated FBWW with raw water) would not impact DBP concentrations introduced to the rapid mix stage of a plant’s treatment train. However, further study is required to evaluate plant design with regards to chlorine injection points to determine the potential impacts on the formation of THM and HAA species due to the introduction of additional organic material in untreated FBWW recycle streams.
Acknowledgements The authors acknowledge funding provided by the Natural Science and Engineering Research Council of Canada (NSERC) for this study. The authors gratefully acknowledge the assistance from the municipalities of Bridgewater, Windsor, Truro, and New Glasgow of Nova Scotia, Canada.
references
American Public Health Association (APHA), 1998. In: Clescert, L., Greenberg, A., Eaton, A. (Eds.), Standard Methods for the Examination of Water and Wastewater, 20th ed. APHA, Washington, USA. Arora, H., Di Giovanni, G., LeChevallier, M., 2001. Spent filter backwash water contaminants and treatment strategies. Journal AWWA (American Water Works Association) 93 (5), 100e112. AWWA, 1999. Water Treatment Plant Residual Management, Water Quality & Treatment, a Handbook of Community Water Supply. McGraw-Hill, New York. 16.1e16.51. Baruth, Edward E, 2005. Water Treatment Plant Design, fourth ed. McGraw-Hill, Inc.. Bellar, T.A., Lichtenberg, J.J., Kroner, R.C., 1974. The occurrence of organohalides in chlorinated drinking waters. Journal AWWA (American Water Works Assiciation) 66, 703. Bourgeois, J.C., Walsh, M.E., Gagnon, G.A., 2004. Comparison of process options for treatment of water treatment residual streams. Journal of Environmental Engineering and Science 3 (6), 408e416. Bull, R.J., Krasner, S.W., Daniel, P.A., Bull, R.D., 2001. Health Effects and Occurrence of Disinfection By-Products. American Water Works Association Research Foundation (AWWARF), Denver, CO, USA. Bull, R.J., Kopfler, F.C., 1991. Health Effects of Disinfectants and Disinfection By-Products. American Water Works Association Research Foundation (AWWARF), Denver, CO, USA. Canada, Health, 1996. A one-year survey of halogenated disinfection by-products in the distribution system of treatment plants using three different disinfection processes. Minister of Public Health Works and Government Services Canada.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 8 1 e4 5 8 9
Cornwell, D., MacPhee, M., McTigue, N., Arora, H., DiGiovanni, G., LeChevallier, M., Taylor, J., 2001. Treatment Options for Giardia, Cryptosporidium, and Other Contaminants in Recycled Backwash Water. American Water Works Association Research Foundation (AWWARF), Denver, CO, USA. Cornwell, D.A., MacPhee, M.J., 2001. Effects of spent filter backwash recycle on Cryptosporidium removal. Journal AWWA (American Water Works Association) 93 (4), 153e162. Council Directive 98/83/EC, 1998. Concerning the Quality of Water Intended for Human Consumption. 3rd November, 1998. Dodds, L., King, W., Allen, A.C., Armson, B.A., Fell, D.B., Nimrod, C., 2004. Trihalomethanes in public water supplies and risk of stillbirth. Epidemiology 15 (2), 179e186. Edzwald, J.K., Becker, W.C., Wattier, K.L., 1985. Surrogate parameters for monitoring organic matter and THM precursors. Journal AWWA (American Water Works Association) 77 (4), 122e132. Edzwald, J.K., Tobiason, J.E., Kelley, M.B., Dunn, H.J., Galant, P.B., Kaminski, G.S., 2001. Impacts of Filter Backwash Recycle on Clarification and Filtration. American Water Works Association Research Foundation (AWWARF), Denver, CO, USA. Gossett, J.M., 1987. Measurement of Henry’s Law constants for C1 and C2 chlorinated hydrocarbons. Environmental Science Technology 21 (2), 202e208.
4589
Gottfried, A., Shepard, A.D., Hardiman, K., Walsh, M.E., 2008. Impact of recycling filter backwash water on organic removal in coagulationsedimentation processes. Water Research 42 (18), 4683e4691. Health Canada, 2009. Guidelines for Canadian Drinking Water Quality. Available from: http://www.hc-sc.gc.ca/ewh-semt/ water-eau/drink-potab/guide/index_e.html [cited 30.07.09]. Rook, J.J., 1974. Formation of haloforms during chlorination of natural waters. Water Treatment Examination 23, 234. Summers, R.S., Hooper, S.M., Solarik, G., Owen, D., 1996. Assessing DBP yield: uniform formation conditions. Journal AWWA (American Water Works Association) 81 (7), 80e93. Tobiason, John E., Edzwald, James K., Levesque, Benjamin R., Kaminski, Gary K., Dunn, Howard J., Galant, Peter B, 2003. Fullscale assessment of waste and filter backwash recycle. Journal AWWA (American Water Works Association) 95 (7), 80e93. U.S. EPA, 1998. Disinfectants and disinfection byproducts; final rule. Federal Register 63 (241), 69390. Walsh, M.E., Gagnon, G.A., Alam, Z., Andrews, R.C., 2008. Biostability and disinfectant by-product formation in drinking water blended with UF treated filter backwash water. Water Research 42 (8e9), 2135e2145. Walsh, M.E., Gagnon, G.A., 2006. Blending membrane treated WTP waste residuals with finished water: impacts to water quality and biofilm formation. Journal of Water Supply: Research & Technology e AQUA 55 (5), 321e334.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 9 7 e4 5 0 4
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Effect of some parameters on the formation of chloroform during chloramination of aqueous solutions of resorcinol Nicolas Cimetiere, Florence Dossier-Berne, Joseph De Laat* Universite´ de Poitiers, Laboratoire de Chimie et Microbiologie de l’Eau, CNRS UMR 6008, Ecole Supe´rieure d’Inge´nieurs de Poitiers, 40, Avenue du Recteur Pineau, 86022 Poitiers Cedex, France
article info
abstract
Article history:
The effects of various factors (N/Cl ratio used to prepare monochloramine, monochloramine
Received 5 February 2010
doses, pH and contact time) on the monochloramine demand and on the chloroform yield
Received in revised form
during chloramination of resorcinol have been investigated. Chloramination experiments
3 May 2010
were carried out at 24 1 C, at pH values ranging from 6.5 to 12 using a bicarbonate/
Accepted 7 June 2010
carbonate buffer and preformed monochloramine solutions prepared at pH 8.5 with N/Cl
Available online 12 June 2010
ratios ([NH4Cl]0/[Total free Cl2]0 ranging from 1.0 to 150 mol/mol). Kinetic experiments
Keywords:
a slow increase of the monochloramine consumption with reaction time. The monochlor-
Monochloramine
amine demands after reaction times of 7 days ([Resorcinol]0 ¼ 100 mM) and 14 days
Monochloramine demand
([Resorcinol]0 ¼ 5 mM) were equal to 8.5 mol of NH2Cl/mole of resorcinol and were higher than
Trihalomethanes
the chlorine demands (z7.3 mol/mol). Chloroform yields from monochloramination of
([Resorcinol]0 ¼ 5 or 100 mM, [NH2Cl]0/[Resorcinol]0 ¼ 20 mol/mol, pH ¼ 8.5 0.1) showed
Free chlorine
resorcinol were lower than 8% (<80 mmol of CHCl3/mole of resorcinol) and were less than
Monochloramine hydrolysis
the yields obtained by chlorination (0.9e0.95 mol/mol). Chloroform productions increased with increasing monochloramine dose and reaction time and decreased with increasing pH values within the pH range 6.5e10. Chloroform formation markedly decreased when the N/Cl ratio increased from 1 to 1.5 mol/mol and was suppressed at N/Cl > 100 mol/mol. The data obtained in the present work suggest that free chlorine released from monochloramine hydrolysis plays a significant role on the formation of chloroform during chloramination of resorcinol at N/Cl ratios close to unity (1.0 < N/Cl < 1.5). ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Chlorination is the most widely used process for drinking water disinfection. However, reactions between chlorine and naturally occurring organic matter, bromide and iodide produce numerous disinfection by-products (DBPs) of public health concern. Since the discovery of trihalomethanes (THMs) in chlorinated drinking waters in 1974 (Bellar et al., 1974; Rook, 1974), more than 600 DBPs have been reported in the literature (Richardson et al., 2007; Hrudey, 2009). Because
of health risks associated with long term exposure to DBPs, most developed nations have set regulations or guidelines to minimize the concentrations of THMs and of other DBPs in finished drinking water. Many utilities have adopted the use of monochloramine as a secondary disinfectant in order to meet the current or future regulations concerning the concentrations of DBPs in drinking water. Monochloramine is an efficient disinfectant. Monochloramine is less reactive with natural organic matter (NOM) and its use provides long lasting residuals in the water
* Corresponding author. Tel.: þ33 5 49 45 39 21; fax: þ33 5 49 45 37 68. E-mail address:
[email protected] (J. De Laat). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.010
4498
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 9 7 e4 5 0 4
distribution systems. Total concentrations of THMs, haloacetic acids or of other DBPs in chloraminated drinking waters or chloraminated solutions of natural organic matter (NOM) are lower than in chlorinated ones (Diehl et al., 2000; Hua and Reckhow, 2007; Yang et al., 2007; Goslan et al., 2009). The production and the speciation of DBPs formed by chloramination of NOM depend on various parameters such as temperature, pH, bromide ion concentrations, nature of NOM, contact time, application modes of monochloramine (introduction of free chlorine into ammonia containing water or introduction of preformed monochloramine solution) and the N/Cl initial ratios used for the production of monochloramine (Qi et al., 2004; Yang et al., 2007, 2008). The kinetics and the reaction pathways for the formation of DBPs by chloramination of NOM are not well documented. As demonstrated by Valentine’s research group, the oxidation of NOM by monochloramine can be attributed to two distinct reaction pathways: direct reactions of monochloramine with NOM and indirect reactions involving free chlorine liberated from monochloramine hydrolysis (Jafvert and Valentine, 1992; Duirk et al., 2005; Duirk and Valentine, 2006): NH2 Cl þ H2 O$HOCl þ NH3
(1)
The value of the equilibrium constant for reaction (1) (K ¼ 5.1 1012 M; Morris and Isaac, 1983) indicates that free chlorine should be present at trace levels in organic-free solutions of monochloramine. Free chlorine concentrations depend on the total concentration of oxidants, pH and the initial N/Cl ratios used to prepare monochloramine solutions. In the presence of organic compounds, free chlorine can be consumed by the organic solutes. This reaction shifts the equilibrium (1) to the right and thus increases the decay rate of monochloramine. The contribution of free chlorine to the degradation of organic solutes and to the formation of DBPs during chloramination depends on various parameters such as the free chlorine concentration and the relative rate constants for the reactions of monochloramine and chlorine with the organic solutes. In order to better understand the role played by free chlorine on the formation of DBPs during chloramination of organic compounds, solutions of resorcinol have been chloraminated in the present study and the production of chloroform has been determined at various N/Cl ratios. Resorcinol, which is a meta-polyhydroxybenzene, has been used in this work as a model compound for the following reasons. Meta-polyhydroxybenzene moieties in NOM have been assumed to be responsible for the formation of THMs upon chlorination (Rook, 1977; Norwood et al., 1980). Resorcinol is one of the most efficient precursors of chloroform upon chlorination, with the formation of 0.9e0.95 mol of chloroform per mole of resorcinol (Norwood et al., 1980; De Laat et al., 1982; Boyce and Hornig, 1983; Gallard and von Gunten, 2002). The rate of formation of chloroform is very fast because chloroform production is possible at very low chlorination doses ([Chlorine]0/[Resorcinol]0 < 1 mol/mol) and can compete with the very fast reaction of formation of monochloramine when solutions of resorcinol are chlorinated in the presence of ammonia (De Laat et al., 1982). Furthermore, the absolute second-order rate constants for the reaction of hypochlorous acid (Rebenne et al., 1996) and monochloramine
(Cimetiere et al., 2009) with the acidebase forms (Ar(OH)2, Ar (OH)O and Ar(O)2) of resorcinol, 4-chlororesorcinol and 4,6dichlororesorcinol are now known. The apparent secondorder rate constants for the reaction of free chlorine and monochloramine with resorcinol in the pH range 7e11 indicate that free chlorine is z5 orders of magnitude more reactive than monochloramine (Cimetiere et al., 2009). These kinetic data suggest that free chlorine reactions may contribute to the formation of chloroform during chloramination of resorcinol. Therefore, this work was designed to examine the effect of some parameters (N/Cl ratio, pH, monochloramine dose, reaction time) on the production of chloroform during chloramination of resorcinol.
2.
Experimental section
2.1.
Chemicals and preparation of solutions
All solutions were prepared from reagent-grade chemicals and purified water delivered by a Millipore system (Milli RX75/ Synergy 185). Stock solutions of resorcinol (SigmaeAldrich, ACS reagent, >99%), ammonium chloride (Fisher Scientific, >99%) and free chlorine (NaOCl, 13%, Acros Organics) were stored in the dark at 4 C. Solutions of monochloramine (0.1e4 mM) were prepared daily by slowly adding free chlorine into an ammonium chloride solution in a well-stirred reactor. The pH of the sodium hypochlorite and ammonium chloride solutions were adjusted to 8.5 0.1 just before use. Solutions of monochloramine were prepared by using a nitrogen to chlorine molar ratio (N/Cl ¼ [Ammonium chloride]0/[Total chlorine]0) ranging from 1.0 to 150 mol/mol. Iodometric titrations and spectrophotometric analyses of freshly prepared solutions of NH2Cl showed that free chlorine introduced in the solution of ammonium chloride was quantitatively converted into NH2Cl at pH z 8.5. Phosphate buffer was not used in the present work because phosphate ions are known to catalyze the autodecomposition of NH2Cl (Valentine and Jafvert, 1988) and are suspected to promote the resorcinol ring-opening during chlorination (Heasley et al., 1989). Due to the low catalytic effect of carbonate ion toward monochloramine disproportionation (Jafvert and Valentine, 1992), a carbonate buffer (20 mM) was used in the present work. Initial pH was adjusted with hydrochloric acid or sodium hydroxide. All glasswares were cleaned with a solution of monochloramine (z1 mM) and were kept under monochloramine before the experiments.
2.2.
Experimental procedures
All chloramination experiments were conducted in batch reactors at 24 1 C. The monochloramine demands of resorcinol were determined at pH 8.5 0.2 (carbonate buffer: 20 mM), with two initial concentrations of resorcinol ([Resorcinol]0 ¼ 5 and 100 mM) and by using an excess of monochloramine ([NH2Cl]0/ [Resorcinol]0 ¼ 20 mol/mol). Reactions were initiated by adding, under vigorous mixing, an appropriate volume of a stock solution of resorcinol (5 or 100 mM) into a freshly prepared carbonate buffered solution of monochloramine (100 mM or
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 9 7 e4 5 0 4
2 mM). Blank solutions ([Resorcinol]0 ¼ 0 mM) were systematically prepared under the same conditions in order to evaluate the monochloramine autodecomposition. To determine the effects of pH (6.5 < pH < 12), N/Cl ratio (1 N/Cl 150 mol/mol) and monochloramine dose (0 [NH2Cl]0/[Resorcinol]0 65 mol/mol) on the chloroform yield from chloramination of resorcinol ([Resorcinol]0 ¼ 10e100 mM), the reaction was initiated by introducing under vigorous mixing an appropriate volume (typically 0.5e2 mL) of a stock solution a resorcinol into 1000 mL of a monochloramine solution. The pH of the monochloramine solution was adjusted to the desired value before the addition of resorcinol. After the addition of resorcinol into the solution, the pH was readjusted if necessary and 10 mL aliquots of solution were immediately transferred into a series of 20 mL headspace vials which were then hermetically sealed with PTFE-faced-silicone septa. At least 250 mL of the remaining chloraminated solution of resorcinol was kept in an amber bottle at 24 1 C and pH measurements performed at various reaction times indicated that pH did not vary by more than 0.2 pH unit. At timed intervals, the reaction of monochloramine with resorcinol in the vials was stopped by adding 100 mL of HCl (1 M) and 10 mL of aqueous sodium sulfite solution (2 M) through the septum. Each data point for chloroform yields represents the mean value of chloroform concentrations from two or three vials. Aqueous solutions of resorcinol have also been chlorinated in the present work in order to validate our experimental methods and to compare oxidant demands and chloroform production from chloramination and chlorination experiments. The data obtained from chlorination experiments of resorcinol have been presented in the Supplementary Content.
2.3.
Analytical methods
Free chlorine and total chlorine concentrations in the stock solutions of sodium hypochlorite and monochloramine were determined iodometrically with aqueous sodium thiosulfate solution (0.01 M Prolabo, >99.9%). Monochloramine concentrations were determined spectrophotometrically using a molar extinction coefficient of 446 M1 cm1 for NH2Cl at 244 nm. At various reaction times, monochloramine was determined spectrophotometrically by the I 3 method. For an initial concentration of NH2Cl of 2 mM, 0.1 mL sample of the solution was rapidly transferred into a 10 mL vial containing 4 mL of KI 0.2 M (Acros Organics, >99% for analysis) and 100 mL of acetic acid 1.7 M (Riedel-de Hae¨n, 99e100%). After a reaction time of 2 or 3 min, the mixture was then transferred into a 1-cm spectrophotometer quartz cell and the concentration of triiodide ion was determined by measuring the absorbance at 351 nm (Shimadzu UVmini 1240 spectrophotometer). The concentration of monochloramine was calculated using a molar extinction coefficient of 2.69 104 M1 cm1 for I 3 at 351 nm (Cimetiere et al., 2009). This value was consistent with values reported in literature (Awtrey and Connick, 1951; Bichsel and von Gunten, 1999). CHCl3 was analyzed using a gas chromatograph (Varian 3300 Star) equipped with an automatic headspace sampler (DANI HSS3950), a transfer line (heated at 250 C) between the
4499
sampler and the injector of the chromatograph, a capillary column (J&W/DB624 30 m 0.53 mm I.D., 3 mm) and an electron capture detector. Separation was carried out with nitrogen gas as carrier gas (1 mL min1). Operating temperatures were 250 C, 40 C and 300 C for the injector, the oven and the detector, respectively. Vials were introduced in the headspace sampler for at least 50 min at 50 C in order to reach gaseliquid equilibrium. Calibration curves were performed with standard chloroform solutions (0.01e1.25 mM). Detection and quantification limits were 0.01 and 0.025 mM, respectively. In a few experiments ([Resorcinol]0 ¼ 100 mM), the total organic carbon (TOC) was determined at the beginning and at the end of the experiment using a Shimadzu TOC analyzer (TOC-VCSH).
3.
Results
3.1.
Monochloramine demand of resorcinol
Fig. 1 presents examples of concentrationetime decays of monochloramine ([NH2Cl]0 ¼ 2 mM) obtained in the absence and in the presence of resorcinol ([Resorcinol]0 ¼ 100 mM). Under the conditions used (carbonate buffer ¼ 20 mM, pH ¼ 8.5 0.1), a 18% decay of monochloramine was observed in organic-free solutions after a reaction time of 7 days. A much lower loss of monochloramine was obtained with an initial concentration of monochloramine of 100 mM (z4% decay after a reaction time of 14 days). The autodecomposition rates of monochloramine in organic-free solutions ([NH2Cl]0 ¼ 100 mM and 2 mM) could be well simulated by the kinetic model of Jafvert and Valentine (1992) which has been reported in Tables S1 and S2 in Supplementary Content. At a given pH, the overall rate of autodecomposition of monochloramine follows an apparent second-order kinetic law with respect to the concentration of monochloramine:
Fig. 1 e Concentrationetime profiles of monochloramine in the absence and in the presence of resorcinol. The dashed line represents the simulated autodecomposition curve of monochloramine in organic-free solution calculated by the model of Jafvert and Valentine (1992). ([Resorcinol]0 [ 0 or 100 mM, N/Cl [ 1.2 mol/mol, [NH2Cl]0 [ 2.0 mM, pH [ 8.5 ± 0.1).
4500
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 9 7 e4 5 0 4
d½NH2 Cl 2 ¼ kapp ½NH2 Cl dt
(2)
As the contribution of the autodecomposition reactions of NH2Cl to the overall consumption of monochloramine by resorcinol cannot be always neglected, the monochloramine demand of resorcinol has been determined by subtracting the autodecomposition of monochloramine. The amount of monochloramine autodecomposed ([NH2Cl]a [NH2Cl]b) between two consecutive sampling times ta and tb have been calculated from the integrated form of Eq. (2): 1 1 ¼ kapp ðtb ta Þ ½NH2 Clb ½NH2 Cla or ½NH2 Clb ¼
kapp ðtb ta Þ þ
(3)
1 ½NH2 Cla
1 (4)
where [NH2Cl]a and [NH2Cl]b represent the monochloramine concentrations in organic-free solution at reaction times ta and tb, respectively, and kapp, the apparent second-order rate constant of decomposition of monochloramine in organicfree solution. Fig. 2 presents monochloramine demands of resorcinol at pH 8.5 0.1 as a function of reaction time. The rates of monochloramine consumption increased with increasing concentrations of reactants. For an initial concentration of resorcinol of 100 mM ([NH2Cl]0 ¼ 2 mM), the data showed that the N/Cl ratio (N/Cl ¼ 1.2 and 5 mol/mol) did not significantly affect the rate of monochloramine consumption and the monochloramine demands. Monochloramine demands of 5.5, 7.7 and 8.6 mol of NH2Cl/mole of resorcinol were measured at reaction times of 24, 72 and 168 h, respectively. For an initial concentration of resorcinol of 5 mM ([NH2Cl]0 ¼ 100 mM; N/Cl ¼ 5 mol/mol), the monochloramine demands were equal to 4.0, 6.5 and 8.4 mol/mol after reaction times of 48, 168 and 480 h, respectively (Fig. 2). Data obtained from chlorination experiments of resorcinol ([Resorcinol]0 ¼ 20 mM; [Free Chlorine]0 ¼ 400 mM; Fig. S1) exhibit a very fast consumption of chlorine because the chlorine demands obtained after a reaction time of 5 min were nearly equal to 5.5 and 6.4 mol of chlorine/mole of resorcinol
in carbonate and in phosphate buffered solutions of resorcinol, respectively. For a reaction time higher than 15 h, the measured chlorine demands (7.2 0.1 mol/mol in phosphate buffer) were in good agreement with previous studies (Norwood et al., 1980; De Laat et al., 1982; Reckhow and Singer, 1986). A comparison between the shapes of the chlorine and monochloramine consumption curves also demonstrates that monochloramine is much less reactive than chlorine. Monochloramine demands did not reach a plateau within a reaction time of 168 h and were higher than the chlorine demand of resorcinol. The complete oxidation of 1 mol of resorcinol into CO2 should require a theoretical amount of 13 mol of monochloramine. Since TOC removals in chloraminated solutions of resorcinol ([Resorcinol]0 ¼ 100 mM) were less than 30% after a reaction time of 7 days, the unexpected high monochloramine demands obtained in this work (z8.5 mol/mol) suggest that oxidation by-products of resorcinol catalyze the decomposition of monochloramine. It is well known that proton donator inorganic species such as bicarbonate and phosphate ions catalyze the monochloramine disproportionation. Among the chloramination byproducts of resorcinol identified by Heasley et al. (1999), chloroketones and chlorocarboxylic acids may catalyze the decomposition of monochloramine.
3.2.
Effect of N/Cl ratio upon chloroform formation
Preliminary experiments showed that chlorination of resorcinol ([Resorcinol]0 ¼ 1e50 mM; [Cl2]0/[Resorcinol]0 ¼ 20e40 mol/mol; 7 < pH < 9, reaction time>1 h) leads to a very fast production of 0.93 0.03 mol of chloroform/mole of resorcinol (Fig. S2). In addition, significant amounts of chloroform can also be formed at very low chlorine/resorcinol ratios (<1 mol/mol), indicating that the initial steps of chlorination of resorcinol represent rate limiting steps for the formation of chloroform (Fig. S3). All these data are in good agreement with literature (De Laat et al., 1982; Reckhow and Singer, 1986). Figs. 3 and 4 present chloroform yields obtained from solutions of resorcinol ([Resorcinol]0 ¼ 50 mM) which have
5
N/Cl = 1.2 N/Cl = 5
8 6
8 6
4
[R]0 = 5 µM [NH2 Cl]0 = 0.1 mM N/Cl = 5
4 2
2
0 0
96
192 288 384 480
Chloroform yield (%)
NH2Cl demand (mol/mol)
10
N/Cl = 1.00 N/Cl = 1.05 N/Cl = 1.1 N/Cl = 10 N/Cl = 100 N/Cl = 152
4 3 2 1
0 0
24
48
72 96 120 Time (h)
144 168
Fig. 2 e Monochloramine demand of resorcinol as a function of reaction time ([Resorcinol]0 [ 100 mM, [HOCl]0 [ 2 mM, N/Cl [ 1.2 and 5.0 mol/mol, pH [ 8.5 ± 0.1). Inset refers to monochloramine demand for [Resorcinol]0 [ 5 mM and [NH2Cl]0 [ 0.1 mM, pH [ 8.5 ± 0.1.
0 0
24
48 72 Time (hours)
96
Fig. 3 e Effects of reaction time on the chloroform yield (%) obtained for the monochloramination of resorcinol at various N/Cl ratios ([Resorcinol]0 [ 50 mM, [NH2Cl]0 [ 1 mM, 1 £ N/Cl £ 152 mol/mol, pH [ 8.5 ± 0.2).
4501
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 9 7 e4 5 0 4
Chloroform yield (%)
0.5 0.4
N/Cl = 1.0 N/Cl = 1.1
0.3
N/Cl = 2.0 N/Cl = 10.0
0.2 0.1 0.0 0
3.3. Effect of monochloramine dose on chloroform formation Solutions of resorcinol ([Resorcinol]0 ¼ 100 mM) were also treated at monochloramine-to-resorcinol ratios ranging from 0 to 10 mol/mol at pH 8.5 and using various N/Cl ratios (1 < N/Cl < 10 mol/mol). For a reaction time of 48 h, gas chromatographic analyses showed that chloroform was detected at concentrations close to the quantification limit in samples chloraminated at [NH2Cl]0/[Resorcinol]0 doses lower than 2 mol/mol (Fig. 5). For [NH2Cl]0/[Resorcinol]0 > 2 mol/mol, chloroform yields increased with monochloramine dose and did not exceed 0.5% for the highest monochloramine dose used.
3.4.
Effect of pH on chloroform formation
To study the effect of pH on the formation of chloroform, solutions of resorcinol ([Resorcinol]0 ¼ 10 mM) were treated at a monochloramine dose of 0.65 mM ([NH2Cl]0/[Resorcinol]0 ¼ 65 mol/mol, N/Cl ¼ 10 mol/mol) and at pH values ranging from 6.5 to 12. Fig. 6 shows that the chloroform yield decreases from 7.7% at pH 6.5 to z0.1% at pH 10. At pH higher
than 10, the concentrations of chloroform were below the quantification limit of 0.025 mM (chloroform yield: <0.25%). It should be noted that the decrease of the production of chloroform when the pH increases from pH 8 to 10 has also been previously observed for chlorination of resorcinol solutions (De Laat et al., 1982).
4.
Discussion
4.1. Contribution of free chlorine to the initial steps of degradation of resorcinol Free chlorine formed by the reversible hydrolysis reaction of monochloramine is always present in monochloramine solutions and can therefore contribute to the oxidation of
8
-8 HOCl + ClO l
-
-9
6
-10 4 -11 2
HOCl
CHCll3
log [Free chlorine]
been chloraminated at pH 8.5 with a [NH2Cl]0/[Resorcinol]0 ratio of 20 mol/mol ([NH2Cl]0 ¼ 1 mM) and N/Cl ratios ranging from 1 to 150 mol/mol. The data show that the amounts of chloroform slowly increased with reaction time (reaction time studied: 0e96 h) and markedly decreased when the N/Cl ratio increased from 1 to 150 mol/mol (Fig. 3), and in particular in range 1.0e1.5 mol/mol (Fig. 4). For a reaction time of 48 h, chloroform yields decreased from z3.5% for N/Cl ¼ 1.02 0.02 mol/mol to 0.4% for N/Cl ¼ 100 mol/mol (a yield of 100% is assumed to correspond to the formation of 1 mol of CHCl3 per mole of resorcinol). It should also be noted that chloroform yield which is equal to 95% for N/Cl ¼ 0 mol/mol (chlorination experiments) decreases to 3.5% when the N/Cl increases from 0 to 1.0 mol/mol.
10
Fig. 5 e Effects of monochloramine dose on the production of chloroform by chloramination of resorcinol at various N/Cl ratios. ([Resorcinol]0 [ 100 mM, pH [ 8.5 ± 0.2, Contact time [ 48 h).
Chlorofo f rm yield (%)
Fig. 4 e Effect of the N/Cl molar ratio on the production of chloroform by monochloramination of resorcinol ([Resorcinol]0 [ 50 mM, [NH2Cl]0 [ 1 mM, 1 £ N/ Cl £ 152 mol/mol).
2 4 6 8 [NH2Cl]0 / [R]0 (mol/mol)
-12
0
-13 6
7
8
9 pH
10
11
12
Fig. 6 e Effect of pH on the production of chloroform by monochloramination of resorcinol ([Resorcinol]0 [ 10 mM, [NH2Cl]0 [ 0.65 mM, N/Cl [ 10 mol/mol, reaction time: 48 h) and theoretical concentrations of free chlorine and hypochlorous acid in equilibrium with monochloramine for organic-free solutions of monochloramine.
4502
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 9 7 e4 5 0 4
organic solutes. In organic-free solutions, the concentration of free chlorine in equilibrium with monochloramine can be calculated from the equilibrium constant of the monochloramine hydrolysis reaction (reaction (1), K1 ¼ 5 1012 M; Morris and Isaac, 1983), the acid dissociation constants of 9.25 ; HOCl (Ka1 ¼ 107.54; Morris, 1966) and NHþ 4 (Ka2 ¼ 10 Morris, 1966) and the total concentration of nitrogen species ([Total Nitrogen] ¼ [NHþ 4 ] þ [NH3] þ [NH2Cl]). By neglecting the concentration of the aqueous chlorine ([Cl2]aq) at pH > 5, the equilibrium concentration of free chlorine is equal to:
½Free Chlorineeq ¼
b þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi b2 4a ½Total Chlorine 2a
(5)
Ka2 Hþ þ þ K1 Ka1 þ H Ka2 þ H
(6)
and b ¼ 1 þ að½Total Nitrogen ½Total ChlorineÞ
(7)
with a ¼
In organic-free aqueous solutions of monochloramine, concentration of free chlorine should be at a minimum at pH z 8.4 (pH ¼ (pKa1 þ pKa2)/2) and would decrease when the concentration of free ammonia increases (or when the N/Cl ratio increases). As depicted in the calculated curves in Fig. S4, the theoretical concentrations of free chlorine in solutions containing 1 mM of monochloramine should be very low: z6 107 M for N/Cl ¼ 1.0 mol/mol, 7 1010 M for N/Cl ¼ 1.5 mol/mol and 4 1011 M for N/Cl ¼ 10 mol/mol. In the presence of organic compounds, free chlorine present in monochloramine solutions can react with organic solutes and these reactions accelerate the overall decay of monochloramine. As we have recently demonstrated (Cimetiere et al., 2009), the importance of these indirect reactions involving free chlorine in the overall rate of degradation of organic solutes by monochloramination depend on various parameters such as N/Cl ratio, pH and the reactivity of free chlorine and monochloramine with the organic solutes. In the case of resorcinol, the absolute rate constants for the reactions of chlorine (Rebenne et al., 1996) and monochloramine (Cimetiere et al., 2009) with resorcinol are known. Therefore, the contribution of free chlorine reactions to the degradation of resorcinol can be calculated by kinetic modelling (Table S2). Fig. 7 presents computed data calculated for the following conditions: mM, [Monochloramine]0 ¼ 4 mM, [Resorcinol]0 ¼ 50 6 < pH < 12, 1 < N/Cl < 2 mol/mol. Fig. 7 shows that the contribution of free chlorine to the degradation of resorcinol is highly sensitive to the N/Cl ratio in the range 1.0e1.5 mol/ mol. For the conditions used in Fig. 7, the contributions of free chlorine reactions to the degradation of resorcinol at pH 8 are equal to 25% for N/Cl ¼ 1.05, 15% for N/Cl ¼ 1.1, 8.2% for N/Cl ¼ 1.2 and 1.8% for N/Cl ¼ 2. These data demonstrate that trace amounts of free chlorine in monochloramine solutions can play a significant role in the degradation of resorcinol at N/Cl ratio close to unity because apparent rate constants for the reaction of free chlorine with resorcinol are about 5 orders of magnitude higher than the rate constants of monochloramine with resorcinol (Cimetiere et al., 2009).
Fig. 7 e Contribution of free chlorine reactions to the degradation of resorcinol as a function of pH and N/Cl. Computed data obtained from kinetic modelling (Table S2).
4.2.
Effect of pH on the formation of chloroform
Kinetic modelling cannot be used to simulate chloroform formation during chlorination and chloramination of resorcinol because the reaction mechanisms and the rate constants of reactions leading to the formation of chloroform are not known. The reaction pathways for the formation of chloroform by chlorination of 1,3-dihydroxybenzene model compounds such as resorcinol and orcinol have been investigated by several researchers (Christman et al., 1978; Norwood et al., 1980; Rook, 1980; Boyce and Hornig, 1983; Heasley et al., 1989; Tretyakova et al., 1994). These studies have shown that resorcinol degradation involves aromatic ring-substitution, oxidation, hydrolysis, decarboxylation, ring-contraction and ring-opening reactions. Chlorination of resorcinol leads to the formation of chlororesorcinols, non-cyclic and cyclic chloroketones and chloroketo-carboxylic acids as intermediates, and chloroform (z0.9e0.95 mol/mol of resorcinol), CO2 and short-chain chlorinated acids as end-products. The reaction of monochloramine with resorcinol in water and in ether has been investigated by Heasley et al. (1999). These authors showed that chloramination of aqueous solutions of resorcinol (pH 7) also leads to the formation of chlororesorcinols and of a pentachloro intermediate which undergoes ringopening with NH2Cl/H2O to produce CO2 and non-cyclic chloroketones without significant production of chloroform (z0.04 mol of CHCl3/mole of resorcinol). The data obtained in the present study also confirmed that little chloroform is formed in the reaction of monochloramine with resorcinol. The highest chloroform yield measured in this work (0.077 mol of CHCl3/mole of resorcinol) was obtained at pH 6.5 (Fig. 6). Compared to chlorination, the rate of chloroform formation by chloramination was very slow. The large differences for chlorination and chloramination with respect to the formation of chloroform (rates of formation and chloroform yields) might be attributed to the fact that free chlorine is much more reactive than monochloramine with resorcinol, its chlorinated derivatives (chlororesorcinols) and with the ring-opened by-products (chloroketones, chloroketocarboxylic acids), the latter represent the precursors of
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 9 7 e4 5 0 4
chloroform through the haloform reaction. As reported above, chlorine is about 5 orders of magnitude more reactive than monochloramine with resorcinol, 4-chlororesorcinol and 4,6dichlororesorcinol, but the rate constants for the reaction of monochloramine with chloroketones and chloroketocarboxylic acids are unknown.
5.
Conclusions
Chloramination of resorcinol leads to the formation of chloroform at much lower yields (<8%) than those observed with the use of chlorination (z90e95%). Chloroform yields increased with increasing monochloramine dose and reaction time and with decreasing pH in the range 6.5e10. The data also showed a remarkable decrease of chloroform production when the N/Cl ratio used to prepare monochloramine increased from 1 to 1.5 mol/mol, i.e., when the concentration of excess ammonia increased. Chloroform production was nearly suppressed at very high N/Cl molar ratio (N/Cl > 100). The data obtained in the present work suggest that free chlorine formed by monochloramine hydrolysis might be largely responsible for the formation of chloroform during chloramination of resorcinol, because conditions which favour the hydrolysis reaction of monochloramine into free chlorine (low N/Cl ratio, neutral pH) were found to enhance chloroform formation. Free chlorine reactions may also contribute to the formation of chloroform and of other DBPs during monochloramination of other organic solutes or of drinking waters. Our current research is examining the effects of the N/Cl ratio on the production of chloroform and of other DBPs during chloramination of other phenolic compounds and of NOM extracts.
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2010.06.010.
references
Awtrey, A.D., Connick, R.E., 1951. The absorption spectra of I2, I 3, I, IO 3 , S4O6 and S2O3 . Heat of the reaction I3 ¼ I2 þ I . Journal of the American Chemical Society 73, 1842e1843. Bellar, T.A., Lichtenberg, J.J., Kroner, R.C., 1974. Occurence of organohalides in chlorinated drinking waters. Journal of American Water Works Association 66, 703e706. Bichsel, Y., von Gunten, U., 1999. Determination of iodide and iodate by ion chromatography with postcolumn reaction and UV/visible detection. Analytical Chemistry 71, 34e38. Boyce, S.D., Hornig, J.F., 1983. Reaction pathways of trihalomethane formation from the halogenation of dihydroxyaromatic model compounds for humic acid. Environmental Science & Technology 17, 202e211. Christman, R.F., Johnson, J.D., Hass, J.R., Pfaender, F.K., Liao, W.T., Norwood, D.L., Alexander, H.J., et al., 1978. Natural and model aquatic humics: reactions with chlorine. In: Jolley, R.L. (Ed.),
4503
Water Chlorination, Environmental Impact and Health Effects, vol. 2, pp. 15e28. Ann Arbor, MI. Cimetiere, N., Dossier-Berne, F., De Laat, J., 2009. Monochloramination of resorcinol: mechanism and kinetic modeling. Environmental Science & Technology 43, 9380e9385. De Laat, J., Merlet, N., Dore, M., 1982. Chlorine demand and reactivity in relationship to the trihalomethane formation. Incidence of ammoniacal nitrogen. Water Research 16, 1437e1450. Diehl, A.C., Speitel Jr., G.E., Symons, J.M., Krasner, S.W., Hwang, C.J., Barrett, S.E., 2000. DBP formation during chloramination. Journal of American Water Works Association 92, 76e90. Duirk, S.E., Gombert, B., Croue, J.P., Valentine, R.L., 2005. Modeling monochloramine loss in the presence of natural organic matter. Water Research 39, 3418e3431. Duirk, S.E., Valentine, R.L., 2006. Modeling dichloroacetic acid formation from the reaction of monochloramine with natural organic matter. Water Research 40, 2667e2674. Gallard, H., von Gunten, U., 2002. Chlorination of phenols: kinetics and formation of chloroform. Environmental Science & Technology 36, 884e890. Goslan, E.H., Krasner, S.W., Bower, M., Rocks, S.A., Holmes, P., Levy, L.S., Parsons, S.A., 2009. A comparison of disinfection byproducts found in chlorinated and chloraminated drinking waters in Scotland. Water Research 43, 4698e4706. Heasley, V.L., Alexander, M.B., DeBoard, R.H., Hanley Jr., J.C., McKee, T.C., Wadley, B.D., Shellhamer, D.F., 1999. Reactions of resorcinol and its chlorinated derivatives with monochloramine: identification of intermediates and products. Environmental Toxicology and Chemistry 18, 2406e2409. Heasley, V.L., Burns, M.D., Kemalyan, N.A., McKee, T.C., Schroeter, H., Teegarden, B.R., Whitney, S.E., Wershaw, R.L., 1989. Aqueous chlorination of resorcinol. Environmental Toxicology and Chemistry 8, 1159e1163. Hrudey, S.E., 2009. Chlorination by-products, public health risk tradeoffs and me. Water Research 43, 2057e2092. Hua, G., Reckhow, D.A., 2007. Comparison of disinfection byproduct formation from chlorine and alternative disinfectants. Water Research 41, 1667e1678. Jafvert, C.T., Valentine, R.L., 1992. Reaction scheme for the chlorination of ammoniacal water. Environmental Science & Technology 26, 577e586. Morris, J.C., Isaac, R.A., et al., 1983. A critical review of kinetic and thermodynamic constants for the aqueous chlorineeammonia system. In: Jolley, R.L. (Ed.), Water Chlorination, Environmental Impact and Health Effects, vol. 4, pp. 49e62. Ann Arbor, MI. Morris, J.C., 1966. The acid ionization constant of HOCl from 5 to 35 C. Journal of Physical Chemistry 70, 3798e3805. Norwood, D.L., Johnson, J.D., Christman, R.F., Hass, J.R., Bobenrieth, M.J., 1980. Reactions of chlorine with selected aromatic models of aquatic humic material. Environmental Science & Technology 14, 187e190. Qi, Y., Shang, C., Lo, I.M.C., 2004. Formation of haloacetic acids during monochloramination. Water Research 38, 2375e2383. Rebenne, L.M., Gonzalez, A.C., Olson, T.M., 1996. Aqueous chlorination kinetics and mechanism of substituted dihydroxybenzenes. Environmental Science & Technology 30, 2235e2242. Reckhow, D.A., Singer, P.C., 1986. Mechanisms of Organic Halide Formation During Fulvic Acid Chlorination and Implications with Respect to Preozonation. Water Chlorination. Environmental Impact and Health Effects, vol. 5, pp. 1229e1257. Richardson, S.D., Plewa, M.J., Wagner, E.D., Schoeny, R., DeMarini, D.M., 2007. Occurrence, genotoxicity, and
4504
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 9 7 e4 5 0 4
carcinogenicity of regulated and emerging disinfection by-products in drinking water: a review and roadmap for research. Mutation Research 636, 178e242. Rook, J.J., 1974. Formation of haloforms during chlorination of natural waters. Water Treatment and Examination 23, 234e243. Rook, J.J., 1977. Chlorination reactions of fulvic acids in natural waters. Environmental Science & Technology 11, 478e482. Rook, J.J., et al., 1980. Possible pathways for the formation of chlorinated degradation products during chlorination of humic acids and resorcinol. In: Jolley, R.L. (Ed.), Water Chlorination, Environmental Impact and Health Effects, vol. 3, pp. 85e98. Ann Arbor, MI.
Tretyakova, N.Y., Lebedev, A.T., Petrosyan, V.S., 1994. Degradative pathways for aqueous chlorination of orcinol. Environmental Science & Technology 28, 606e613. Valentine, R.L., Jafvert, C.T., 1988. General acid catalysis of monochloramine disproportionation. Environmental Science & Technology 22, 691e696. Yang, X., Shang, C., Lee, W., Westerhoff, P., Fan, C., 2008. Correlations between organic matter properties and DBP formation during monochloramination. Water Research 42, 2329e2339. Yang, X., Shang, C., Westerhoff, P., 2007. Factors affecting formation of haloacetonitriles, haloketones, chloropicrin and cyanogen halides during chloramination. Water Research 41, 1193e1200.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 1 e4 3 7 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Enhancement of waste activated sludge aerobic digestion by electrochemical pre-treatment Li-Jie Song a, Nan-Wen Zhu a,*, Hai-Ping Yuan a, Ying Hong b, Jin Ding a a b
School of Environmental Science and Engineering, Shanghai Jiao Tong University, Shanghai 200240, PR China 4208 Belle Grove Ct, Orlando, FL 32812, United States
article info
abstract
Article history:
Electrochemical technology with a pair of RuO2/Ti mesh plate electrode is first applied to
Received 15 March 2010
pre-treat Waste Activated Sludge (WAS) prior to aerobic digestion in this study. The effects
Received in revised form
of various operating conditions were investigated including electrolysis time, electric
26 May 2010
power, current density, initial pH of sludge and sludge concentration. The study showed
Accepted 29 May 2010
that the sludge reduction increased with the electrolysis time, electric power or current
Available online 10 June 2010
density, while decreased with the sludge concentration. Additionally, higher or lower pH than 7.0 was propitious to remove organic matters. The electrochemical pre-treatment
Keywords:
removed volatile solids (VS) and volatile suspended solids (VSS) by 2.75% and 7.87%,
Waste activated sludge (WAS)
respectively, with a WAS concentration of 12.9 g/L, electrolysis time of 30 min, electric
Electrochemical pre-treatment
power of 5 W and initial sludge pH of 10. In the subsequent aerobic digestion, the sludge
Aerobic digestion
reductions for VS and VSS after solids retention time (SRT) of 17.5 days were 34.25% and
Scanning electron microscope (SEM)
39.59%, respectively. However, a SRT of 23.5 days was necessary to achieve equivalent
Infrared (IR) spectra
reductions without electrochemical pre-treatment. Sludge analysis by Scanning Electron Microscope (SEM) images and infrared (IR) spectra indicated that electrochemical pretreatment can rupture sludge cells, remove and solubilize intracellular substances, especially protein and polysaccharide, and consequently enhance the aerobic digestion. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Waste Activated Sludge (WAS) generated during wastewater treatment should be stabilized sufficiently to reduce its organic content, pathogen contamination and odor problems prior to ultimate disposal (Vlyssides and Karlis, 2004; Fytili and Zabaniotou, 2009; Li et al., 2009). The most common methods of sludge stabilization are biological processes of anaerobic digestion and aerobic digestion. Compared with anaerobic digestion, simplicity of process and lower capital costs are the advantages of aerobic process. Aerobic digestion has been a popular option for small or medium-sized wastewater treatment plants because of these merits (Barbusinski and
Koscielniak, 1997; Bernard and Gray, 2000). However, conventional aerobic digestion still requires large digestion tanks due to its relatively long retention time (15e30 days) (Jin et al., 2009). During sludge digestion, the hydrolysis of large organic molecules associated with microbial cells has been proven as the rate-limiting step (Eastman and Ferguson, 1981; Shimizu et al., 1993; Tiehm et al., 2001; Gronroos et al., 2005; Park and Novak, 2007; Appels et al., 2008). Therefore, the pre-treatment which disintegrates sludge flocs and disrupt microbial cell walls of sludge was developed to improve subsequent biological digestion. Several successful sludge disintegration technologies include alkaline treatment (Lin et al., 1997, 1998;
* Corresponding author. Tel.: þ86021 54742817; fax: þ86 021 34203732. E-mail address:
[email protected] (N.-W. Zhu). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.05.052
4372
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 1 e4 3 7 8
Li et al., 2008), thermal treatment (Kim et al., 2003; Bougrier et al., 2006; Salsabil et al., 2010), alkaline combined with thermal hydrolysis (Neyens et al., 2003a; Vlyssides and Karlis, 2004), ultrasonic treatment (Wang et al., 1999; Neis et al., 2000; Tiehm et al., 2001; Sangave et al., 2007; Yu et al., 2008; Jin et al., 2009), ozone oxidation(Arodi et al., 2007; Dytczak and Oleszkiewicz, 2008), hydrogen peroxide (Neyens et al., 2003b) and biological hydrolysis with enzymes (Ucisik and Henze, 2008). As a result of the more stringent environmental regulations on the discharge of industrial and municipal wastewater, electrochemical technology is considered a powerful means of pollution control and has been widely used. It has shown great versatility, high removal efficiency, lower temperature requirement and environmental compatibility. The main regent, the electron, is clean (Rajeshwar et al., 1994). In some situations, the technology may be the indispensable step for the treatment of industrial effluents which contain bio-refractory organic pollutants, such as landfill leachate (Deng and Englehardt, 2007), phenol (Yavuz and Koparal, 2006), cyanides (Arellano and Martı´nez, 2007), cigarette industry wastewater (Bejankiwar, 2002), textile (dye) wastewater (Vlyssides et al., 2000; Ko¨rbahti, 2007), tannery wastewater (Costa et al., 2008), etc. Complete mineralization or partial degradation of organic pollutants depends on the anode materials. It was reported that the use of Ti/RuO2 anode produced a series of electrochemical steps which converted high biopolymer substances to low-molecular-weight products. The low-molecular-weight products then can be easily removed by the subsequent biological treatment (Torresa et al., 2003). It was indicated that the combination of electrochemical and biological technology might be a promising choice for the industrial wastewaters that contain recalcitrant compounds. Although the literature on electrochemical treatment of activated sludge is rare, the capability of electrochemical technique to decompose organic macromolecules to small molecules observed from above studies may justify its application in WAS treatment. In this study, electrochemical method is first applied to pre-treat WAS, aiming to enhance subsequent aerobic digestion. Firstly, the effects of electrochemical treatment on sludge reduction and solubilization were evaluated and optimized under different electrochemical conditions. Secondly, the performances of aerobic digestion of treated sludge and untreated sludge were compared and assessed in terms of sludge reduction. The feasibility of electrochemical pre-treatment on the enhancement of WAS aerobic digestibility was finally discussed.
storage period was one week. Table 1 shows the characteristics of sludge samples.
2.2.
Electrochemical pre-treatment
All electrochemical experiments of waste activated sludge were carried out in a 500 mL single-compartment glass cell. Both the anode and the cathode were a pair of Ti/RuO2 mesh plate electrodes of 7.0 10.0 cm2 size. The current was supplied by a highly stable power unit (WYJ. 5 A 60V DC. REGUL. ATED. POWER SUPPLY, Shanghai, China). Copper wires were used for electrical circuit. During the experiments, air was bubbled with an air pump (AIR PUMP, X-6500) to avoid sludge settling and alleviate anode passivation. Experiments were carried out at ambient temperature. All samples were performed in triplicate and average, standard deviation were calculated for each sample.
2.3.
Aerobic digestion reactor
Aerobic digestion experiments were carried out in two plexiglass cylinders with an effective volume of 5 L each (Fig. 1). One reactor was filled with control sample, and the other one was filled with electrochemical pretreated sludge. After adjusting the sludge pH to 7.0 approximately, inoculations were performed with microbial consortia of 2% (V/V) of fresh activated wastewater sludge at a solids concentration of 25 gL1. The digesters were aerated by an air compressor (AIR PUMP, X-6500) to maintain an uniform oxygen concentration of 2 mgO2L1 and good mixing between the electrochemically treated sludge and the biomass. The digesters were operated at room temperature (20e28 C) for 28 days. Oxygen concentration was monitored by an O2 probe at the top of the column. Periodic samples (150 mL each) were taken from the biological reactor, filtered and analyzed for total solids (TS), volatile solids (VS), suspended solids (SS) and volatile suspended solid (VSS). During the period of incubation, the volume loss due to evaporation was readjusted with distilled water.
2.4.
Analytical methods
All analyses were evaluated using chemicals of analytical grade. pH, TS, VS, SS, VSS, and SCOD were determined by the Standard Methods (APHA et al., 1998). TP was determined with the ammonium molybdate spectrophotometric method. TN
Table 1 e Characteristics of sludge samples.
2.
Materials and methods
2.1.
Sludge samples
In this study, WAS was obtained from the sludge return well of the secondary clarifier of a municipal wastewater treatment plant in Shanghai, China. The plant treats 50,000 m3d1 of wastewater with the anaerobic-anoxic-aerobic process. The sludge samples were thickened to required solid concentrations and stored at 4 1 C prior to use. Maximum sludge
Parameter pH Moisture content (%) Conductivity (mScm1) Chemical oxygen demand (COD) (mgL1) Soluble chemical oxygen demand (SCOD) (mgL1) Total solid (TS) (gL1) Volatile solid (VS) (gL1) Suspended solid (SS) (gL1) Volatile suspended solid (VSS) (gL1) Organic content (VSS/SS) (%)
Value 6.69e7.07 99.2e98.2 794e1148 17,462e18,990 36e52 8.0e18.2 5.5e12.7 7.6e17.0 5.4e12.6 0.68e0.74
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 1 e4 3 7 8
4373
conditions in terms of organic matter solubilization and the possible improvement of digestion.
3.1.1.
Effect of electrolysis time
Investigation of the effect of electrolysis time on the organic degradation was carried out on sludge samples with a concentration of 8.2 gL1, at a constant electric power (5 W) and pH 6.90. As shown in Fig. 2, the differences between VSS and VS removals indicated the quantity of organic matters solubilized to the liquid. It increased with the time, which was consistent with the change of soluble chemical oxygen demand. It was noted that the removal efficiencies of VS and VSS increased rapidly within the time frame of less than 30 min, then increased at a slower pace than before. The removal efficiencies of VS and VSS for the first 30 min were 2.4% and 4.9%, respectively, while they were only 3.8% and 8.3% after 240 min. Therefore, an electrolysis time of 30 min was advisable for subsequent studies due to the consideration of reactor volume and power cost.
3.1.2.
Fig. 1 e Experimental device for aerobic digestion. was measured with the alkaline potassium persulphate digestion-UV spectrophotometric method, and ammonia nitrogen was determined by the Nessler’s reagent spectrophotometric method. The sludge pH was measured using a pH meter (pHs-3C, Leici Co., Ltd., Shanghai, China). Conductivity was determined by a conductivity meter (DDSJ-308A, Leici Co., Ltd., Shanghai, China). The samples were centrifuged at 5000g for 30 min and then filtered through a 0.45 mm membrane. The filtrate was collected to measure SCOD, TP, TN and NHþ 4 -N.
2.5.
Evaluation
The efficiencies of electrochemical pre-treatment and aerobic digestion were evaluated by measuring the changes in terms of SCOD, VS and VSS. VS or VSS reductions were calculated as follow: VS removal ¼
VS0 VS 100% VS0
VSS removal ¼
VSS0 VSS 100% VSS0
Effect of electric power
Electric power reflects the energy input to the sludge treatment system and its influence on the sludge disintegration and reduction was investigated with a concentration of 12.7 gL1, electrolysis time of 30 min, at an initial sludge pH of 6.70. The removal efficiencies of VS and VSS increased with the increasing of electric power as shown in Fig. 3, and similar trend was observed for SCOD. The results can be explained by the composition of suspended solids: the mineral matters constituted 20e30%, and the organic matters was 70e80%. Only a small fraction of organic matters in the suspended solids could be hydrolysable with an electric attack probably due to refractory organic compounds. Higher power supplied could lead to more organic matters decomposed and solubilized. When electric power was greater than 5 W, the effect of power on sludge reduction seems to level off. At the power input of 5 W, the VS and VSS removal were 2.3% and 4.8%, respectively, while they only increased to 3.5% and 7.5% at power input of 14 W. Therefore, 5 W of electric power was selected for the subsequent studies.
(1)
(2)
where VS0, VSS0 represented the VS, VSS concentration prior to electrochemical pre-treatment or aerobic digestion.
3.
Results and discussion
3.1. Optimizing of electrochemical pre-treatment conditions The effects of electrochemical treatment on sludge disintegration were studied to evaluate: the optimal pre-treatment
Fig. 2 e Effect of the electrolysis time on the sludge organic degradation (Note: error bars represent standard deviation).
4374
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 1 e4 3 7 8
Fig. 3 e Effect of the electric power on the sludge treatment (Note: error bars represent standard deviation).
3.1.3.
Effect of current density
Different dosages of Na2SO4 were added to the sludge (13.5 gL1) to investigate the effect of current density on sludge reduction. The experiment was carried out with 30 min electrolysis time and a constant electric voltage of 12 V, at an initial sludge pH of 6.70, and temperature of 19 C. The results were shown in Fig. 4. Both VS and VSS removal efficiencies increased with the increasing of the current density as indicted from Fig. 4. The removal efficiencies of VS and VSS were sharply increased at low Na2SO4 concentration range of 0.004e0.04 molL1, followed by a placid increase with increasing the Na2SO4 concentration to 0.16 molL1 where the removal efficiencies were 4.75% and 16.1%, respectively. As a result, the variation of Na2SO4 caused the increase of electric power. Moreover, the temperature after electrolysis increased from the beginning of 19 C to 37 C with the current density, indicating electric energy converting to heat during the electrolysis process. The temperature change is not obvious in the pre-treatment without Na2SO4. The addition of Na2SO4 is to investigate the impact of current density on electrochemical pre-treatment with a constant electric voltage. It was not applied to the subsequent optimizing experiments due to temperature increase and additional chemical costs.
3.1.4.
Effect of the initial sludge pH
The initial pH of sludge sample was adjusted to 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12 and 13 using 1 molL1 of sodium hydroxide or sulfuric acid solution, respectively. The operating conditions were as following: sludge concentration 11.7 mgL1, electric power (4.9e5.2) W, electrolysis time 30 min and at ambient temperature. The removal of the organic solid at each pH is schematically shown in Fig. 5. As was evident, both of VS and VSS removal efficiencies increased with the increasing of pH under alkaline condition or the decreasing of pH in acid range. When pH was 12.94 or 1.99, the higher removal efficiencies of VS and VSS could be achieved, 5.08% and 17.81%, 5.16% and 8.13% respectively. While the initial sludge pH was neutral, 6.97, those of VS and VSS were the lowest and were 2.35% and 4.70%, respectively. That was to say, alkali and acid could
Fig. 4 e Effect of the addition of Na2SO4 on the removal of VS or VSS: (a) VS and VSS removal efficiencies (Note: error bars represent standard deviation); (b) current density and temperature after electrolysis.
enhance the electrochemical pre-treatment of the sludge. In alkaline treatment, hydroxy anions can destroy floc structures and cell walls, resulting in natural shape losing of proteins, saponification of lipid and hydrolysis of RNA. Neyens et al. (2004) and Erdincler and Vesilind (2000) examined chemical degradation and ionization of the hydroxyl groups (eOH/eO) could cause extensive swelling and subsequent solubilization of gels in sludge, and after the destruction of extracellular polymer substances, the cell walls being exposed to a high pH could not withstand the appropriate turgor pressure resulting in the disruption of cells and release of intracellular substances. Gasco et al. (2007) also observed that acid treatment caused the modifications in the organic matter composition of sewage sludge. Additionally, the increasing of ion concentrations in the reactors increased due to the addition of NaOH or H2SO4 perhaps contributed to the sludge reduction. In view of removal efficiency and chemicals cost, the pH of 10 was chosen as the optimal pH, where the removal efficiencies were 2.9% and 8.4%, respectively.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 1 e4 3 7 8
4375
3.2. Performances assessment of aerobic digestion of electrolyzed sludge The ability of electrochemical pre-treatment to solubilize or remove particulate organic matter has been demonstrated by the experiments in the first part of the paper, the comparison and analysis of the treated sludge aerobic biodegradability were required to optimize the coupling process of electrochemical pre-treatment and biodegradation.
3.2.1. Comparison of aerobic digestion performances between electrochemical pre-treatment and control
Fig. 5 e Effect of the initial sludge pH on the removal of VS or VSS (Note: error bars represent standard deviation).
3.1.5.
Effect of the initial sludge concentration
To evaluate the impact of the initial sludge concentration on the degradation of organic matters by electrochemical treatment, experiments were conducted by varying initial solid concentrations with a sludge pH of 10, 30 min electrolysis time and a constant electric power of 5 W. Sludge samples with the certain concentration could be obtained by being centrifuged at 2000g for 15 min or being diluted with the supernatant. Fig. 6 indicated the effect of initial sludge concentration on electrochemical treatment. Both the removal efficiencies of VS and VSS decreased with the sludge concentration increased. The results suggested that at higher organic solid concentrations, mass transfer limitation may be inhibit the rate of electrochemical degradation.
Fig. 6 e Effect of initial sludge concentration on electrochemical treatment (Note: error bars represent standard deviation).
The electrochemical pre-treatment was carried out in the following conditions: sludge concentration 12.9 gL1, electrolysis time 30 min, pH 10.0 and electric power 5 W. Fig. 7(a) shows the changes in characteristics of sludge samples after electrochemical pre-treatment. Since a small portion of the organic solids was degraded or solubilized into the supernatant after electrochemical pre-treatment, the initial VS and VSS concentrations of sludge samples after electrochemical pre-treatment were slightly lower than the control samples but different (P < 0.05), and their reduction percentages were 2.2% and 5.9%, respectively. In accordance with it, COD, TP, TN and NHþ 4 -N of the supernatant after electrolysis (shown in Table 2) increased in some extent.
Fig. 7 e Comparison of electrochemical pretreated sludge and control sludge (Note: error bars represent standard deviation) (a) electrochemical pre-treatment (Note: the P values of TSS, VS and VSS are 0.04, 0.03, 0.02, respectively); (b) aerobic digestion.
4376
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 1 e4 3 7 8
Table 2 e Comparison of the water quality of supernatant (mgLL1).
Control Electrochemical pre-treatment P
SCOD
TN
TP
NHþ 4 -N
44 5 348 8
12.8 0.3 26.5 1.6
1.35 0.1 13.8 0.7
7.8 0.4 9.2 0.6
4.7E-07
1.1E-04
5.5E-06
0.033
Fig. 7(b) depicts the comparison of sludge aerobic digestion performances between electrochemical pretreated sludge and the control sample. Pretreated samples showed higher removal efficiency from the very beginning of digestion. During 1.65 d digestion, VS and VSS removed about 10.2% and 12.25%, respectively, for samples with electrochemical pretreatment, while VS and VSS of control samples reduced about 6.13% and 8.38%, respectively. Pretreated sludge reduction for VS and VSS was 34.25% and 39.59%, respectively, with electrochemical pre-treatment compared with 28.32% and 33.79%, respectively, with control at the digestion time of 17.5 d. This is probably due to the release of soluble organic carbon sources which were more biodegradable. In order to achieve a VSS reduction greater than 38%, the requirement of United States Environmental Protection Agency (U.S. EPA) regulation, 23.5 d was the minimum digestion time for control sludge. Therefore, the electrochemical pre-treatment could significantly enhance sludge biodegradability and aerobic digestion efficiency.
3.2.2. Hypothesis on mechanism of improved aerobic digestion by electrochemical pre-treatment Fig. 8 shows the scanning electron microscope (SEM) images of control and electrochemical pretreated sludge cells. The difference in cell appearance was obvious. The surface of sludge cells (Fig. 8(a)) was relatively round and smooth, while that of electrochemical pretreated sludge cell was deformed, indicating that the sludge cell was broken by electrochemical treatment and intracellular substances would be then solubilized into the solution which could be readily utilized by aerobic microorganisms. As shown in Fig. 9, the Infrared (IR) spectra of the control sludge sample reveal a number of absorption peaks, indicating the complex nature of the sludge. The main absorption band at 3800e2500 cm1 was a symbol of eOH in the carboxyl group (Tirkistani, 1998; Padmavathy et al., 2003; Choi and Yun, 2006). The absorption peak at 1384 cm1 could be attributed to the symmetrical stretching of the carboxylate anion (Choi and Yun, 2006). The existence of phosphonate group was proved from some absorption bands (P]O stretching at 1164 cm1; PeOH stretching at 941 cm1; PeOeC stretching at 1047 cm1) (Pagnanelli et al., 2000). The IR spectra of the Sludge also displayed some characteristic absorption bands of amine group (Choi and Yun, 2006): NeH bending band at 1662 cm1; NeH out of plane bending band near 700 cm1; and CeN stretching band at 1238 cm1. The stretching band of NeH in the range of 3500e3300 cm1 was completely shielded by the strong and large band of carboxyl group in the range of 3800e2500 cm1 Fig. 9 also shows the effect of electrochemical pre-treatment on sludge. It was apparent that the characteristic absorption
Fig. 8 e SEM images of sludge cells: (a) control; (b) electrochemical pretreated sludge.
bands above-mentioned became significantly weaker after electrochemical pre-treatment, which was the indicative of the degradation or solubilization of organic solids by the electrochemical pre-treatment, especially carboxylate, protein and polysaccharide.
3.3.
Cost analysis
Electrochemical pre-treatment is feasible and can be cost effectively based on the literature and this work. The total cost of pre-treatment sludge technology would be less than that of non-treated sludge as the energy consumption costs of the two methods are comparable while the capital cost of the digestion reactor with sludge pre-treatment is less than that of non-treated sludge as the footprint of the former method is smaller than the latter. 1) Regarding energy consumption, Conventional Aerobic Digestion (CAD) requires 18e25 d of sludge retention time, sometimes even 30 d, with the conditions of 20 C temperature, an oxygen concentration of no less than 2 mg O2L1 and 2% solid content. The total energy consumption of CAD under such conditions is approximately 10e15 kW h m3 (Environmental Protection Department, 2010). Based on this experiment results, the total energy consumptions of pre-treatment technology would be
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 1 e4 3 7 8
4377
Fig. 9 e Infrared spectra of sludge samples after electrochemical pre-treatment.
approximately 12e16 kW h m3, with electrolysis of approximately 5 kW h m3 and aerobic digestion of 7e11 kW h m3 2) Most importantly, electrochemical pre-treatment can significantly decrease aerobic digestion time. The digestion time after electrochemical pre-treatment decreased from 23.5 d to 17.5 d as shown in the paper, in other words, electrochemical pre-treatment could reduce roughly 26% of aerobic digestion reactor volume, which is crucial for utilities with limited site spaces, especially in Shanghai. The purpose of this bench experiment is to reduce the digestion time and reactor footprint. The detailed cost of the technology will be evaluated in the subsequent pilot-scale experiments.
4.
supernatant and enhanced its biodegradability. It is recommended that additional work such as the biochemical analysis should be done in the future studies. Electrochemical pre-treatment could decrease the cost of sludge aerobic digestion as a result of comparable energy consumption and smaller aerobic digestion reactor volume requirement.
Acknowledgements This study has been supported by China Postdoctoral Science Foundation (No.: 20090450698) and Shanghai Science and Technology Commission (No.: 09dz1204104).
Conclusions references
The removals of VS and VSS and SCOD after electrochemical pre-treatment increased with the increase of electrolysis time, electric power or the addition of Na2SO4 chemicals. Additionally, the sludge reduction efficiencies decreased when sludge concentration increased. The VS and VSS removal efficiencies were 2.75% and 7.87%, respectively, for WAS with a concentration of 12.9 gL1, pH of 10, by 30 min electrochemical pre-treatment with an electric power of 5 W. In the subsequent aerobic digestion, the sludge reductions for VS and VSS were 34.25% and 39.59%, respectively, after an aerobic digestion time of 17.5 d. However, an aerobic digestion time of 23.5 d was necessary for the control samples to achieve equivalent reductions. The application of electrochemical technique to WAS was proved to enhance the subsequent aerobic digestion. SEM images indicated that sludge cells were ruptured by electrochemical pre-treatment, and intracellular substances were solubilized into the solution which was readily utilized by aerobic microorganisms. Additionally, compared with the IR spectra of control sludge samples, electrochemical pretreatment could weaken the characteristic absorption bands of eOH in the carboxyl group, carboxylate anion, phosphonate group and amine group. It revealed that organic matters, such as protein and polysaccharide, solubilized into the
APHA, AWWA and WEF, 1998. Standard Methods for the Examination of Water and Wastewater, twentieth ed.. Americal Public Health Association/AmericanWaterWorks Association/ Water Environment Federation, Washington, DC, USA. Appels, L., Baeyens, J., Degre`ve, J., Dewil, R., 2008. Principles and potential of the anaerobic digestion of waste-activated sludge. Progress in Energy and Combustion Science 34, 755e781. Arellano, C.A.P., Martı´nez, S.S., 2007. Indirect electrochemical oxidation of cyanide by hydrogen peroxide generated at a carbon cathode. International Journal of Hydrogen Energy 32 (15), 3163e3169. Arodi, B.M., He´le`ne, C., Dominique, P., Delgene`s, J.P., 2007. Ozone pre-treatment as improver of PAH removal during anaerobic digestion of urban sludge. Chemosphere 68, 1013e1019. Barbusinski, K., Koscielniak, H., 1997. Activated sludge floc structure during aerobic digestion. Water Science and Technology 36 (11), 107e114. Bejankiwar, R.S., 2002. Electrochemical treatment of cigarette industry wastewater: feasibility study. Water Research 36 (17), 4386e4390. Bernard, S., Gray, N.F., 2000. Aerobic digestion of pharmaceutical and domestic wastewater sludges at ambient temperature. Water Research 34 (3), 725e734. Bougrier, C., Albasi, C., Delgene`s, J.P., Carre`re, H., 2006. Effect of ultrasonic, thermal and ozone pre-treatments on waste
4378
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 7 1 e4 3 7 8
activated sludge solubilisation and anaerobic biodegradability. Chemical Engineering and Processing 45 (8), 711e718. Choi, S.B., Yun, Y.S., 2006. Biosorption of cadmium by various types of dried sludge: an equilibrium study and investigation of mechanisms. Journal of Hazardous Materials B138, 378e383. Costa, C.R., Botta, C.M.R., Espindola, E.L.G., Olivi, P., 2008. Electrochemical treatment of tannery wastewater using DSA electrodes. Journal of Hazardous Materials 153 (1e2), 616e627. Deng, Y., Englehardt, J.D., 2007. Electrochemical oxidation for landfill leachate treatment. Waste Management 27 (3), 380e388. Dytczak, M.A., Oleszkiewicz, J.A., 2008. Performance change during long-term ozonation aimed at augmenting denitrification and decreasing waste activated sludge. Chemosphere 73, 1529e1532. Eastman, J.A., Ferguson, J.F., 1981. Solubilization of particulate organic carbon during the acid phase of anaerobic digestion. Journal of Water Pollution and Control Facility 53 (3), 352e366. Environmental Protection Department, 2010. Guideline on Best Available Technologies of Pollution Prevention and Control for Treatment and Disposal of Sludge from Municipal Wastewater Treatment Plant (On Trial) Beijing, China. Erdincler, A., Vesilind, P.A., 2000. Effect of sludge cell disruption on compactibility of biological sludge. Water Science and Technology 42 (9), 119e126. Fytili, D., Zabaniotou, A., 2009. Utilization of sewage sludge in EU application of old and new methodsdA review. Renewable and Sustainable Energy Reviews 12, 116e140. Gasco, G., Cueto, M.J., Me´ndez, A., 2007. The effect of acid treatment on the pyrolysis behavior of sewage sludges. Journal of Analytical and Applied Pyrolysis 80, 496e501. Gronroos, A., Kyllonen, H., Korpijarvi, K., Pirkonen, P., Paavola, T., Jokela, J., Rintala, J., 2005. Ultrasound assisted method to increase soluble chemical oxygen demand (SCOD) of sewage sludge for digestion. Ultrasonics Sonochemistry 12 (1e2), 115e120. Jin, Y.Y., Li, H., Mahar, R.B., Wang, Z.Y., Nie, Y.F., 2009. Combined alkaline and ultrasonic pretreatment of sludge before aerobic digestion. Journal of Environmental Sciences 21, 279e284. Kim, J., Park, C., Kim, T., Lee, M., Kim, S., Kim, S., Kim, S., Lee, J., 2003. Effects of various pretreatments for enhanced anaerobic digestion with waste activated sludge. Journal of Bioscience and Bioengineering 95, 271e275. Ko¨rbahti, B.K., 2007. Response surface optimization of electrochemical treatment of textile dye wastewater. Journal of Hazardous Materials 145 (1e2), 277e286. Lin, J.G., Chang, C.N., Chang, S.C., 1997. Enhancement of anaerobic digestion of waster activated sludge by alkaline solubilization. Bioresource Technology 62, 85e90. Lin, J.G., Ma, Y.S., Huang, C.C., 1998. Alkaline hydrolysis of the sludge generated from a high-strength, nitrogenouswastewater biological-treatment process. Bioresource Technology 65, 35e42. Li, H., Jin, Y.Y., Mahar, R.B., Wang, Z.Y., Nie, Y.F., 2008. Effects and model of alkaline waste activated sludge treatment. Bioresource Technology 99, 5140e5144. Li, X.S., Ma, H.Z., Wang, Q.H., Matsumoto, S., Maeda, T., Ogawa, H. I., 2009. Isolation, identification of sludge-lysing strain and its utilization in thermophilic aerobic digestion for waste activated sludge. Bioresource Technology 100, 2475e2481. Neis, U., Nickel, K., Tiehm, A., 2000. Enhancement of anaerobic sludge digestion by ultrasonic disintegration. Water Science and Technology 42 (9), 73e80. Neyens, E., Baeyens, J., Dewil, R., De Heyder, B., 2004. Advanced sludge treatment affects extracellular polymeric substances to
improve activated sludge dewatering. Journal of Hazardous Materials 106 (2e3), 83e92. Neyens, E., Baeyens, J., Creemers, C., 2003a. Alkaline thermal sludge hydrolysis. Journal of Hazardous Materials 97 (1e3), 295e314. Neyens, E., Baeyens, J., Weemaes, M., Deheyder, B., 2003b. Pilotscale peroxidation (H2O2) of sewage sludge. Journal of Hazardous Materials B98, 91e106. Padmavathy, V., Vasudevan, P., Dhingra, S.C., 2003. Thermal and spectroscopic studies on sorption of nickel (II) ion on protonated Baker’s yeast. Chemosphere 52 (10), 1807e1817. Pagnanelli, F., Papini, M.P., Toro, L., Trifoni, M., Veglio, F., 2000. Biosorption of metal ions on Anthrobacter sp.: biomass characterization and biosorption modeling. Environmental Science and Technology 34 (13), 2773e2778. Park, C., Novak, J.T., 2007. Characterization of activated sludge exocellular polymers using several cation-associated extraction methods. Water Research 41 (8), 1679e1688. Rajeshwar, K., Ibanez, J.G., Swain, G.M., 1994. Electrochemistry and environment. Journal of Applied Electrochemistry 24, 1077e1109. Salsabil, M.R., Laurenta, J., Casellasa, M., Dagot, C., 2010. Technoeconomic evaluation of thermal treatment, ozonation and sonication for the reduction of wastewater biomass volume before aerobic or anaerobic digestion. Journal of Hazardous Materials 174, 323e333. Sangave, P.C., Gogate, P.R., Pandit, A.B., 2007. Ultrasound and ozone assisted biological degradation of thermally pretreated and anaerobically pretreated distillery wastewater. Chemosphere 68 (1), 42e50. Shimizu, T., Kudo, K., Nasu, Y., 1993. Anaerobic waste activated sludge digestion e a bioconversion and kinetic model. Biotechnology Bioengineering 41, 1082e1091. Tiehm, A., Nickel, K., Zellhorn, M., Neis, U., 2001. Ultrasonic waste activated sludge disintegration for improving anaerobic stabilization. Water Research 35, 2003e2009. Tirkistani, F.A.A., 1998. Thermal analysis of some chitosan Schiff bases. Polymer Degradation and Stability 60, 67e70. Torresa, R.A., Sarria, V., Torres, W., Peringera, P., Pulgarina, C., 2003. Electrochemical treatment of industrial wastewater containing 5-amino-6-methyl-2-benzimidazolone: toward an electrochemicalebiological coupling. Water Research 37, 3118e3124. Ucisik, A.S., Henze, M., 2008. Biological hydrolysis and acidification of sludge under anaerobic conditions: the effect of sludge type and origin on the production and composition of volatile fatty acids. Water Research 42 (14), 3729e3738. Vlyssides, A.G., Karlis, P.K., 2004. Thermal-alkaline solubilization of waste activated sludge as a pre-treatment stage for anaerobic digestion. Bioresource Technology 91, 201e206. Vlyssides, A.G., Papaioannou, D., Loizidoy, M., Karlis, P.K., Zorpas, A.A., 2000. Testing an electrochemical method for treatment of textile dye wastewater. Waste Management 20 (7), 569e574. Wang, Q.H., Kuninobub, M., Kakimoto, K., Ogawa, H.I., Kato, Y., 1999. Upgrading of anaerobic digestion of waste activated sludge by ultrasonic pretreatment. Bioresource Technology 68, 309e313. Yavuz, Y., Koparal, A.S., 2006. Electrochemical oxidation of phenol in a parallel plate reactor using ruthenium mixed metal oxide electrode. Journal of Hazardous Materials 136 (2), 296e302. Yu, G.H., He, P.J., Shao, L.M., Zhu, Y.S., 2008. Extracellular proteins, polysaccharides and enzymes impact on sludge aerobic digestion after ultrasonic pretreatment. Water Research 42, 1925e1934.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 1 6 e4 6 2 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Evaluation on the impacts of predators on biomass components and oxygen uptake in sequencing batch reactor and continuous systems Bing-Jie Ni a, Bruce E. Rittmann b, Han-Qing Yu a,* a b
Lab of Environmental Engineering, Department of Chemistry, University of Science & Technology of China, Hefei 230026, China Center for Environmental Biotechnology, Arizona State University, Tempe, AZ 85287-5701, USA
article info
abstract
Article history:
An expanded unified model for the biomass fractions, soluble-organic fractions, and
Received 25 April 2010
oxygen-uptake rates considering extracellular polymeric substances (EPS), intracellular
Received in revised form
storage products (XSTO), and predators for activated sludge is used to study the impacts of
27 May 2010
predators on biomass components and oxygen uptake. The new model is applied to
Accepted 29 May 2010
evaluate how predation affects the oxygen-uptake rate (OUR) and the different forms of
Available online 17 June 2010
biomass: active bacteria (XH), XEPS, and XSTO, under dynamic feast-and-famine and continuous conditions. For the dynamic conditions of a sequencing batch reactor (SBR),
Keywords:
eliminating predators from the model increases XH and XEPS fractions significantly, and
Activated sludge
this causes the substantial increases in OUR and MLVSS once the famine period begins. An
Biomass
analysis of how the OUR is distributed among the several respiration processes shows that
Continuous activated sludge
the predation of XH is the most significant oxygen utilization rate process in the system
Oxygen-uptake rate (OUR)
under famine conditions of an SBR. Application of the model to simulate the long-term
Predators
operation of an SBR indicates that predators reach their maximum fraction in the MLVSS
Sequencing batch reactor (SBR)
(w4% of MLVSS) at a solids retention time of about 13 days, but they are washed out at a solids retention time less than w3 days. Simulation for a continuous system indicates that predators take more time (about 800 h) to reach steady state and reach their maximum fraction (w5.5%) at an SRT of w14 days. Comparison of SBR and continuous systems reveals that the predators have greater impact in the continuous system because the permanent near-famine condition accentuates predation processes. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
The sequencing batch reactor (SBR) has been widely used to treat municipal wastewater (Irvine and Ketchum, 1989; Furumai et al., 1999; de Kreuk et al., 2005; Otawa et al., 2006), landfill leachates (Timur and Ozturk, 1999), and various industrial wastewaters (Irvine and Ketchum, 1989; Segar et al., 1995; Kortekaas et al., 1998) since it was invented by Irvine and his co-workers in the 1970s (Irvine and Busch, 1979). In SBRs,
wastewater is mixed with the aerated activated sludge in a pulse-feed mode (Beun et al., 2001, 2002; Schwarzenbeck et al., 2004; Moussa et al., 2005; de Kreuk and van Loosdrecht, 2006). The highly dynamic feed regime is a practical example of feast-and-famine conditions that promote the formation of solid microbial products, particularly intracellular storage products (XSTO) (van Loosdrecht et al., 1997; Pratt et al., 2004; Karahan et al., 2006; Ni et al., 2009, 2010). Although bacteria form the metabolic foundation in SBRs,
* Corresponding author. Fax: þ86 551 3601592. E-mail address:
[email protected] (H.-Q. Yu). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.05.048
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 1 6 e4 6 2 2
predators often are associated with good performance and should not be neglected. A new expanded unified model, described in detail elsewhere (Ni et al., 2010), describes the biomass fractions, soluble-chemical oxygen demanded (COD) fractions, and oxygen-uptake rates by considering extracellular polymeric substances (EPS), intracellular storage products (XSTO), and predators during dynamic operating conditions in activated sludge. The foundation of the model is the expanded unified model that describes the dynamics of EPS, XSTO, and soluble microbial products (SMP) (Ni et al., 2009). The kinetics for predation of active bacteria (XH), EPS, and XSTO were added by Ni et al. (2010). Model calibration using batch experimental data provided the new parameter values for predation processes and insight into mechanisms involving predators (Ni et al., 2010). The maximum specific growth rate of the predators is much slower than the bacterial maximum specific growth rate (mH,S), confirming that predators are relatively slow growers. However, the predators’ endogenous-decay rate and oxygen half-maximum-rate concentration are similar to those of heterotrophic bacteria. The model with calibrated parameters was able to simulate the results of independent experiments, including those in which predators were eliminated by a highsalt shock. Ni et al. (2009) indicated that the strong feast-and-famine conditions in an SBR accentuate the differences in growth and decay mechanisms among the different biomass components, including XH, EPS, XSTO. Ni et al. (2010) further illuminated that predators have important impacts on biomass components and oxygen uptake; thus, they should not be neglected, even though the bacteria form the metabolic foundation in SBRs. However, the detail impacts of predators on biomass components and oxygen uptake in SBR or continuous systems still remain unexplored. In this study, we exploit the model to interpret the role of predation in activated sludge experiments operated under feast-and-famine conditions. We also use the model to predict the fate and influence of predators for the long-term performance of an SBR and compare the impacts to those in a continuous, steady state activated sludge process. This study explores three important details related to predation processes. First, we gain insights of how predators affect the different biomass components (XH, EPS, XSTO, and XI) and oxygen uptake in activated sludge. Second, we comprehensively analyze the long-term performance of SBR and continuous systems in the presence of predators and as it is affected by the SRT. Finally, we compare SBR and continuous systems in the presence of predators to reveal their key difference. Information provided in this paper will be useful for understanding the role of predation in activated sludge and for optimizing the design and operation of activated sludge system in the presence of predators.
2.
Materials and methods
2.1.
Batch experiments
The activated sludge used for predation tests in this work was the same as described in Ni et al. (2010). For the respiration
4617
tests in the presence of the predators, activated sludge was kept under the famine condition to deplete it of internal storage products and washed twice with distilled water to remove external soluble-organic material (SCOD). Then, a predetermined content of external substrate (soybean-processing wastewater) was added, and the OUR was monitored until the substrate became depleted and the endogenous activity resumed. Samples taken every 5e10 min were analyzed for mixed liquor volatile suspended solids (MLVSS). The initial MLVSS concentration in each experiment was 1340 mg/L, and the experiments were conducted at temperature of 20 C and at a fixed pH of 7.0, controlled using HCl or NaOH additions by using a pH meter and auto-titrator that compensated/ neutralized the protons produced or consumed by adding equivalent amounts of base or acid. The OUR and MLVSS were determined twice for each test, and the averages are reported here.
2.2.
Analysis methods
Measurements of COD, mixed liquor suspended solids (MLSS), and MLVSS followed Standard Methods (APHA, 1995). OUR was measured using a respirometer that consisted of a 300 mL glass vessel with a pot at the top for insertion of a DO probe (MO128, Mettler-Toledo Gmbh, Switzerland). A magnetic stirring bar and a stirring plate provided internal mixing of the liquor and sludge. The system was kept at pH of 7.0 0.1 using NaOH or HCl solutions and an auto-titrator, which enabled the pH to be maintained at about 7.0. For each respirometric test, 250 mL of sludge was collected from the reactors and added to the respirometer. The slope of the DO concentration decline was measured over 15 min. The OUR, calculated from the slope though linear regression, was determined twice for each test, and the averages are reported here.
2.3.
Modeling methods
In our model (Ni et al., 2010), predators consume EPS, XSTO, and active biomass (XH), since all are part of the solid matrix of the MLVSS and are subject to phagocytosis. When protozoa graze on active bacteria, they convert the non-biodegradable fraction of XH into inert biomass (XI). Predators grow aerobically on the degradable fraction of the active bacteria, converting part of it to predator biomass (Xpred) and oxidizing the rest to consume dissolved oxygen. The predation rate is a multiplicative Monod function of dissolved oxygen and the predator’s substrate, which is XH in this case. The rate also is a function of the protozoan concentration, Xpred. Predators also prey on EPS and XSTO for their aerobic growth with similar kinetics. When EPS or XSTO is consumed for growth, part of it is converted into Xpred, and the remaining fraction is used for energy, consuming an equivalent amount of dissolved oxygen, SO. However, predation of XEPS or XSTO does not generate inert COD, as XEPS and XSTO do not have non-biodegradable fractions. First, the role of predators in the 4-h feast-and-famine batch experiments was explored by model simulations (scenario I). The expanded unified model of Ni et al. (2010) was run with and without predators and using the same kinetic and stoichiometric parameter values and same initial conditions. The initial
4618
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 1 6 e4 6 2 2
900 850 XH (mg/L)
150 100
800 Absence of predators Presence of predators
750
50
700 60
A
0 1600
B
50 XEPS (mg/L)
VSS (mg/L)
A
Absence of predators Presence of predators
1500
40 30 20 10
1400
0 570
B
1300 0
1
2 Time (h)
3
C
560
4 X I (mg/L)
OUR (mg/L.h)
200
Fig. 1 e Model predictions of the batch experimental (A) OUR and (B) MLVSS with the model absent predators and with predators. The batch experimental data are in the presence of predators. The initial MLVSS is 1340 mg/L for model simulations.
550 540 530 520 80
conditions corresponded to the batch experimental conditions: T ¼ 20 C, pH ¼ 7.0, SS ¼ 320 mg/L, and MLVSS ¼ 1340 mg/L. Second, the new model was used to simulate long-term SBR performance using the parameters in Ni et al. (2010) (scenario II). The simulation was conducted with the following the initial conditions: SBR volume ¼ 2 L, SRT ¼ 10 days, cycle time ¼ 4 h, T ¼ 20 C, XH(0) ¼ 1000 mg/L, Xpred(0) ¼ 50 mg/L, filling time ¼ 3 min, filling volumetric flux ¼ 20 L/h, react time ¼ 222 min, settling time ¼ 10 min, draw time ¼ 5 min, and influent SS ¼ 400 mg/L. We ran the simulation for 2000 h, which allowed the system to come to a global steady state for all the soluble and solid components. Third, we also ran parallel modeling experiments for a continuous activated sludge system (scenario III). For the mass-balance equations of the continuous activated sludge system, we added to each mass balance an advective term, QS0/VQS/V; where V is the liquid volume, Q is the flow rate, and S0 and S represent influent and reactor concentrations, respectively. Rate terms in each mass balance and the parameters for model simulation are the same as for the SBR simulations (Ni et al., 2010). The initial conditions were: reactor volume ¼ 2 L, SRT ¼ 10 days, T ¼ 20 C, XH(0) ¼ 1000 mg/L, Xpred(0) ¼ 50 mg/L, volumetric flux ¼ 0.25 L/ h, and influent SS ¼ 400 mg/L. We also ran the simulation for 2000 h, which allowed the system to come to a global steady state for all the soluble and solid components. All model simulations were performed with a nonlinear least-squares algorithm in the AQUASIM software package which offers a free definition of the biokinetic model, flow scheme, and process control strategies, graphic support of the
XSTO (mg/L)
D 60 40 20 0 0
1
2 Time (h)
3
4
Fig. 2 e Model-predictions for the solid components in the activated sludge in the presence or absence of predators and corresponding to Fig. 1: (A) XH; (B) XEPS; (C) XI; and (D) XSTO. The initial XH and XI are 820, and 520 mg/L, respectively.
simulation, experimental data, and communication with spreadsheet programs (Reichert, 1998; Siegrist et al., 2002).
3.
Results and discussion
3.1.
Model-based analysis of the influence of predation
Scenario I explores the role of predators by model simulations. The simulation and experimental results in the 4-h feast-andfamine batch experiments are compared in Fig. 1. The model absent predators simulated the high OUR value accompanying the initial strong increase in the VSS concentration, but
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 1 6 e4 6 2 2
4619
160 SS utilization XH respiration SBAP utilization SUAP utilization XSTO utilization
OUR (mg/L.h)
120 80
A
40 0 30
OUR (mg/L.h)
25
Predation of XH
20
B
15 Xpred respiration
10
Predation of XEPS
5
Predation of XSTO
0 0
1
2
3
4
Time (h) Fig. 3 e Model-predictions for the OURs attributable to each oxygen-consuming process corresponding to the withpredation conditions of Fig. 1: (A) Non-predation respiration processes; and (B) Predation processes.
predicted OUR values substantially too low once the famine period began (w50 min). Likewise, the simulated VSS results were too high for the predator-absent model after about 1 h. The simulation results from the model with predators were able to describe the OUR and VSS profiles without systematic error for the entire batch experiment. The results for all the biomass fractions during the 4-h simulation of the feast-and-famine batch experiments corresponding to Fig. 1 are illustrated in Fig. 2. Although XSTO had only small differences between the models (Fig. 2D), the presence of predators had profound impact on the other three biomass fractions. Elimination of predators increased the active biomass (XH, Fig. 2A) and EPS (XEPS, Fig. 2B) fractions significantly, since they are food sources for the predators. XH decreased by about 15% with predation, while its loss was only about 3% without predation. Likewise, XEPS declined 20% with predation, but did not decline without predation. The lack of impact of predation on XSTO is explained by the fact that the main loss of XSTO is its utilization by XH, which is very rapid and overwhelms predation. Contrary to the other forms of biomass, XI increased greatly (Fig. 2C), because the excreted fecal pellets contain inert biomass. Fig. 3 presents an in-depth analysis of how OUR was distributed among the different oxygen-consuming processes for the with-predation simulation of Figs. 1 and 2 in the 4-h feast-and-famine batch experiments. The OUR for each respiration process was obtained by multiplying the SO-term
Fig. 4 e Model prediction of soluble and solid components of 2000 h of SBR operation from start-up to steady state: (A) SS, SCOD, SSMP, SUAP, and SBAP; and (B) MLVSS, EPS, XH, Xpred, XI and XSTO. The simulation conditions are: Reactor volume [ 2 L, SRT [ 10 days, cycle time [ 4 h, T [ 20 C, XH(0) [ 1000 mg/L, Xpred(0) [ 50 mg/L, filling flux [ 20 L/h, filling time [ 3 min, react time [ 222 min, settling time [ 10 min, draw time [ 5 min, and influent SS [ 400 mg COD/L.
stoichiometry of the relevant process by the corresponding kinetic rate (Ni et al., 2010). As expected, OUR associated with utilization of SS to support synthesis of XH was very large during the feast period (Fig. 3A). Once the system entered the famine period (after about 50 min), utilization of XSTO joined endogenous respiration of XSTO and predation of XH as the significant OUR processes, although the XSTO process declined over the rest of the cycle. The OURs due to predation of XH and utilization of XSTO jumped as the substrate was depleted. OURs for XH utilization of SUAP and SBAP and for predation on XEPS and XSTO were low at all times.
3.2.
Long-term simulation for the SBR
Fig. 4 presents the simulation results for MLVSS, XH, Xpred, XEPS, XI, XSTO, SCOD, SS, SSMP, SUAP, and SBAP out to 2000 h of SBR operation. The soluble components (Fig. 4A) SS, SUAP, SBAP, SSMP, and SCOD all took about three SRTs (w30 days or 720 h) to approach plateaus. For the biomass components (Fig. 4B), XSTO stabilized after 30 days of operation, the active biomass (XH) and XEPS reached plateaus at about four SRT (1000 h), Xpred took less time (about 750 h) to reach steady state, and the inert residual biomass (XI) took at least 1900 h (8 times SRT) to reach steady state. Accordingly, the MLVSS (¼XH þ XI þ XEPS þ XSTO þ Xpred)
4620
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 1 6 e4 6 2 2
Fig. 5 e Model prediction of the effect of SRT on (A) MLVSS; and (B) the fraction in XH, XI, XEPS, Xpred, and XSTO at global steady state of SBR operation. Panels A and B are for an SRT up to 100 days. Panel C included in Panel A is an expanded scale for an SRT £ 1 day, the system washes out at SRT < SRTmin [ 0.19 day.
also needed at least 8 SRT to reach its plateau. At steady state, the active microorganisms were about 24% of MLVSS (Fig. 4B). The remaining biomass was comprised of inert biomass at 50%, XEPS at 20%, Xpred at 5%, and XSTO at w1%. Thus, XH, XEPS and XI became the main fractions of MLVSS for the SRT simulated (10 days). The effects of the SRT on MLVSS and the fractions of XEPS, XSTO, Xpred, XH, and XI of SBR operation are illustrated in Fig. 5. As expected, MLVSS increased with increasing SRT when SRT was greater than the washout SRT of 0.19 day (Fig. 5A). The active biomass became a smaller fraction with increasing SRT, while the inert biomass became increasingly dominant. At SRT ¼ 100 days, XH was 3.0%, and XI was 93% (Fig. 5B). Fig. 5B illustrates vividly that the other forms of non-inert biomass in SBR did not behave the same as XH. EPS generally tracked XH, but its highest accumulation occurred when XH was declining: EPS was almost as high as active heterotrophs (w40%) at an SRT of about 6 days, and it declined in parallel to heterotrophs for larger SRT; at SRT ¼ 100 days, XEPS was 2.0% of MLVSS. The predators reached their maximum fraction (w4.2%) at an SRT of w13 days. They washed out for SRT < w3 days, whereas active heterotrophs washed out at for SRT < 0.19 day. This difference is a natural consequence of the predators being slower growers than heterotrophs. XSTO was a major fraction (up to 30%) for SRT near washout of the heterotrophs, but declined to a small fraction for SRT > 5 days.
Fig. 6 e Model prediction of soluble and solid components of 2000 h of continuous activated sludge system operation from start-up to steady state: (A) SS, SCOD, SSMP, SUAP, and SBAP; and (B) MLVSS, EPS, XH, Xpred, XI and XSTO. The simulation conditions are: Reactor volume [ 2 L, SRT [ 10 days, T [ 20 C, XH(0) [ 1000 mg/L, Xpred(0) [ 50 mg/L, volumetric flux [ 0.25 L/h, and influent SS [ 400 mg/L.
3.3. Simulation for the continuous system and comparison with the SBR Fig. 6 presents the simulation results for MLVSS, XH, Xpred, XEPS, XI, XSTO, SCOD, SS, SSMP, SUAP, and SBAP out to 2000 h of operation of the continuous activated sludge system. The soluble components (Fig. 6A) SS, SUAP, SBAP, SSMP, and SCOD all took about 2.5 SRTs (w25 days or 600 h) to approach plateaus. For the biomass components (Fig. 6B), XSTO stabilized after 35 days of operation, the active biomass (XH) and XEPS reached plateaus at about 3 SRT (720 h), Xpred took more time (about 800 h) to reach steady state, and the inert residual biomass (XI) took at least 1700 h (7 times SRT) to reach steady state. Again, the MLVSS (¼XH þ XI þ XEPS þ XSTO þ Xpred) also needed at least 7 SRT to reach its plateau. At steady state, the active microorganisms were about 21% of MLVSS (Fig. 6B). The remaining biomass was comprised of inert biomass at 54.5%, XEPS at 17%, Xpred at 6.5%, and XSTO at w1%. Also, XH, XEPS and XI became the main fractions of MLVSS for the SRT simulated (10 days). The effects of the SRT on MLVSS and the fractions of XEPS, XSTO, Xpred, XH, and XI of the continuous activated sludge system operation are illustrated in Fig. 7. Similar to the SBR, MLVSS increased with increasing SRT when SRT was greater than the washout SRT of 0.21 day (Fig. 7A). The active biomass became a smaller fraction with increasing SRT, while the inert biomass became increasingly dominant. At SRT ¼ 100 days, XH was 2.3%, and XI was 95% (Fig. 7B).
4621
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 1 6 e4 6 2 2
Table 1 e Comparison of the effects of predators on the key points of the continuous activated sludge system and SBR operation for an SRT of 10 days. Characteristics
Time to steady state (d)
Fractions in MLVSS at SRT of 10 days (%)
OUR values (mg/L/h)
Fig. 7 e Model prediction of the effect of SRT on (A) MLVSS; and (B) the fraction in XH, XI, XEPS, Xpred, and XSTO at global steady state of continuous activated sludge system operation.
Fig. 7B illustrates the behavior of all forms of biomass in the continuous activated sludge system. EPS generally tracked XH, but its highest accumulation occurred when XH was declining, and it declined in parallel to heterotrophs for larger SRT; at SRT ¼ 100 days, XEPS was 1.3% of MLVSS. The predators reached their maximum fraction (w5.5%) at an SRT of w14 days. They washed out for SRT < w3 days, whereas active heterotrophs washed out at for SRT < 0.21 day. XSTO was
70 60
OUR (mg/L.h)
50 40 30 SBR operation Continuous activated sludge system
20 10 0 0
20
40
60
80
100
SRT (d) Fig. 8 e Model predictions of the effect of SRT on OUR at global steady state for the continuous activated sludge system and the SBR.
Continuous Sequencing activated batch reactor sludge system (SBR) Soluble components XSTO XEPS XH XI Xpred MLVSS XH XEPS XSTO Xpred XI
25
30
35 30 30 70 35 70 21 17 w1 6.5 54.5 60
30 40 40 80 30 80 24 20 w1 5.0 50 51
a major fraction (up to 35%) for SRT near washout of the heterotrophs, but declined to a small fraction for SRT > 3 days. Comparing Figs. 4 and 6, the soluble components (SS, SUAP, SBAP, SSMP, and SCOD) took less time (2.5 vs. 3 times SRTs) to approach steady state for the continuous system. Likewise, the active biomass, XEPS, and XI reached plateaus in less time in the continuous system. However, Xpred, XSTO, and MLVSS took slightly more time in continuous system to reach steady state. At steady state, the active microorganisms were about 21% of MLVSS in continuous system, which is lower than that of SBR (24%). Although the XSTO at steady state was similar in both systems, the continuous system had obviously higher Xpred and XI, but lower XEPS compared to the SBR. Comparing Figs. 5 and 7 identifies the differential effects of the SRT in the two systems. One consistent difference is that the MLVSS of the continuous system was higher than that of the SBR at any SRT, and it was accompanied by a lower fraction of active microorganisms: e.g., the highest fraction of active microorganisms in the SBR was 52% at an SRT of about 3 days, while the maximum was 45% in continuous system at an SRT of 2.5 days. In contrast, the continuous system had a clearly higher fraction of predators, because the continuous activated sludge system was almost in permanent nearfamine condition for any realistic SRT. Fig. 8 shows the model-predictions for OUR at global steady state for the continuous activated sludge system and the SBR operated over the range of at SRT. The OUR levels in the SBR increased with increasing operating SRT initially from 0 to 51 mg/L/h at SRT of 10 days and later gradually decreased to 46 mg/L/h at SRT of 100 days; the OUR in the continuous activated sludge system initially increased to 60 mg/L/h and later gradually decreased to 53 mg/L/h with increasing operating SRT. The most important observations were that all OUR values in the continuous activated sludge system were higher than those in the SBR for any realistic SRT, since the continuous activated sludge system had more predators (see the fraction of predators in Figs. 5 and 7). Finally, Table 1 summarizes and compares the key effects of predators for the continuous activated sludge system and
4622
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 1 6 e4 6 2 2
SBR operated at SRT ¼ 10 days. Because predators have more profound impacts on biomass components and oxygenuptake rates under famine conditions, the continuous system is affected more significantly than the SBR system: higher Xpred and OUR, but lower XH and XEPS. This greater impact comes about because the continuous system operated at all times in a situation of near-famine for any realistic SRT.
4.
Conclusions
The following conclusions can be drawn from the studies on the impacts of predators on biomass components and oxygenuptake rates in activated sludge under feast-and-famine conditions: C Eliminating predators from the model increased XH and XEPS fractions significantly, and this caused the predictions for OUR to be substantially too low and for MLVSS to be too high (compared to experimental results and to predictions that included predation) once the famine period began. An analysis of OUR distribution shows that predation of XH was the most significant OUR process under famine conditions. C When used to analyze the operation of an SBR, the new model showed that the soluble and biomass components required different time periods to reach steady state for an SBR. XSTO stabilized after 30 days of operation, XH and XEPS reached plateaus at about 40 days (4 times SRT), Xpred took less time (about 30 days) to reach steady state, and XI took at least 80 days (8 times SRT) to reach steady state. MLVSS increased with increasing SRT when SRT was greater than the washout SRT of 0.19 day, but the active biomass became a smaller fraction with increasing SRT, while the inert biomass became increasingly dominant. Perhaps most important is that predators constituted their maximum fraction of the MLVSS at an SRT of about 13 days, but washed out for SRT < w3 days. C Simulation for the continuous system indicated that XH and XEPS reached plateaus at about 3 SRT, Xpred took more time (about 800 h) to reach steady state, and XI took at least 1700 h (7 times SRT) to reach steady state. At SRT ¼ 100 days, XH was 2.3%, and XI was 95%. The predators reached their maximum fraction (w5.5%) at an SRT of w14 days. C Comparison of the predation in the SBR and the continuous system reveals that the predators have a larger presence and more of an impact in continuous systems because the continuous activated sludge system is in permanent near-famine. As a consequence, the continuous system has higher Xpred and OUR, but lower XH and XEPS.
Acknowledgements The authors wish to thank the Natural Science Foundation of China (50625825 and 50738006), the National Key Project for Water Pollution Control (2008ZX07316-002 and 2008ZX07010005) for the partial support of this study.
references
APHA, 1995. Standard Methods for the Examination of Water and Wastewater, nineteenth ed.. American Public Health Association, Washington, DC. Beun, J.J., Heijnen, J.J., van Loosdrecht, M.C.M., 2001. N-removal in a granular sludge sequencing batch airlift reactor. Biotechnol. Bioeng. 75, 82e92. Beun, J.J., van Loosdrecht, M.C.M., Heijnen, J.J., 2002. Aerobic granulation in a sequencing batch airlift reactor. Water Res. 36, 702e712. Furumai, H., Kazmi, A.A., Furuya, Y., Sasaki, K., 1999. Effect of sludge retention time (SRT) on nutrient removal in sequencing batch reactors. J. Envir. Sci. Health 34, 317e328. Irvine, R.L., Busch, A.W., 1979. Sequencing batch biological reactorean overview. J. Water Pollut. Control Fed. 51, 235e243. Irvine, R.L., Ketchum, L.H., 1989. Sequencing batch reactors for biological wastewater treatment. Crit. Rev. Environ. Control 18, 255e294. Karahan, O., Martins, M., Orhon, D., van Loosdrecht, M.C.M., 2006. Experimental evaluation of starch utilization mechanism by activated sludge. Biotechnol. Bioeng. 93, 964e970. Kortekaas, S., Vidal, G., He, Y.L., Lettinga, G., Field, J.A., 1998. Anaerobiceaerobic treatment of toxic pulping black liquor with upfront effluent recirculation. J. Ferment. Bioeng. 86, 97e110. de Kreuk, M.K., Heijnen, J.J., van Loosdrecht, M.C.M., 2005. Simultaneous COD, nitrogen, and phosphate removal by aerobic granular sludge. Biotechnol. Bioeng. 90, 761e769. de Kreuk, M.K., van Loosdrecht, M.C.M., 2006. Formation of aerobic granules with domestic sewage. J. Environ. Engin. 132, 694e697. van Loosdrecht, M.C.M., Pot, M., Heijnen, J., 1997. Importance of bacterial storage polymers in bioprocesses. Water Res. 35, 41e47. Moussa, M.S., Hooijmans, C.M., Lubberding, H.J., Gijzen, H.J., van Loosdrecht, M.C.M., 2005. Modeling nitrification, heterotrophic growth and predation in activated sludge. Water Res. 39, 5080e5098. Ni, B.J., Fang, F., Rittmann, B.E., Yu, H.Q., 2009. Modeling microbial products in activated sludge under feast-famine conditions. Environ. Sci. Technol. 43, 2489e2497. Ni, B.J., Rittmann, B.E., Yu, H.Q., 2010. Modeling predation processes in activated sludge. Biotechnol. Bioeng. 105, 1021e1030. Otawa, K., Asano, R., Ohba, Y., Sasaki, T., Kawamura, E., Koyama, F., Nakamura, S., Nakai, Y., 2006. Molecular analysis of ammonia-oxidizing bacteria community in intermittent aeration sequencing batch reactors used for animal wastewater treatment. Environ. Microbiol. 8, 1985e1996. Pratt, S., Yuan, Z., Keller, J., 2004. Modelling aerobic carbon oxidation and storage by integrating respirometric, titrimetric, and off-gas CO2 measurements. Biotechnol. Bioeng. 88, 135e147. Reichert, P., 1998. Aquasim 2.0-User Manual, Computer Program for the Identification and Simulation of Aquatic Systems. EAWAG, Dubendorf, Switzerland, ISBN 3 906484 16 5. Schwarzenbeck, N., Erley, R., McSwain, B.S., Wilderer, P.A., Irvine, R.L., 2004. Treatment of malting wastewater in a granular sludge sequencing batch reactor (SBR). Acta Hydrochim. Hydrobiol. 32, 16e24. Segar, R.L., Dwys, S.L., Speitel, G.E., 1995. Sustained trichloroethylene cometabolism by phenol-degrading bacteria in sequencing batch reactors. Water Environ. Res. 67, 764e774. Siegrist, H., Vogt, D., Garcia-Heras, J.L., Gujer, W., 2002. Mathematical model for meso- and thermophilic anaerobic sewage sludge digestion. Environ. Sci. Technol. 36, 1113e1123. Timur, H., Ozturk, I., 1999. Anaerobic sequencing batch reactor treatment of landfill leachat. Water Res. 33, 3225e3230.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Review
Fate of antibiotics during municipal water recycling treatment processes N. Le-Minh a, S.J. Khan a,*, J.E. Drewes a,b, R.M. Stuetz a a b
UNSW Water Research Centre, School of Civil and Environmental Engineering, University of New South Wales, NSW 2054, Australia Advanced Water Technology Center (AQWATEC), Colorado School of Mines, Golden, CO 80401, USA
article info
abstract
Article history:
Municipal water recycling processes are potential human and environmental exposure
Received 5 January 2010
routes for low concentrations of persistent antibiotics. While the implications of such
Received in revised form
exposure scenarios are unknown, concerns have been raised regarding the possibility that
5 May 2010
continuous discharge of antibiotics to the environment may facilitate the development or
Accepted 8 June 2010
proliferation of resistant strains of bacteria. As potable and non-potable water recycling
Available online 15 June 2010
schemes are continuously developed, it is imperative to improve our understanding of the fate of antibiotics during conventional and advanced wastewater treatment processes
Keywords:
leading to high-quality water reclamation. This review collates existing knowledge with
Pharmaceutically active compounds
the aim of providing new insight to the influence of a wide range of treatment processes to
Antibiotics
the ultimate fate of antibiotics during conventional and advanced wastewater treatment.
Wastewater treatment
Although conventional biological wastewater treatment processes are effective for the
Advanced treatment
removal of some antibiotics, many have been reported to occur at 10e1000 ng L1
Potable reuse
concentrations in secondary treated effluents. These include b-lactams, sulfonamides, trimethoprim, macrolides, fluoroquinolones, and tetracyclines. Tertiary and advanced treatment processes may be required to fully manage environmental and human exposure to these contaminants in water recycling schemes. The effectiveness of a range of processes including tertiary media filtration, ozonation, chlorination, UV irradiation, activated carbon adsorption, and NF/RO filtration has been reviewed and, where possible, semi-quantitative estimations of antibiotics removals have been provided. ª 2010 Elsevier Ltd. All rights reserved.
Contents 1. 2. 3.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Analytical methods for determining antibiotics in wastewater . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Removal of antibiotics during conventional sewage treatment processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. b-Lactams . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Sulfonamides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Trimethoprim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
* Corresponding author. Tel.: þ61 2 93855082; fax: þ61 2 93138624. E-mail address:
[email protected] (S.J. Khan). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.020
4296 4296 4297 4300 4300 4304
4296
4.
5.
1.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
3.4. Macrolides . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. Fluoroquinolones . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6. Tetracyclines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7. Nitroimidazoles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.8. Other antibiotic groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.9. Effects of antibiotics on wastewater microbial consortia/processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Fate of antibiotics during advanced treatment processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Membrane filtration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2. Adsorptive treatment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1. Activated carbon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2. Ionic adsorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3. Chemical and photochemical oxidation processes for the removal of antibiotics . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.1. Chlorination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.2. Ozonation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.3. Ultraviolet irradiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.4. Advanced oxidation processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Introduction
Municipal water recycling for industrial, agricultural, and non-potable municipal uses is an increasingly important component of water resources management practices in many parts of the world (Exall, 2004; Vigneswaran and Sundaravadivel, 2004; Wintgens et al., 2005). In some countries, such as the USA, Singapore, Mexico and Belgium, treated effluents are intentionally used to supplement drinking water supplies, a process known as planned indirect potable reuse (planned IPR) (Drewes and Khan, 2010; Rodriguez et al., 2009). Planned IPR is rapidly emerging as an important water supply strategy for a number of Australian cities (Khan, 2009). In Windhoek, Namibia, direct potable reuse of highly treated effluents for drinking water supply has been practiced since 1969 (du Pisani, 2006) and it is possible that other countries may adopt this strategy in the future. Pharmaceuticals including antibiotics are present in municipal sewage, largely as a result of human excretion. Many active antibiotics are not completely metabolised during therapeutic use and thus enter sewage through excretion in an unchanged form (Hirsch et al., 1999). The intentional disposal of unused drugs into the sewer (Kummerer, 2003) and veterinary use (Diaz-Cruz et al., 2003) also contribute to the quantities of antibiotics found in sewage. Discharges from veterinary clinics and runoff from agricultural applications into municipal sewers are also potential sources of veterinary antibiotics in wastewater. The reported levels of specific antibiotic drugs detected in raw sewage appear to differ between countries, possibly reflecting variable prescription practices (Miao et al., 2004) and differences in per-capita water consumption leading to various degrees of dilution (Drewes et al., 2008). Seasonal variations in sewage concentrations of antibiotics have also been reported (Alder et al., 2006). Antibiotic drugs have been identified as a particular category of trace chemical contaminants, which warrant close scrutiny (Watkinson et al., 2007). Much of the concern regarding the presence of antibiotics in wastewater and their
4307 4307 4308 4308 4308 4309 4309 4309 4312 4312 4314 4314 4314 4315 4316 4316 4316 4317 4317
persistence through wastewater treatment processes is related to concerns that they may contribute to the prevalence of resistance to antibiotics in bacterial species in wastewater effluents and surface water near wastewater treatment plants (WWTPs) (Adelowo et al., 2008; Auerbach et al., 2007; Baquero et al., 2008; Jury et al., in press). Reusing treated effluents for non-potable or potable purposes increases the range of human and environmental exposure scenarios for bacteria potentially harbouring antibiotics resistance. Accordingly, a thorough understanding of the effectiveness of treatment processes employed in water recycling schemes is warranted.
2. Analytical methods for determining antibiotics in wastewater Many antibiotics are non-volatile with high molecular weight, which tends to render them more suited to analysis by liquid chromatography (LC) rather than gas chromatography (GC) (Choi et al., 2007a). The determination of antibiotic residues by LC with spectrophotometric detection has been reported including fluorescence and ultraviolet (UV) absorbance (Choi et al., 2007a; Esponda et al., 2009; Golet et al., 2002b; Jen et al., 1998; Li et al., 2007; Peng et al., 2008). However, a survey of literature by Herna´ndez et al. (2007) revealed the impressive progress and focus on method development using liquid chromatography-mass spectrometry (LC-MS) and particularly liquid chromatography-tandem mass spectrometry (LC-MS/MS) to determine antibiotics in complex matrices such as municipal wastewater. This review also summarised the important factors affecting the analyses of different classes of antibiotics, such as pH adjustment, sample container materials, storage conditions, the addition of chelating agents, solvent types and matrix interference. Sample extraction for both clean-up and enrichment is commonly undertaken with typical concentration factors in the range 100e1000 required for the necessary low limits of detection (LOD). Although a variety of techniques have been
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
employed for extracting antibiotic, such as lyophilisation (Hirsch et al., 1998) or liquideliquid extraction (LLE) (Jen et al., 1998), the most frequently used technique is solid phase extraction (SPE) with hydrophilic e lipophilic balanced (HLB) cartridges. Fatta et al. (2007) have surveyed the use of several common SPE sorbents and confirmed that Oasis HLB was among the best performing, yielding high recoveries for multiple analyses of acid, neutral and basic analytes including antibiotics. Polar solvents such as acetone, methanol and acetonitrile have been employed for extraction and LC separation. Most recently, solid phase micro extraction (SPME) (Balakrishnan et al., 2006) and SPE coupled online to LC-MS/MS (Stoob et al., 2005) have emerged as alternative approaches for the analysis of antibiotics. Although SPME often yields poorer LODs than the conventional SPE extraction for wastewater samples, the benefits are reduction in solvent usage and significant savings of the analytical time (Fatta et al., 2007). Comparatively, online SPE methods appear to provide improved extraction recovery, reduction in costs of consumable extraction materials and analysis time and similar sensitivity to conventional offline SPE methods (Stoob et al., 2005). Regarding the analysis of antibiotics in solid matrices, a review by Kim and Carlson (2005) summarised different techniques including LLE, accelerated solvent extraction (ASE), ultrasonic solvent extraction (USE) and pressurised liquid extraction (PLE) to desorb antibiotics from wastewater sludge, manure, soil and sediment. During their method development study, Gobel et al. (2005b) compared extraction efficiencies for antibiotics in activated sludge between PLE (50% methanol 50% water) and USE (methanol and acetone). The latter technique was reported to be slightly less efficient for macrolides and trimethoprim, while the extraction efficiency for sulfonamides was significantly lower (by 20e60%) compared to PLE. Analytical methods developed to determine trace concentrations of antibiotics in wastewater and sludge are summarised in Table 1. The sensitivity, recovery and range of antibiotics included in individual methods are highly variable. This variation is largely due to the wide structural variability of these compounds and their inability to respond similarly to the same extraction and analytical procedures, differences in the use of internal standards and analytical instruments, and differences in concentration factors employed in each extraction procedure. However, most methods employing SPE-LCMS/MS have achieved good sensitivity with LODs reaching the nanogram or even sub-nanogram per litre range. Meanwhile, a few methods have been reported to analyse antibiotics in sewage sludge (Gobel et al., 2005b; Golet et al., 2002b). For these methods, LODs between 0.6 and 5.1 ng g1 were reported for macrolides, sulfonamides and trimethoprim using PLESPE-LC-MS/MS (Gobel et al., 2005b) and 450 ng g1 for fluorquinolones using ASE-LC-FLD (Golet et al., 2002b).
3. Removal of antibiotics during conventional sewage treatment processes The occurrence of some common antibiotics and performance of conventional WWTPs for removing them as reported in the literature are summarised in Table 2. It is apparent that removal of antibiotics during conventional wastewater
4297
treatment processes is highly variable. However, some activated sludge (AS) processes appear to be effective for the efficient removal of several of these compounds. During treatment, antibiotics can be transformed or removed from the aqueous phase by hydrolysis, biotransformation, or sorption to primary and secondary sludges. Hydrophobic chemicals are expected to occur at higher concentration in primary sludge than hydrophilic chemicals because they have a greater affinity to solids and hence concentrate in the organic-rich sewage sludge (Beausse, 2004). Pharmaceuticals can also be removed from aqueous solution onto solid particulates by ion exchange, complex formation with metal ions, and polar hydrophilic interactions (Diaz-Cruz et al., 2003). Pharmaceuticals, adsorbed to flocs, suspended solids and/or activated (microbial) sludge, will be removed from the aqueous phase by sedimentation and subsequent disposal of excess sludge. The affinity of antibiotics adsorbed to sludge has been represented by sludge sorption constants Kd (L kg1), shown in Table 3. The greater Kd values represent the greater adsorption of the compounds to sludge. The sludge may be delivered to anaerobic digesters for stabilisation before being used in agricultural soil amendment and/or biogas generation. Those antibiotics, which are hydrophilic and highly resistant to most conventional biological treatment processes, are expected to mainly remain in the aqueous phase of the treated effluent. The operating conditions of a wastewater treatment process such as temperature, solids retention time (SRT), and hydraulic retention time (HRT), can significantly affect the removal efficiency of many pharmaceutical contaminants. While ambient temperature is not practical to control, SRT and HRT can be adjusted to some degree in order to optimise removal efficiencies. Increasing SRTs were reported to enhance the removal of several pharmaceutical compounds during aerobic biological processes (Clara et al., 2005a; Kim et al., 2005; Yasojima et al., 2006). The extended SRTs have been suggested to allow for the enrichment of slower growing bacterial species and therefore, to provide greater diversity of enzymes, some of which are capable of breaking down the pharmaceutical compounds (Jones et al., 2007). Similarly, longer HRT is expected to provide sufficient reaction time for biotransformation and sludge sorption to occur in order to reach maximum efficiency at equilibrium. Consistent with this expectation, a reduction in the removal of some pharmaceuticals during wastewater treatment processes has been observed to be generally correlated with reducing HRT (TauxeWuersch et al., 2005; Vieno et al., 2007a). These observations were made during rainfall events, resulting in increased dilution and reduced HRT. Kim et al. (2005) pointed out that during dry weather operation, shorter HRT (implying a greater substrate loading) will result in higher biomass concentration which could enhance the removal, thus compensating for shorter reaction times. Accordingly, it is difficult to generalise the expected relationship between HRT and pharmaceutical removal without accounting for the effect on biomass concentration. A number of published studies have reported SRT and HRT in conjunction with observed removal of antibiotics during membrane bioreactor (MBR) and activated sludge (AS) treatment processes as summarised in Table 4. The studies were selected to include only those based on 24 h composite sampling regimes in order to consider the specific HRT, thus
Target compounds
Sample matrices
Extraction, (sample size), pH
Solvents
Internal standards
Instruments
Recoveries (relative)
LOD (ng/L) or (ng/g)
Reference
Wastewater
SPE Oasis HLB (200 mL), pH 7.5
Methanol Acetonitrile Formic trifluoroacetic acid
Penicillin G
LC-MS/MS (þve ESI) LCQ Duo Ion Trap
>70% Amoxicillin 10%
13e18 (Inf); 8e15 (Eff)
(Cha et al., 2006)
16 Sulfonamides Trimethoprim
Wastewater
SPE Oasis HLB (250 - 500 mL) Silica gel
Methanol Formic acid
13C6 - sulfamethoazine
LC-MS/MS (þve ESI) Water QqQ
62e102%
0.02e0.2 (Inf); 0.016e0.12 (Eff)
(Chang et al., 2008)
6 Sulfnonamides Trimethoprim 5 Macrolides
Wastewater
SPE Oasis HLB (50 - 250 mL), pH 4
Methanol Formic acid Ethyl acetate
Isotope label sulfonamides 13C2-erythromycin
LC-MS/MS (þve ESI) TSQ Quantum QqQ
91e108% 30e47% 78e124%
11e68 (Inf); 1.2e9.6 (Eff) 4.5e8.1 (Inf); 0.9e2.7 (Eff) 0.36e3.9 (Inf); 0.09e2.9 (Eff)
(Gobel et al., 2004)
2 Macrolides Sulfamethoxazole Trimethoprim Ofloxacin
Wastewater
SPE Oasis HLB (100 - 200 mL)
Methanol/ acetonitrile Ammonium acetate
13C-phenacetin Carbamazepine-d10
LC-MS/MS (þve ESI) Water QqQ
40e116% 50e80% 88e111% 95e106%
6e7 (Inf); 3e6 (Eff) 42 (Inf); 20 (Eff) 25 (Inf); 10 (Eff) 43 (Inf); 43 (Eff)
(Gros et al., 2006)
Sulfnonamides Trimethoprim Macrolides
Sewage sludge
PLE, 0.2 g, pH 4 SPE Oasis HLB
Methanol Formic acid Ethyl acetate
Isotope label for sulfonamides 13C2-erythromycin for macrolides
LC-MS/MS (þve ESI) TSQ Quantum QqQ
79e106% 78% 91e142%
0.9e15 2.7e5.1 0.6e2.4
(Gobel et al., 2005b)
7 Sulfonamides
Swine wastewater
LLE, (50 mL), pH 6.6
Nicotinamide Ethyl acetate
N/A
LC-UV
86e99%
4000e15 000
(Jen et al., 1998)
2 Fluoroquinolones
Sewage sludge
ASE
Acetonitrile
LC-FLD
82e94%
450
(Golet et al., 2002b)
5 Sulfonamides
Wastewater
SPE mixed hemimicelles column, pH 2
Methanol Acetinitrile
N/A
LC-UV
89e113%
150e350
(Li et al., 2007)
5 Fluoroquinolones 3 Sulfonamides Trimethoprim
Wastewater
SPE Oasis HLB, (1000 mL), pH 2.5
Methanol Acetonitrile
Sulfamerazine Standard addition
LC-MS
90e129% 37e65% 98e109%
20e40 (Eff) 40e90 (Eff) 40e50 (Eff)
(Renew and Huang, 2004)
5 Macrolides 2 Ionophores Tiamulin
Liquid manure
LLE, (15 g), pH 8
Ethyl acetate Acetonitrile Ammonium acetate
(E)-9-[O-(2-methyloxime)]erythromycin
LC-MS/MS (þve ESI) TSQ7000 QqQ
78e94% 119% 123%
0.4e27.9 3.2e17.9 0.4
(Schlu¨sener et al., 2003)
Gentamicin
Hospital wastewater
SPE Widepores CBX, (20e50 mL), pH 7e8
Methanol Kanamycin Acetic acid Heptafluorobutyric acid
LC-MS/MS (þve ESI) API 365
107e111%
200
(Lo¨ffler and Ternes, 2003)
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
5 Beta lactams
4298
Table 1 e Summary of analytical methods for determination of antibiotics in wastewater and sludge.
Wastewater
SPE Anpel MEP, (100e500 mL), pH 3
Methanol Formic acid Acetonitrile Tetrabutyl ammonium bromide
N/A
LC-FLD
10 Sulfonamides
Wastewater
SPME Supelco CW/TRP (25 mL)
Methanol Formic acid Ammonium acetate
13C6sulfamethazine
LC-MS/MS (þve ESI) Quattro Ultima QqQ
20 Quinolones and Fluoroquinolones
Wastewater
SPE Oasis HLB (200e400 mL), pH 3 2nd SPE Water WCX
Methanol Formic acid
Norfloxacin-d5
LC-MS/MS (þve ESI) Waters Premier XE
3 Macrolides
Wastewater
SPE Oasis HLB (120 mL), pH 5
Methanol Acetonitrile Formic acid
Simatone
6 Tetracyclines 5 Sulfonamides
Wastewater
SPE Oasis HLB (120 mL), pH 3
Methanol Acetonitrile Formic acid
7 Macrolides 2 Fluoroquinolones Sulfamethoxazole Oxytetracycline Amoxycillin
Wastewater
SPE Oasis MCX (500 mL), pH 2 SPE Lichrolut EN (500 mL), pH 7
2 Fluoroquinolones 2 Macrolides Sulfamethoxazole Trimethoprim
Wastewater
5 Sulfonamides Trimethoprim
7 Beta lactams 3 Sulfonamides
(Shi et al., 2009)
9040e55300 (Inf)
(Balakrishnan et al., 2006)
1.6e50 (Inf); 0.6e50 (Eff)
(Xiao et al., 2008)
LC-MS/MS (þve ESI) 83e86% Finnigan LCQ Ion Trap
30e70 (Eff)
(Yang and Carlson, 2004)
Simatone
LC-MS/MS (þve ESI) 78e95% Finnigan LCQ Ion Trap 91e104%
40e70 (Inf); 30e50 (Eff) 40e60 (Inf); 30e40 (Eff)
(Yang et al., 2005)
Methanol Ethyl acetate Acetone Acetonitrile
Salbutamol-d3 Ibuprofen-d3
LC-MS/MS (þve ESI) API3000 QqQ
47e76% 31e32% 65% 73% 36%
0.2e1.4 (Eff) 1.3e1.8 (Eff) 1.5 (Eff) 1.2 (Eff) 2.1 (Eff)
(Castiglioni et al., 2005)
SPE Strata-X þ XC (250 mL), pH 3
Methanol Formic acid Acetonitrile
Lomefloxacin Josamycin Diaveridine
LC-MS/MS (þve ESI) TSQ Quantum QqQ
76e97% 92e100% 68% 104%
4e21 (Inf) 0.3e12 (Inf) 22 (Inf) 7 (Inf)
(Segura et al., 2007)
Wastewater
SPE Oasis HLB (50 mL), pH 4
Methanol Acetonitrile Formic acid
Sulfathiazole-d4 Sulfamethoxazole-d4
LC-MS-MS (þve ESI) TSQ Quantum QqQ
72e110% 80e103%
7e10 (Eff) 7 (Eff)
(Botitsi et al., 2007)
Wastewater
SPE Oasis HLB (250 mL), Methanol pH 3
Caffeine
UPLC-MS/MS (þve ESI) 56e93%
4.1e84 (Inf); 3.8e60 (Eff) 3.0e3.3 (Inf); 1.0 (Eff) 4.6e7.0 (Inf); 2.8e5.0 (Eff) 6.8e14 (Inf); 5.1e8.1 (Eff) 0.5e37 (inf); 0.3e26 (Eff) 2.7 (Inf); 1.1 (Eff)
(Li et al., 2009)
64e127% piromidic acid <29%
80e104%
3 Fluoroquinolones
86e105%
3 Tetracyclines
83e96%
3 Macrolides
73e93%
Trimethoprim
90e97%
Inf: influent; Eff: Effluent; SPE: Solid phase extraction; LLE: Liquideliquid extraction; ASE: Accelerated solvent extraction, PLE: Pressurised liquid extraction; SPME: Solid phase micro extraction; FLD: Fluorescence detector; UPLC: Ultra pressurised liquid chromatography; þve ESI: Positive electro-spray ionisation.
4299
100e1060
Acquity TQ QqQ
79e109%
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
8 Fluoroquinolones
4300
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
accounting for influent variability. It can be seen from Table 4, while some relationship between SRT and antibiotic removal has been reported within some well controlled studies (Clara et al., 2005b; Gobel et al., 2007; Kim et al., 2005), no clear trends are evident between these studies. This suggests that other unreported factors, such as biomass concentration and diversity as well as substrate/biomass ratios, might be important to explain the variation. MBR systems have been reported to be equal to or more effective in removing several antibiotics compared to AS systems of similar SRT and HRT, possibly due to higher biomass concentration in the MBRs. Higher biomass concentration implies the reduced sludge loading (i.e., ratio between substrate and sludge concentration), which was reported to enhance biotransformation of several antimicrobial compounds (Gobel et al., 2007). Unfortunately, it is not possible to interpret this data in further detail by consideration of treatment design and operational parameters since there are only few studies which comprehensively presented operating parameters of the MBR processes. One very good example, which confirms the importance of these parameters and illustrates how this data could be provided, is given by Gobel et al. (2007).
3.1.
b-Lactams
b-Lactams, such as penicillins and cephalosporins, are narrow spectrum antibiotics, which are highly effective against the Gram-positive genera Streptococcus, Gonococcus, and Staphylococcus (Todar, 2002). These antibiotics act as bacteriostatics by inhibiting bacterial peptidoglycan cell wall synthesis (Marzo and Dal Bo, 1998). The four-membered ring, which all b-lactam drugs feature, is a strained, cyclic amide that is highly susceptible to chemical or enzymatic hydrolysis (Deshpande et al., 2004). The hydrolysed b-lactam drugs result in an inactive product when the ring is broken. The degradation of b-lactam antibiotics such as penicillin, takes place under acidic and alkaline conditions or by reactions with weak nucleophiles, such as water or metal ions (Aksu and Tunc, 2005; Hou and Poole, 1971). Alternatively, penicillin can be enzymatically hydrolysed by b-lactamase enzyme via the same way as acid hydrolysis. b-lactamases are the widespread enzymes in bacteria, and are produced by many species to inactivate the pharmacological effects of the beta-lactam antibiotics (Neu, 1992). Historically, in order to make ‘orally available’ penicillins, some chemical modifications were developed to reduce the susceptibility of penicillins to acid hydrolysis in stomach, which resulted in the availability of acid-resistant penicillins including amoxicillin and ampicillin (Sneader, 2005). However, since these drugs are still susceptible to enzymatic hydrolysis, they do not appear to persist during sewage treatment (Table 2). The study by Li et al. (2008) focusing on Penicillin G reported concentration of 153 mg L1 in raw sewage and 1.68 mg L1 in treated effluent, respectively, revealing that Penicillin G had undergone partial transformation during the anaerobic, aerobic and hydrolysis processes at the WWTP. A number of studies have reported that the presence of blactams in treated wastewater samples are generally not detected or only at very low concentrations, despite them being among the most commonly prescribed antibiotics
(Cahill et al., 2004; Cha et al., 2006; Costanzo et al., 2005; Hirsch et al., 1999; Watkinson et al., 2007; Zuccato et al., 2005). Although b-lactam antibiotics have been reported to dominate the overall antibiotic concentration in some sewage influents, they tend to be significantly reduced in concentrations during biological processes (Watkinson et al., 2007). The analysis of five b-lactams in WWTP influent revealed that while chloxacillin and oxacillin were observed in three out of 72 influent samples (at concentration less than 20 ng L1), none of these blactams were detected in effluent samples (Cha et al., 2006). The significant removal (>96%) of cephalexin from 2000 ng L1 to 78 ng L1 has been reported through conventional WWTP processes in Australia (Costanzo et al., 2005). Similarly, Morse and Jackson (2004) concluded that amoxicillin, a representative b-lactam drug, is quite susceptible to microbial degradation and therefore is not likely to remain in significant concentration after biological treatment systems.
3.2.
Sulfonamides
Sulfonamides and trimethoprim are bacteriostatic agents that synergistically target and inhibit two pathway steps in bacterial folic acid synthesis (Masters et al., 2003; Skold, 2001). Folate derivatives are essential cofactors in the biosynthesis of purines, pyrimidines and bacterial DNA in all living cells. Therefore, blocking this pathway inhibits the production of reduced folates and eventually the synthesis of nucleic acid, which in turn affects bacterial growth. When combined, sulfonamides and trimethoprim afford an effective treatment against a variety of potential bacterial infections. Sulfonamides are not completely metabolised during use and are excreted via urine into sewage, partly as unchanged parent compounds and partly as metabolites (Gobel et al., 2005a; Hirsch et al., 1999). The major metabolites of sulfonamides entering sewage are biologically inactive N4-acetylated products, for which transformations back to the active parent compounds during sewage treatment has been reported (Gobel et al., 2005a). This phenomenon may have led to apparent negative removal of some sulfonamides, particularly sulfamethoxazole, during biological wastewater treatment (Gobel et al., 2007; Karthikeyan and Meyer, 2006). Sulfamethoxazole is among the most frequently detected sulfonamides in municipal sewage (Brown et al., 2006; Choi et al., 2007a; Gobel et al., 2007; Levine et al., 2006; Yang et al., 2005). However, concentrations of this drug in WWTP influents and effluents vary significantly, depending on antibiotic consumption patterns and the types of wastewater treatment processes employed. For example, sulfamethoxazole was reported with concentrations as high as 7.91 mg L1 in sewage influent in China, where the compound is one of the top 15 pharmaceuticals sold (Peng et al., 2006). Sulfamethoxazole removal efficiencies by conventional WWTPs have been reported to range from 279% to 100% (Table 2). Sulfamethoxazole’s acetylated metabolite, N4acetylsulfamethoxazole usually accounts for greater than 50% of an administered dose in human excretion (Gobel et al., 2004) and can occur in WWTP influents at concentrations 2.5e3.5 times higher than concentrations of the parent compound (Gobel et al., 2007). Despite occurring at high concentration in the raw influent, N4-acetylsulfamethoxazole does not appear to partition significantly into sludge (Gobel et al., 2005a).
Table 2 e Occurrence of common antibiotics in WWTPs. Analytes/Location
Main treatment practices
Removal Efficiency (%)
Reference
Cephalexin Australia Australia China Taiwan
2000 5600 670e2900 1563e4367
80 <(2) 240e1800 10e994
Activated sludge system (AS) AS Chemical enhanced/Secondary treatment Secondary treatments/UV or chlorination
96a e 9 to 89b 36 to 99.8b
(Costanzo et al., 2005) (Watkinson et al., 2007) (Gulkowska et al., 2008) (Lin et al., 2009)
Amoxicillin Australia
280
<(3) 30
AS
e
(Watkinson et al., 2007)
Cloxacillin Australia USA
<(1)e320 <(13)e15
<(1) <(9)
AS Secondary treatment/Chlorination
e e
(Watkinson et al., 2007) (Cha et al., 2006)
Penicillin G Australia
<(2)
<(2)
AS
e
(Watkinson et al., 2007)
Penicillin V Australia
160
80
AS
e
(Watkinson et al., 2007)
Sulfamethoxazole USA Korea China Croatia Switzerland Mexico USA Sweden Spain Spain Sweden Austria China Taiwan
1090 450 5450e7910 590 230e570 390 <(50)e1250 20 580 0.60 <(80)e674 24e145 10e118 179e1760
210 <(30) <100 0.39 210e860 0.31 <(50)e210 70 250 NA <(80)e304 18e91 9e78 47e964
AS/chlorination AS AS/filtration/chlorination AS AS/Sand filtration AS AS AS/Chemical P removal AS AS Chemical P removal/AS AS AS or Chemical enhanced/UV or chlorination Secondary treatments/UV or chlorination
81b >93a >98a 33a e 20b 18e100b e 67b 57b 42b 279 to 66b 34e63b 26e88b
(Yang et al., 2005) (Choi et al., 2007a) (Peng et al., 2006) (Gros et al., 2006) (Gobel et al., 2005a) (Brown et al., 2006) (Karthikeyan and Meyer, 2006) (Bendz et al., 2005) (Carballa et al., 2004) (Carballa et al., 2005) (Lindberg et al., 2005) (Clara et al., 2005b) (Xu et al., 2007) (Lin et al., 2009)
N4-sulfamethoxazole Switzerland
850e1600
<(20)e180
AS/Sand filtration
e
(Gobel et al., 2005a)
Sulfathiazole Korea
10 570
180
AS
98a
(Choi et al., 2007a)
Sulfamethazine USA Korea
150 4010
<(30) <(30)
AS/chlorination AS
>80a >99a
(Yang et al., 2005) (Choi et al., 2007a)
Sulfadimethoxine USA Korea
70 460
<(30) <(30)
AS/chlorination AS
> 57a >93a
(Yang et al., 2005) (Choi et al., 2007a) (continued on next page)
4301
Effluent conc. (ng/l)
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
Influent conc. (ng/l)
Analytes/Location
Influent conc. (ng/l)
Effluent conc. (ng/l)
Main treatment practices
Removal Efficiency (%)
Reference
5100e5150 <(1)e72
<(150) <(1)e36
AS/filtration/chlorination AS or Chemical enhanced/UV or chlorination
> 97a 50b
(Peng et al., 2006) (Xu et al., 2007)
Sulfamerazine Korea
1530
< (30)
AS
> 98a
(Choi et al., 2007a)
Trimethoprim Croatia Switzerland Mexico USA Sweden UK Sweden China Taiwan
1172 210e440 0.59 0.14e1.10 80 213e300 99e1300 120e320 259e949
290 20e310 180 <(50)e550 40 218e322 66e1340 120e230 203e415
AS AS AS AS AS/chemical P removal Trickling filter/AS/UV Chemical P removal/AS Chemical enhanced/Secondary treatment Secondary treatments/UV or chlorination
75a 64b 70b 50 to 100b 49b 3b 3b 17 to 62b 22e56b
(Gros et al., 2006) (Gobel et al., 2005a) (Brown et al., 2006) (Karthikeyan and Meyer, 2006) (Bendz et al., 2005) (Roberts and Thomas, 2006) (Lindberg et al., 2005) (Gulkowska et al., 2008) (Lin et al., 2009)
Doxycycline USA Korea Sweden
210 220 <(64)e2480
70 30 <(64)e915
AS/chlorination AS Chemical P removal/AS
67b 86a 70a
(Yang et al., 2005) (Choi et al., 2007a) (Lindberg et al., 2005)
Tetracycline USA Korea USA China Taiwan
200 110 240e790 96e1300 46e234
< (30) <(0.03) <(50)e160 180e620 16e38
AS/chlorination AS AS Chemical enhanced/Secondary treatment Secondary treatments/UV or chlorination
>85a >73a 68 to 100b 88 to 73b 66e90b
(Yang et al., 2005) (Choi et al., 2007a) (Karthikeyan and Meyer, 2006) (Gulkowska et al., 2008) (Lin et al., 2009)
Chlortetracycline USA Korea
270 970
60 40
AS/chlorination AS
78a 96a
(Yang et al., 2005) (Choi et al., 2007a)
Oxytetracycline Korea
240
<(30)
AS
>88a
(Choi et al., 2007a)
Ciprofloxacin USA Sweden Australia Switzerland Sweden China
<(50)e310 90e300 90 320e570 320 80
<(50)e60 <(6)e60 130 60e90 31.5 27
AS Chemical P removal/AS AS AS/Fe flocculation AS/Chemical P removal/sand filtration Secondary treatment
22e100b 87b 44a 83a 90b 66a
(Karthikeyan and Meyer, 2006) (Lindberg et al., 2005) (Costanzo et al., 2005) (Golet et al., 2003) (Zorita et al., 2009) (Xiao et al., 2008)
Norfloxacin Switzerland Sweden Sweden
340e520 66e174 18
40e60 <(7)e37 <(5.5)
Act. Sludge/Fe flocculation Chemical P removal/AS AS/Chemical P removal/sand filtration
88a 87b >70a
(Golet et al., 2003) (Lindberg et al., 2005) (Zorita et al., 2009)
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
Sulfadiazine China China
4302
Table 2 (continued).
110e460 54e263 339
85e320 27e85 85
Chemical enhanced/Secondary treatment AS or Chemical enhanced/UV or chlorination Secondary treatment
20 to 78b 50 to 82b 75a
(Gulkowska et al., 2008) (Xu et al., 2007) (Xiao et al., 2008)
<(43) 3520e5560 470 <(6)e287 22.5 80e368 115e1274 1208
<(43) <(80)e740 110 <(6)e45 10 41e165 53e991 503
AS AS/filtration/chlorination AS Chemical P removal/AS AS/Chemical P removal/sand filtration AS or Chemical enhanced/UV or chlorination Secondary treatment/UV or chlorination Secondary treatment
e >85b 77b 86b 56b 40 to70b <88b 58a
(Gros et al., 2006) (Peng et al., 2006) (Brown et al., 2006) (Lindberg et al., 2005) (Zorita et al., 2009) (Xu et al., 2007) (Lin et al., 2009) (Xiao et al., 2008)
Erythromycin Croatia Switzerland USA UK China China Taiwan
<(20) 60e190 <(50)e1200 71e141 470e810 253e1978 226e1537
<(20) 60e110 <(50)e300 145e290 520e850 216e2054 361e811
AS AS AS or aerated lagoon Trickling filter/Act. sludge/UV Chemical enhanced/Secondary treatment AS or Chemical enhanced/UV or chlorination Secondary treatments/UV or chlorination
e e 44 to 100b 79b 12 to 19b 15e45b <56b
(Gros et al., 2006) (Gobel et al., 2005a) (Karthikeyan and Meyer, 2006) (Roberts and Thomas, 2006) (Gulkowska et al., 2008) (Xu et al., 2007) (Lin et al., 2009)
Roxithromycin Switzerland USA Austria China
10e40 1500 25e117 75e164
10e30 870 36e69 35e278
AS/sand filtration AS or aerated lagoon AS AS or Chemical enhanced/UV or chlorination
e 42a 80 to 44b 53 to 76b
(Gobel et al., 2005a) (Karthikeyan and Meyer, 2006) (Clara et al., 2005b) (Xu et al., 2007)
Clarithromycin Switzerland Japan Taiwan
330e600 492e883 59e1433
110e350 266e444 12e232
AS/sand filtration AS Secondary treatments/UV or chlorination
21b 43b <0 to 99b
(Gobel et al., 2005a) (Yasojima et al., 2006) (Lin et al., 2009)
Azithromycin Japan
199e371
88e219
AS
49b
(Yasojima et al., 2006)
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
China China China Ofloxacin Croatia China Mexico Sweden Sweden China Taiwan China
Value in the parenthesis is the limit of detection in each study. a Removal efficiencies, not reported by authors in the cited study, are calculated from the average influent and effluent concentrations which were stated in the study. b Removal efficiency, either reported by authors in the cited study or calculated from the results of each sampling case.
4303
4304
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
Table 3 e Sorption constants of antibiotics to wastewater sludge.
Sulfapyridine Sulfamethoxazole
Trimethoprim Roxythromycin Azithromycin Clarithromycin Norfloxacin Ciprofloxacin Tetracycline
Log Kd
Condition
Reference
2.3e2.6 0.8e1.8 2.1e2.6 2.2e2.7 2.2e2.6 1.1e1.9 2.2e2.7 2.5e2.7 2.5e2.6 4.2 4.3 3.9
Activated Sludge Digested Sludge Activated Sludge Activated Sludge Activated Sludge Digested Sludge Activated Sludge Activated Sludge Activated Sludge Activated Sludge Activated Sludge Activated Sludge
(Gobel et al., 2005b) (Carballa et al., 2008) (Gobel et al., 2005b) (Joss et al., 2005) (Gobel et al., 2005b) (Carballa et al., 2008) (Joss et al., 2005) (Gobel et al., 2005b) (Gobel et al., 2005b) (Golet et al., 2003) (Golet et al., 2003) (Kim et al., 2005)
From the literature data, sulfonamides appear to be only partially removed by conventional WWTPs (Table 2). This partial removal may be attributed to moderate sorption to sludge and limited biodegradability. As a general rule, chemicals with logD < 2.5 are considered to have low hydrophobic sorption potential (Drewes, 2007). Therefore, sulfonamides are expected to be quite soluble and have a low potential for hydrophobic partitioning based on LogDpH6e8 between 1.8 and 1.3 (Table 5). Nonetheless, it has recently been demonstrated that the removal of sulfamethoxazole during MBR treatment is highly pH-dependant between pH 5e9 (Tadkaew et al., 2010). Since sulfonamides are neutral or negatively charged under typical WWTP operating conditions (pH 7e8) (Quiang and Adams, 2004), their binding to biomass via cation exchange with anionic sites or by metal complexation are also likely to be minimal. Consistent with this, Peng et al. (2006) observed insignificant partitioning to particulates during primary wastewater treatment processes. However, for those sulfonamides that are partially adsorbed to sludge, anaerobic sludge digestion has been shown to provide effective biotransformation (Carballa et al., 2006). In a laboratory-scale study by Sponza and Demirden (2007), a sequential upflow anaerobic sludge blanket system was reported to have a good removal performance for sulfamerazine (above 97%). Ingerslev and Halling-Sørensen (2000) have reported that while sulfonamides are ultimately biodegradable in AS systems, they exhibit a lag phase of 6e12 days at 20 C and 34e47 days at 6 C. This lag phase is the period after spiking, during which there was no significant degradation of the sulfonamides added to the AS system. However, by serial spiking tests, they observed that once bacteria in the reactors adapted to an initial introduction of four different sulfonamides, the bacterial culture rapidly degraded either a subsequent spike of the same four compounds or four other previously unexposed sulfonamides, illustrated by shorter degradation half-lives and no lag-phase, respectively. This implies that once the bacteria have adapted to degrade one sulfonamide, they may also be capable of efficiently degrading others. Some studies have observed that conventional WWTPs were effective in removing sulfamethoxazole (Choi et al., 2007a), while others appear to contradict this (Brown et al.,
2006). The contradiction may possibly be explained by the differences in WWTP operating conditions, such as SRT, HRT, and temperature. Moreover, differences in reported removal efficiencies may, in some cases, be attributed to limitations of employed mass balance techniques. For example, short-term variations of pharmaceutical loads in influent can be significant (Gobel et al., 2005a; Khan and Ongerth, 2005), thus care must be taken when comparing influent and effluent concentrations. Collecting composite samples over a period that is longer than the hydraulic retention time may improve the comparability between influent and effluent samples (Roberts and Thomas, 2006). A further complication is that sulfonamides are easily able to undergo reversible intertransformation with their respective metabolites (Gobel et al., 2007), such that if these metabolites are not considered, removal efficiencies may be underestimated or overestimated.
3.3.
Trimethoprim
Trimethoprim has been reported to occur in raw sewage of a number of countries including the USA (Karthikeyan and Meyer, 2006), Croatia (Gros et al., 2006), and Mexico (Brown et al., 2006). The presence of trimethoprim can generally be correlated to that of sulfamethoxazole since the two drugs are often administered in combination at a ratio 1:5 (Gobel et al., 2005a). Perez et al. (2005) reported that the concentration of trimethoprim in the primary effluent of a WWTP was around four times lower than that of sulfamethoxazole, which is relatively consistent with the typical medication ratio. Biodegradation experiments undertaken by HallingSorensen et al. (2000) showed the strong persistence of trimethoprim in AS batch reactors. The removal of trimethoprim during conventional biological wastewater treatment has been reported to significantly vary but is often incomplete (Brown et al., 2006; Gobel et al., 2007; Gros et al., 2006; Levine et al., 2006; Paxeus, 2004). Sorption to biomass appears to be negligible (Gobel et al., 2005a; Lindberg et al., 2005), as reflected by the low hydrophobic partitioning coefficient (LogDpH6e8) of the compound (Table 5). Only minor removal of trimethoprim during primary and secondary treatment has been reported (Gobel et al., 2005a; Perez et al., 2005). However, improved removal has been achieved by subsequent biologically active
Table 4 e Removal efficiency (%) of antibiotics by secondary treatment processes from studies reporting SRT and HRT, based on 24 h composite samples. Study
(Gobel et al., 2007)
(Golet et al., 2003) (Kobayashi et al., 2006) (Vieno et al., 2007a) (Yasojima et al., 2006)
SRT (d)
HRT (h)
NOR
CIP
AS AS AS (N) Lab MBR (S) AS MBR AS (N/DN) MBR MBR MBR AS (N/DN) AS (N/DN) AS AS/MF, UV AS (N/DN), Fe AS AS AS AS AS AS
5.6e8.2 10e12 7e16 72 2 10 10e12 16 33 60e80 21e25 20 10 12 10 9 7 7 7 8 5
15e22 9 12e20 12 2 12 5 13 13 13 31 12 7 9 9.6 12 10 7 7 6 4
50 75 80 to 87
66 79 to 86
ROX
AZI
CLA
ERY
SPY
SMX
53
15
<0
77 27 >62 <0 to 38 39 62 59 5 to 38
91
53 <0 61 <0 to 9 37 38 37 <0 to 60
<0 5 24 22 to 55
<0 to 9 57 41 88 4 to 20
<0 to 6 34 26 87 <0
<0 60 50 58 49 to 72
SMX þ AceSMX
TRI
36
<0 to 50 87 74 68 61 to 76
<0 to 14 30 34 87 <0 to 20
92 <0 >89
11 27
89 <0
15
21 20 61 29 26 41
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
(Xu et al., 2007) (Xiao et al., 2008) (Golet et al., 2002a) (Reif et al., 2008) (Clara et al., 2005b)
Main Treatment
AS: Activated sludge, N: nitrification; DN: denitrification, FBR: Fixed bed reactor, S: synthetic feed; Fe: Ferric chloride addition, MF: membrane filtration, MBR: membrane bioreactor, SRT: sludge retention time, HRT: hydraulic retention time, NOR: norfloxacin, CIP: ciprofloxacin, ROX: roxithromycin, AZI: azithromycin, CLA: clarithromycin, ERY: erythromycin, SPY: sulfapyridine, SMX: sulfamethoxazole, AceSMX: N4-aceytl sulfamethoxazole, TRI: trimethoprim. Removal efficiency <0 (less than zero) indicate that concentration of the effluent is greater than that of influent.
4305
4306
Table 5 e Acid dissociation constants and partitioning coefficients of several antibiotics. Antibiotic Category Beta-lactam
Sulfonamides
Functional groups corresponding to
Antibiotics
pKa1
pKa2
pKa4
Log Kow [5]
LogD pH6e8 [5] 1.9 to 2.8
6.88 [7]
0.65
1.9 to 2.8
5.86 6.77 7.42 5.57 7.07 6.28 8.40
1.5 0.34 0.80 0.89 0.047 0.12 0.034
1.3 to 0.10 0.30 to 0.68 0.79 to 0.16 0.49 to 0.90 0.02 to 0.76 0.23 to 1.5 0.08 to 0.16
2.829 na 3.159 3.33
0.72 to 2.4 na 1.1 to 2.8 0.65 to 2.3
1.501 0.325 1.470 0.540
4.0 2.9 4.0 3.1
1.313 1.478
1.1 to 0.95 0.90 to 0.78
0.013
0.01
0.791
0.42 to 0.73
Amoxicillin
2.4
7.4
Cephalexin
2.56
pKa1: basic amine group (-NH2)
Sulfadimethoxine Sulfamerazine Sulfamethazine Sulfamethoxazole Sulfathiazole Sulfadiazine Sulfapyridine
1.87 2.17 2.28 1.83 2.08 2.10 na
9.6 [6]
[1] [1] [1] [1] [1] [1] [2]
Macrolides
pKa1: basic dimethylamino group [-N(CH3)2]
Erythromycin Roxithromycin Clarithromycin Azithromycin
8.90 [3] 9.17 [3] 9.0 [8] 8.59 [5]
Tetracyclines
pKa1: acidic tricarbonyl group pKa2/pKa3: basic dymethylamino group acidic b-dikentone group (simultaneous dissociation)
Oxytetracycline Chlortetracycline Tetracycline Doxycycline
3.30 3.30 3.30 3.50
7.30 7.40 7.70 7.70
9.10 [3] 9.30 [3] 9.70 [3] 9.50 [3]
Fluoroquinolones
pKa1: acidic carboxylic group connected with ring 1 pKa2/pKa3/pKa4: are assigned in order to three basic nitrogen sites starting from ring 1 (nalidixic acid group) to ring 3 (fluoro group)
Ciprofloxacin Norfloxacin
3.01 3.11
6.14 6.10
8.70 8.60
Nitroimidazoles
pKa1 basic imidazole (heterocyclic aromatic ring with 3C and 2 N)
Metronidazole
2.50 [4]
Other antibiotics
pKa1 and pKa2 are assigned to basic N3 and N1 site in the pyrimidine ring
Trimethoprim
3.23
6.76 [3]
10.58 [3] 10.56 [3]
to 4.9 to 3.8 to 4.7 to 3.7
[1] (Maria and Reginald, 1993) [2] (Huber et al., 2005) [3] (Quiang and Adams, 2004) [4] (Loke et al., 2000). [5] (American Chemical Society, 2009) [6] (Delgado and Remers, 1991) [7] (Dutta et al., 1999) [8] (Gustavson et al., 1995).
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
0.61
pKa1: acidic carboxylic group pKa2: basic amine group (-NH2) pKa3: acidic hydroxide group (-OH) pKa1: acid carboxylic group pKa2: basic amine group (-NH2)
pKa2: acidic amide group (-NH-)
pKa3
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
media filtration (Gobel et al., 2005a). Previous studies have indicated that nitrification microorganisms appear to be capable of degrading trimethoprim (Batt et al., 2006; Perez et al., 2005). This suggests an important role for aerobic conditions for the biotransformation of trimethoprim. Consistent with this, removal efficiency of trimethoprim appears to be enhanced by long SRT during biological treatment, which is conducive to nitrification (Batt et al., 2007).
3.4.
Macrolides
Macrolide antibiotics, such as erythromycin, are active against most Gram-positive bacteria by binding reversibly to 50 S ribosomal subunits and inhibiting protein synthesis in microorganisms (Marzo and Dal Bo, 1998; Todar, 2002). After administration, macrolides are largely excreted into sewage in their unchanged forms at excretion rates greater than 60% (Hirsch et al., 1999). This implies that sewage entering WWTPs may contain high concentration of macrolides, particularly in countries in which macrolides are highly prescribed antibiotics. Gobel et al. (2005a,b) reported the concentration of macrolides in raw sewage from Switzerland to vary between 0.01 and 0.6 mg L1, while Karthikeyan and Meyer (2006) found that WWTP influent in the USA can contain macrolides at concentrations as high as 1.5 mg L1. Erythromycin is among the principal representatives of the macrolide antibiotics for clinical use (Kirst, 2002). An important difference between erythromycin and other macrolides, such as clarithromycin and roxythromycin, is the sensitivity of erythromycin to pH. Under acidic conditions, erythromycin is unstable and is transformed into an inactive anhydro-form by the loss of one H2O molecule (Gobel et al., 2004). At the ambient operational pH ranges (6.5e8) of most municipal WWTPs, erythromycin can exist in both its active original form and as the inactive erythromycin-H2O. Reported regional differences in sewage concentrations of macrolides may be a reflection of variable prescription and consumption patterns (Gobel et al., 2005a; Miao et al., 2004). In Switzerland, clarithromycin is more often detected in WWTPs effluents at higher concentrations than erythromycin-H2O and roxythromycin, which is well correlated to consumption data (Gobel et al., 2007). Conversely, in Canada erythromycinH2O is more frequently detected (and prescribed), followed by clarithromycin and roxythromycin (Miao et al., 2004). Macrolide antibiotics are often incompletely removed by conventional WWTPs (McArdell et al., 2003). Erythromycin (including erythromycin-H2O), has been removed between 43% and >99% by secondary wastewater treatment processes employing either AS or aerated lagoons (Karthikeyan and Meyer, 2006). Average removal of about 50% for macrolide antibiotics, such as clarithromycin and azithromycin, were reported from three conventional WWTPs in Japan (Kobayashi et al., 2006). Hirsch et al. (1999) reported that macrolides were found in all investigated WWTP discharges in Germany at concentrations in excess of 100 ng L1. Studies using 24-h composite samples have revealed that the removal of macrolides by conventional AS treatment varied from 80% to 44% (Clara et al., 2005b; Gobel et al., 2007). It has been suggested that negative removals of macrolides are likely to be due to the release of these compounds from excreted bile and faeces
4307
during the biological treatment rather than the presence of deconjugable metabolites (Gobel et al., 2007). Sampling variation may also have contributed to this negative removal as reported by Clara et al. (2005b), where the collection of effluent samples was not time-adjusted to account for long HRTs. Sorption of macrolides to wastewater biomass is mainly attributed to hydrophobic interactions (Gobel et al., 2005a). This is expected due to their high Log DpH 6e8 partitioning coefficients (Table 5). Macrolides may also adsorb to biomass via cation exchange processes due to the fact that under typical wastewater conditions, many are positively charged through the protonation of the basic dimethylamino group (pKa > 8.9) and the surface of activated sludge is predominantly negatively charged (Carberry and Englande, 1983). However, in general, sorption to sludge accounts for only minor components of most macrolide drugs in WWTPs (Gobel et al., 2005a). Greater adsorption of azithromycin to biomass compared to clarithromycin has been reported (Kobayashi et al., 2006).
3.5.
Fluoroquinolones
Fluoroquinolones are antibiotics effective against several types of Gram-negative and Gram-positive bacteria (Turiel et al., 2003). These antibiotics act by inhibiting essential enzyme function for DNA production (Marzo and Dal Bo, 1998). The occurrence of fluoroquinolones in WWTP effluents has been reported in Australia, Canada, China, Italy, Mexico, Sweden, and the USA (Brown et al., 2006; Costanzo et al., 2005; Karthikeyan and Meyer, 2006; Lindberg et al., 2006; Miao et al., 2004; Zorita et al., 2009; Zuccato et al., 2005). When screening 12 human antibiotics in five WWTPs in Sweden, Lindberg et al. (2005) reported fluoroquinolones to be the most frequently detected antibiotics above analytical quantitation limits. In that study, norfloxacin and ciprofloxacin were detected in 97% and ofloxacin in 50% of the analysed samples. Fluoroquinolones appeared not be readily biodegradable in controlled batch tests (Kummerer et al., 2000). Removal efficiencies of fluoroquinolones from the aqueous phase during wastewater treatment in Sweden were reported to be 80% for norfloxacin and 78% for ciprofloxacin. Grit removal/ferrous precipitation achieved approximately 55%e58%, while AS treatment removed about 34% and 44% of norfloxacin and ciprofloxacin, respectively (Lindberg et al., 2006). A later study reported the removal of ciprofloxacin (90%), oxfloxacin (56%), and norfloxacin (70%) during AS treatment followed by chemical coagulation/flocculation (Zorita et al., 2009). The predominant removal mechanism of fluoroquinolones has been suggested by several authors to be adsorption to sludge and/or flocs rather than biodegradation (Batt et al., 2007; Golet et al., 2003; Lindberg et al., 2006; Zorita et al., 2009). A mass balance study revealed that a conventional wastewater treatment process resulted in the removal of 88e92% of fluoronoquinolones from the aqueous phase, due to the adsorption to sludge (Golet et al., 2003). They also observed that no significant removal of the compounds occurred under methanogenic conditions of the anaerobic sludge digester (i.e. biodegradation) and 75e83% of the compounds’ input mass remained in the digested sludge. Lindberg et al. (2006) reported that more than 70% of norfloxacin and ciprofloxacin passed through the treatment plant and remained in digested sludge. These findings
4308
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
implied sludge is the main reservoir of fluoroquinolones, potentially releasing the antibiotics into the environment via biosolids application to agricultural land. Sorption behaviour of fluoroquinolones is somewhat pH-dependant (Belden et al., 2007; Cardoza et al., 2005). However, the sorption of these compounds to sludge is not significantly affected by the narrow range of pH variability normally observed for WWTPs (Lindberg et al., 2006).
3.6.
Tetracyclines
The tetracyclines consist of eight related broad spectrum antibiotics, which are bacteriostatic and are active against Gram-positive and Gram-negative bacteria (Todar, 2002). Tetracyclines inhibit protein synthesis in the microorganisms by binding to the 30 S ribosome and preventing the access of aminoacyl tRNA to the acceptor site on the mRNA-ribosome complex (Marzo and Dal Bo, 1998). Tetracycline is one of the most frequently detected antibiotics in wastewater (Kim et al., 2007). Tetracyclines were reported in the raw WWTP influent in the USA at concentrations between 0.1 and 0.6 mg L1 (Kim et al., 2005). In Canada, the remaining concentration of tetracycline in WWTP effluent was reported to be nearly 1.0 mg L1 (Boussu et al., 2007; Miao et al., 2004). Removal efficiency of >68% has been reported for tetracycline during conventional secondary treatment (Karthikeyan and Meyer, 2006). In the USA, chlorotetracycline and doxycycline have been reported after secondary treatment and chlorination with removal efficiencies of 78% and 67%, respectively (Yang et al., 2005). Tetracyclines removal appears not to be significantly affected by changes in HRT of AS process (Kim et al., 2005). However, these authors reported a significant reduction in removal efficiency with decreased SRT from 10 days to 3 days, indicating that the change in the nature of biomass may affect the removal via solids adsorption. Controlled sorption tests have suggested that some tetracyclines have significant potential for adsorption onto biomass (Kim et al., 2005), whereas oxytetracycline is quite soluble in aqueous solutions and poorly adsorbed to biomass (Rabolle and Spliid, 2000). Despite low logDpH 6e8 values for these chemicals, non-hydrophobic mechanisms, such as ionic interactions, metal complexation, hydrogen bond formation or polarisation, likely play a significant role in the sorption of tetracyclines to many solids (Tolls, 2001). Since pH and temperature have been reported to have an effect on hydrolysis rates of tetracyclines (Loftin et al., 2008), it is possible that this mechanism may further contribute to the degradation of these chemicals in wastewater, particularly in tropical regions where temperatures are commonly above 35 C.
3.7.
Nitroimidazoles
Nitroimidazoles, such as metronidazole, are microbicidal drugs that are active against most anaerobic bacterial species (Theron et al., 2004) and a range of pathogenic anaerobic protozoa causing infections such as Giardiasis (Schneider, 1961), amoebiasis (Powell et al., 1966), and trichomoniasis (Cosar and Julou, 1959; Upcroft et al., 1999). There are only a few studies that have examined the occurrence of nitroimidazoles in sewage or WWTP effluents
since these drugs are dispensed in relatively small quantities as compared to other antibiotics. Thus, significant concentrations of nitroimidazoles are not expected to occur in sewage (Khan and Ongerth, 2004). However, Lindberg et al. (2005) observed that the concentrations of metronidazole in hospital sewage can be as high as 90.2 mg L1, implying that higher concentrations may be found in sewage catchment receiving large contributions from hospitals. Metronidazole appears not to be readily biodegradable in laboratory-based batch experiments (Alexy et al., 2004; Ingerslev and Halling-Sørensen, 2000; Kummerer et al., 2000). Furthermore, it is relatively hydrophilic (Loke et al., 2000). As a result of these factors, metronidazole is not expected to be effectively removed during conventional wastewater treatment (Carballa et al., 2004; Khan and Ongerth, 2004).
3.8.
Other antibiotic groups
Aminoglycosides and ionophores are other antibiotic classes of interest. Aminoglycoside antibiotics are widely used in hospitals for treatment of serious human infection by gram-negative and gram-positive bacteria (Lo¨ffler and Ternes, 2003) and in veterinary medicine (Salisbury, 1995). Their antimicrobial action is by the inhibition of microorganism protein synthesis (Marzo and Dal Bo, 1998). Aminoglycosides are mostly nonmetabolised after being administered; hence they will be excreted via urine unchanged (Marzo and Dal Bo, 1998). The analysis of wastewater from a hospital in Germany revealed that the concentration of aminoglycoside antibiotic gentamicin was between 0.4 and 7.6 mg L1 (Lo¨ffler and Ternes, 2003). There is little other information available on the occurrence and fate of aminoglycosides in wastewater and through treatment processes. However, due to their high sorption properties, it has been suggested that aminoglycoside antibiotics in wastewater would be adsorbed onto solid particles and colloidal organic matter and significantly removed from aqueous phase by filtration (Lo¨ffler and Ternes, 2003). Ionophore antibiotics are used in agricultural applications as feed additives to treat or prevent infections in poultry and livestock and as growth promoters for ruminants (Cha et al., 2005; Khan et al., 2008; Schlu¨sener et al., 2003). The pharmacological activity of ionophore antibiotics results from their ability to readily form electrically neutral psuedomacrocyclic complexes with polar mono and divalent cations in solution and transport the cations across the cell membrane (Cha et al., 2005). Several studies reported the occurrence of ionophore antibiotics such as monensin, salinomycin and narasin at concentrations up to 40 ng L1 in surface water near the livestock feeding operations or agricultural lands (Cha et al., 2005; Kim and Carlson, 2006). A study by Watkinson et al. (2009) showed that the detection frequency and concentrations of monensin and salinomycin in wastewater were much lower than those in environmental waters. The behaviour of ionophore antibiotics through WWTP processes is little known due to the less likely occurrence of these antibiotics in domestic wastewater except where there is runoff from agricultural lands into sewers. A study by Donoho (1984) indicated that monensin is biodegradable in manure and soil where the primary degradation occurred in 33 days under aerobic
4309
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
condition, but much longer under anaerobic condition (60e70% after 10 weeks).
be more effective circumstances.
3.9. Effects of antibiotics on wastewater microbial consortia/processes
4.1.
While concentrations of antibiotics reported in domestic wastewater (in ng L1 and mg L1 ranges) are not considered sufficiently high to cause noticeable effects on wastewater treatment processes, a few studies have reported observed inhibition of wastewater microbial activities at elevated antibiotics concentrations (Amin et al., 2006; Gartiser et al., 2007; Ingerslev and Halling-Sørensen, 2000). For sulfonamide antibiotics, concentrations of 10e400 mg L1 were reported to inhibit microbial activities in activated sludge by more than 20% (Ingerslev and Halling-Sørensen, 2000). A study by Amin et al. (2006) showed that the presence of erythromycin at concentration of 1 mg L1 reduced COD removal efficiency and biogas production in anaerobic treatment by about 5%. A study of antibiotic biodegradability using the ISO closed bottle test revealed that a metronidazole concentration of 6 mg L1 could reduce anaerobic activity by 69% (Gartiser et al., 2007). This is perhaps not surprising since an important pharmacological property of metronidazole is to target anaerobic microorganisms (Leiros et al., 2004). The presence of antibiotics in wastewater has been suspected to contribute to the development and dissemination of antibiotic resistant species since a significant number of antibiotic resistant genes have been found in WWTPs (Szczepanowski et al., 2009). Several bacteria such as fecal coliforms, E.coli and enterococci found in wastewater influent and effluent have exhibited resistance to ciprofloxacin, trimethoprim, sulfamethoxazole and vancomycin (Nagulapally et al., 2009). A study using a sequencing batch reactor by Kim et al. (2007) showed that the exposure to tetracycline at concentration of 1 mg L1 increased concentrations and production rates of tetracycline resistant bacteria in the reactor. Nonetheless, the actual significance of low concentrations of antibiotics in WWTPs, in term of resistance propagation, is yet to be conclusively determined (Jury et al., in press).
4. Fate of antibiotics during advanced treatment processes Conventional secondary wastewater treatment processes appear to be highly variable in their ability to remove most antibiotics, with performance apparently dependent upon specific operational conditions, such as SRT. Accordingly, tertiary and advanced treatment processes may be necessary to provide further reduction of these compounds, in order to minimise environmental and human exposure. Semi-qualitative estimations of antibiotic removal by tertiary media filtration, ozonation, chlorination, UV irradiation, activated carbon adsorption, and NF/RO filtration as reported in the literature are presented in Table 6. Generally, it can be observed that the removals of most antibiotics by tertiary sand filtration and UV disinfection are poor while ozonation, chlorination, activated carbon and NF/RO filtration appear to
when
operated
under
optimum
Membrane filtration
Rejection of chemical contaminants by high-pressure membranes, such as nanofiltration (NF) and reverse osmosis (RO), is ultimately determined by complex interactions of electrostatic and other physical forces acting between a specific solute (chemical contaminant), the solution (water and other solutes present), and the membrane itself (Bellona et al., 2004; Nghiem et al., 2005). The key rejection mechanisms for these solutes are steric hindrance (size exclusion), electrostatic interactions (attractive or repulsive), and hydrophobic interaction (diffusion and partitioning) between compounds and the membrane. The nature of these forces is dependent on numerous chemical/physical properties of the solute (molecular size, pKa, polarity or hydrophobicity), solution (pH, ionic strength), and the membrane (materials, pore size). A useful guide for the classification of organic contaminants for estimations of their rejection by high-pressure membrane processes has been proposed by Bellona et al. (2004) (see Fig. 1). This diagram was derived as the result of a comprehensive review of published studies reporting rejection behaviour of a wide range of organic solutes by various commercially available membranes. Based on this rejection diagram, qualitative predictions can be derived to provide an estimation of treatment efficiencies by NF or RO for specific antibiotics as shown in Table 7. This table was developed with the assumptions of a highly negatively charged RO membrane with a nominal molecular weight cutoff (MWCO) of 100 Da at pH 7. Effective molecular width (MWd) of each compound was estimated by MMPplus software (ChemSW, 2005) based on the energy-minimised chemical structure generated by ChemBio3D Ultra (CambridgeSoft, 2007). Molecular charge states were determined according to the pKa values provided in Table 5. From the qualitative predictions (Table 7), RO treatment appears to be a promising process for the effective removal of most antibiotics. The predicted behaviours are reasonably well matched with quantitative data obtained from previous studies (Adams et al., 2002; Baumgarten et al., 2007; Dolar et al., 2009; Kosutic et al., 2007). These studies showed that >99% rejection is often achieved by RO and some NF membranes for several antibiotics including fluoroquinolones, sulfonamides, tetracyclines, and trimethoprim. A study undertaken by Li et al. (2004) on the treatment of an initially very high concentration of oxytetracycline in wastewater from pharmaceutical manufacturing industry found that RO treatment effectively reduced oxytetracycline concentration from 1000 mg L1 to below 80 mg L1 (>92% removal). Despite the usefulness of the membrane rejection diagram, there are some limitations in its applicability for predicting chemical behaviour in real full-scale treatment systems. MWCO is the manufacturer’s rating of the membrane ability to reject an uncharged standard compound (such as dextran or a protein) based on the molecular weight, while other organic compounds with different molecular shapes and sizes may partially permeate the membrane although their molecular weights exceeds the MWCO (Porter,
4310
Table 6 e Semi-quantitative estimations of the removal of antibiotics by tertiary and advanced treatment processes. Treatment Processes Group
Tertiary Sand Filtration Removal
Ref.
Ozonation Removal
Conditions
Chlorination Ref.
Free chlorine conc. 1.0e1.2 mg/L Removal
na
VG-E
O3 (3e5 mg/L), DOC (5.3 mg/L) pH (7.7), secondary effluent
(Dodd et al., 2006)
na
Sulfonamides
VP
(Batt et al., 2007; Gobel et al., 2007)
E
O3 (2.0e7.1 mg/L) HRT (1.5e18.0 min); pH (7e8); DOC (6.6e23 mg/L); secondary effluent and river water
(Adams et al., 2002; Huber et al., 2005; Ternes et al., 2003)
F-E
Free Chlorine (1e1.2 mg/L); pH (8); HRT (up to 1 day); Drinking water; River water
Macrolides
VP-P
(Gobel et al., 2007)
E
O3 (2e5 mg/L); HRT (4e18 min); pH (7e8); DOC (6.6e23 mg/L), secondary effluent and river water
(Huber et al., 2005; Radjenovic et al., 2009; Ternes et al., 2003)
VP
Tetracyclines
VP
(Batt et al., 2007)
E
O3 (3 mg/L), DOC (5.3 mg/L(pH (7.7), secondary effluent
(Dodd et al., 2006)
Fluoroquinolones
VP-P
(Batt et al., 2007; Golet et al., 2003)
E
O3 (3 mg/L); DOC (5.3 mg/L), pH (7.7), secondary effluent
(Dodd et al., 2006)
Nitroimidazoles Trimethoprim
VP-G*
(Batt et al., 2007)
na E
O3 (5e7.1 mg/L); HRT (Adams et al., 2002; (1.5e18 min); pH Dodd et al., 2006; (7e8); DOC Ternes et al., 2003) (7.7e23 mg/L)
Ref.
Removal
Conditions
Ref.
na
(Chamberlain and Adams, 2006; Gibs et al., 2007)
E
Free chlorine (3.5e3.8 mg/L), pH (7e8); DOC (3.0e3.5 mg/L); CaCO3 (80e307 mg/L): river water
(Westerhoff et al., 2005)
Free Chlorine (1.2 mg/ (Gibs et al., L); pH (8); HRT (1 day); 2007) CaCO3 (63 mg/L) Drinking water
E
Free chlorine (3.5e3.8 mg/L), pH (7e8); DOC (3.0e3.5 mg/L); CaCO3 (80e307 mg/L): river water
(Westerhoff et al., 2005)
E
Free Chlorine (1.2 mg/ Gibs et al., L); pH (8); HRT (1 day); 2007 CaCO3 (63 mg/L) Drinking water
na
P
Free Chlorine (1.2 mg/ Gibs et al., L); pH (8); HRT (1 day); 2007 CaCO3 (63 mg/L) Drinking water
na
Free chlorine (3.5e3.8 mg/L), pH (7e8); DOC (3.0e3.5 mg/L); CaCO3 (80e307 mg/L): river water
(Westerhoff et al., 2005)
na VG
Free Chlorine (1 mg/L); (Adams et al., HRT (>40 min); DOC 2002) (10.3 mg/L); River water
na E
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
Beta-lactam
Conditions
Free chlorine conc. >3 mg/L
Treatment Processes Group
UV irradiation Typical disinfection dose Removal
Ref.
Activated Carbon Adsoprtion
RO/NF Membrane Filtration
Higher Dose (20e100 times) Removal
Conditions
Ref.
VP
(Batt et al., 2007)
na
Sulfonamides
VP-P
(Drewes et al., 2008; F-VG Le-Minh et al., 2010)
UV dose (2760e3000 (Adams et al., 2002; mJ/cm2); HRT Kim et al., 2009) (5e30 min); DOC (3.5e10.7 mg/L)
Macrolides
VP
(Drewes et al., 2008) P
UV dose (2760 mJ/ cm2); HRT (5 min); DOC (3.5 mg/L)
(Kim et al., 2009)
Tetracyclines
na
VG-E
UV dose (2760 mJ/ cm2); HRT (5 min); DOC (3.5 mg/L)
Fluoroquinolones
VP
(Vieno et al., 2007b)
VG-E
UV dose (2760 mJ/ cm2); HRT (5 min); DOC (3.5 mg/L)
Nitroimidazoles
VP
(Shemer et al., 2006)
Trimethoprim
VP
(Drewes et al., 2008; F Le-Minh et al., 2010)
Conditions
Ref.
F-VG
pH 6e7, GAC dose (20 mg/L), Co (10ug/ L), based on Freundlich isotherm, equlibrium state
(Aksu and Tunc, 2005; Dutta et al., 1999; Putra et al., 2009)
G-VG
Removal
Ref.
E
(Morse and Jackson, 2004)
pH 7.7e7.9, PAC dose (Adams et al., 2002; (20 mg/L), Co (up to Westerhoff et al., 50ug/L), DOC 2005) (3.5e10.7 mg/L), 4 h contact time; river water
G-E
(Dolar et al., 2009; Kimura et al., 2004)
VG
pH 7.9, PAC dose (20 mg/L), Co (50e100 ng/L), DOC (3.5 mg/L), 4 h contact time, river water
E
(Dolar et al., 2009)
(Kim et al., 2009)
E
pH 5.8, PAC (20 mg/ (Ji et al., 2009) L), Co (10 ug/L), based on Freundlich isotherm, equilibrium state
E
(Kosutic et al., 2007)
(Kim et al., 2009)
G-VG
PAC (50 mg/L); Co (25ug/L); 15 min contact time; MBR permeate
E
(Baumgarten et al., 2007; Dolar et al., 2009)
E
(Rivera-Utrilla et al., na pH 6e7, PAC dose (20 mg/L), Co (10ug/ 2009) L), based on Freundlich isotherm, equlibrium state
G-VG
pH 7.7e7.9, PAC dose (Adams et al., 2002; (20 mg/L), Co (up to Westerhoff et al., 50 mg/L), 2005) experiments in riverwater, DOC (3.5e10.7 mg/L), 4 h contact time
UV dose (2760e3000 (Adams et al., 2002; mJ/cm2); HRT Kim et al., 2009) (5e30 min); DOC (3.5e10.7 mg/L)
(Baumgarten et al., 2007)
VG-E
(Dolar et al., 2009; Kosutic et al., 2007)
4311
VP: very poor (<20%); P: Poor (20e45%); F: Fair (45e65%); G: Good (65e80%); VG: Very Good (80e95%); E: Excellent (>95%); na: not available.
(Westerhoff et al., 2005)
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
Beta-lactam
Removal
4312
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
Fig. 1 e Rejection diagram for organic micropollutants during membrane treatment (Bellona et al., 2004). MW [ molecular weight, pKa [ acid.
1990). Studies have shown that the rejection of antibiotics is better correlated to membrane pore size than to MWCO (Dolar et al., 2009; Kimura et al., 2004). Therefore, when size exclusion is the main mechanism for rejecting compounds, the comparison of effective MWd and membrane pore size distribution should be considered in order to optimise rejection estimations. Most investigations of the relationships between physicalechemical properties of solutes and membrane interactions have been conducted using unfouled ‘virgin’ membranes and thus their conclusions are unlikely to be quantitatively extendable to full-scale systems subjected to long-term operation (Agenson et al., 2003; Nghiem et al., 2004, 2005; Schafer et al., 2003; Simon et al., 2009). During normal operation, membranes are prone to fouling by the build-up of precipitated chemicals or by the growth of microbial biomass (Nghiem et al., 2006; Nghiem and Schafer, 2006; Oschmann et al., 2005). Fouling can lead to significant changes in physicochemical properties of the membrane surface, affecting the separation mechanisms including size exclusion and electrostatic interactions and thus in the way in which the membranes interact with water and solutes (Nghiem et al., 2009; Nghiem and Hawkes, 2007). In many cases, fouling is regarded as a hindrance since it decreases membrane permeability and thus requires elevated pressures to maintain operational flux. However, some investigations reveal that fouling can also lead to improved rejection of many solutes (Drewes et al., 2006; Schafer et al., 1994; Xu et al., 2006). This observation is believed to be due to increased negative surface charge leading to increased electrostatic rejection of ionic species; along with simultaneously increased adsorptive capacity for non-ionic solutes (Xu et al., 2006). Furthermore,
membrane degradation due to exposure to residual chlorine can impact rejection of some antibiotics (Simon et al., 2009). A further limitation of the membrane rejection diagram is that it does not account for the role of electrostatic attraction (as opposed to repulsion). The surfaces of most NF/RO membranes (made of polyamide or cellulose acetate) are negatively charged under neutral conditions due to the presence of ionisable sulphonic and carboxylic functional groups (Nghiem et al., 2005; Schafer et al., 2003; Simon et al., 2009). Negatively charged solutes are rejected mainly via electrostatic repulsion; while the positively charged species are removed by combination of attractive electrostatic interaction with the membrane surface and Donnan equilibrium (Schaep et al., 2001; Verliefde et al., 2008).
4.2.
Adsorptive treatment
4.2.1.
Activated carbon
Adsorptive treatment with activated carbon can be used for removing many hydrophobic pharmaceuticals from water (Snyder et al., 2003). The removal effectiveness of the activated carbon adsorptive treatment system depends on the properties of the adsorbent (specific surface area, porosity, surface polarity, and physical shape of the material) and the characteristics of the compound (shape, size, charge and hydrophobicity). Adsorption mechanisms consist of the chemical (electrostatic interaction) and physical bindings of molecules to the surface of an adsorbent. The latter is often more important due to the capability to form multi-layer bindings (Snyder et al., 2003). In fact, it was recently reported that the greatest removal of amoxicillin by activated carbon
Table 7 e Qualitative rejection prediction based on LogKow*, MW, and charge state for antibiotics by reverse osmosis assuming MWCO [ 100, pH [ 7, negatively charge reverse. Name
MWd (nm)
Charge State at pH 7
Dominant rejection mechanisms
Amoxicillin Cephalexin Sulfadimethoxine Sulfamerazine Sulfamethazine Sulfamethoxazole Sulfathiazole Sulfadiazine Sulfapyridine Ciprofloxacin Norfloxacin Erythromycin
365 347 310 264 278 253 255 250 249 331 319 734
1.32 1.39 1.59 1.47 1.47 1.43 1.31 1.36 1.33 1.23 1.18 1.59
72% zwitterionic; 28% negative 57% zwitterionic; 43% negative 7% neutral; 93% negative 37% neutral; 63% negative; 72% neutral; 28% negative 4% neutral; 96% negative 54% neutral; 45% negative 16% neutral; 84% negative 96% neutral; 4% negative zwitterionic and positive zwitterionic and positive 99% positive
Roxithromycin
837
2.18
99% positive
Clarithromycin
748
1.61
99% positive
Azithromycin
749
1.18
97% positive
Oxytetracycline Chlortetracycline Doxycycline Tetracycline Metronidazole Trimethoprim
460 479 444 444 171 290
1.31 1.56 1.50 1.62 1.01 1.42
Zwitterionic and negative Zwitterionic and negative Zwitterionic and negative Zwitterionic and negative 100% neutral 63% neutral; 37% positive
dominantly electrostatic repusion and size exclusion dominantly electrostatic repusion and size exclusion dominantly electrostatic repulsion and less from size exclusion mainly electrostatic repuslion, and less from size exclusion mainly steric hindrance; and less from electrostatic repusion dominantly electrostatic repusion setric hindrance and electrostatic repulsion dominantly electrostatic repusion; and steric hindrance dominantly steric hindrance steric hindrance, electrostatic attraction steric hindrance, electrostatic attraction electrostatic attraction, mainly steric hindrance and hydrophobic interaction electrostatic attraction, mainly steric hindrance and hydrophobic interaction electrostatic attraction, mainly steric hindrance and hydrophobic interaction electrostatic attraction, mainly steric hindrance and hydrophobic interaction dominantly electrostatic repusion dominantly electrostatic repusion dominantly electrostatic repusion Dominantly electrostatic repusion Mainly steric hindrance Steric hindrance, electrostatic attraction
Qualitative rejection prediction from rejection diagram very high very high very high very high moderate very high moderate very high moderate very high very high moderate to high, but depends on partitioning and diffusion moderate to high depends on partitioning and diffusion moderate to high, but depends on partitioning and diffusion moderate to high, but depends on partitioning and diffusion very high very high very high very high moderate moderate
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
MW
*LogKow values from Table 5. MW: Molecular weight; MWd: Molecular width; MWCO: Membrane molecular weight cut-off.
4313
4314
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
was achieved under pH conditions corresponding to a zero net charge on the activated carbon surface (Putra et al., 2009). It has been known for many years that adsorption to activated carbon can be used to purify antibiotics for commercial production (Bansal and Goyal, 2005) and for the treatment of poisoning and overdose of drugs (Cooney, 1995). In water and wastewater treatment, several studies on the removal of antibiotics by activated carbon have been reported (Adams et al., 2002; Choi et al., 2008; Putra et al., 2009; Rivera-Utrilla et al., 2009). With dosages between 10 mg L1 and 20 mg L1 powdered activated carbon (PAC), the concentrations of several antibiotics in river water have been reduced by 49%e 99% after 4 h contact time (Adams et al., 2002; Westerhoff et al., 2005). After 1 day contact time, similar removals of sulfonamide and tetracycline antibiotics from river water have been achieved with the lower dosage of 1 mg L1 PAC (Choi et al., 2008). The sorption efficiencies of antibiotics to activated carbon may be significantly altered by several factors, such as the types of activated carbon used, the initial concentrations of target compounds and the pH, temperature and dissolved organic carbon (DOC) concentration of the solution (Aksu and Tunc, 2005; Choi et al., 2008; Dutta et al., 1999; Ji et al., 2009; Rivera-Utrilla et al., 2009). A number of previous studies have reported the prediction of activated carbon adsorption capacity for antibiotics including amoxicillin, penicillin, tetracycline, and nitromidazole using Freundlich or Langmuir isotherms (Aksu and Tunc, 2005; Dutta et al., 1999; Ji et al., 2009; Putra et al., 2009; RiveraUtrilla et al., 2009). The capacity of activated carbon to adsorb a particular compound can, to some extent, be predicted based on the ‘hydrophilic’ or ‘hydrophobic’ nature of the chemical (Snyder et al., 2003). The hydrophobic (non-polar) or hydrophilic (polar) properties of antibiotics can be determined from their LogD (or pKa-adjusted Log Kow) values (Table 5). It has been reported that non-polar antibiotics with Log Kow > 2, may be effectively removed with activated carbon by hydrophobic interaction (Snyder et al., 2003). However, the adsorption of more polar or charged compounds to activated carbon is much more difficult to predict due to the additional effects of polar interactions and ion exchange (Snyder et al., 2003). A study undertaken by Westerhoff et al. (2005) examined the relationship between activated carbon capacity and the Log Kow of 62 different micropollutants including some antibiotics such as sulfamethoxazole, erythromycin-H2O and trimethoprim. The study showed that the degree of removal of non-volatile neutral compounds after contact by PAC (5 mg L1, 4 h contact time) can be reasonably well predicted by the following relationship: [percentage removal] ¼ 15 [Log Kow] þ 27%. However, the removal of protonated and deprotonated compounds did not follow this trend (Westerhoff et al., 2005). Accordingly, caution must be exercised in the prediction of activated carbon removal efficiency by Log Kow values and the effect of solution pH as well as pKa values of the compounds must be carefully considered in order to derive LogDpH instead. LogD calculation takes into account both ionisation constant (pKa) and log Kow of various species formed in solution at different pH. Therefore, the use of LogD is more appropriate to estimate the partitioning properties of all compounds including ionisable species than log Kow (Bhal et al., 2007).
4.2.2.
Ionic adsorption
Many antibiotics, including tetracyclines and sulfonamides are often present in negatively charged form at normal operating pH conditions (Quiang and Adams, 2004). Therefore, the use of ionic treatment processes may be effective for the removal of these anionic micropollutants (Robberson et al., 2006). A study by Choi et al. (2007b) revealed that the anionic MIEX resin is effective for the removal of fourteen antibiotics from the sulfonamide and tetracycline classes, which are in zwitterionic and anionic forms at pH 7. Accordingly, significant removal of 10 mg L1 sulfonamides and tetracyclines spiked in 0.8 mg L1 DOC water was achieved with the addition of 3.6e13 mL L1 MIEX. Ion exchange is the main mechanism in the ionic treatment for negatively charged antibiotics though the removal of antibiotics by agglomeration could also occur in the presence of metal oxides and natural organic matter (Boyer and Singer, 2005; Choi et al., 2007b). To the knowledge of the authors, no full-scale treatment of ionic adsorption for antibiotics has been reported in the scientific literature. One explanation for this apparent lack of interest may be a lack of cost-effectiveness since the presence of other organic contaminants in the aqueous stream can compete with targeted antibiotics for the ion exchange sites and reduce the antibiotic removal efficiency. Regardless of the efficiency, ion exchange based processes are not targeting neutral compounds and additional processes would be required to provide a more comprehensive removal of the full spectrum of antibiotics. Full-scale studies are required to determine the optimal configuration and operating conditions of adsorptive systems, which are effective and economically feasible for antibiotics removal.
4.3. Chemical and photochemical oxidation processes for the removal of antibiotics Oxidative processes may be used to transform any organic constituents of wastewaters, which appear to be both biologically recalcitrant and poorly rejected by membranes or activated carbon. Strong chemical oxidants, such as ozone (von Gunten, 2003), potassium permanganate (Adam et al., 2004; Chen et al., 2005) and chlorine (Chamberlain and Adams, 2006), have been shown to be effective for the transformation of various chemical contaminants in water.
4.3.1.
Chlorination
Chlorination can inactivate active chemical compounds via one of two general mechanisms. One possibility is by chorine substitution or addition reactions, which may alter active functional groups (Crain and Gottlieb, 1935). Alternatively, chlorine radicals may oxidise (break down) the target compound such as antibiotic drugs into smaller molecules, which may or may not possess the active properties (Crain and Gottlieb, 1935). The effective removal of antibiotics by chlorination from drinking water requires sufficient free chlorine concentration and contact time. With the use of free chlorine at 1.0 mg L1 (as Cl2), 90% removal has been reported with contact times greater than 16 min for most sulfonamides and greater than 40 min for trimethoprim in river water (Adams et al., 2002). A
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
study by Gibs et al. (2007) on the persistence of 98 pharmaceuticals, including 23 antibiotics of sulfonamide, tetracycline, macrolide and quinolone classes in chlorinated drinking water also revealed that the presence of free chlorine in drinking water is an effective means for the transformation of some pharmaceutical compounds during distribution. With a free chlorine concentration of 1.2 mg L1 and an initial spiked concentration of 0.5 mg L1 for each antibiotic in drinking water, reductions of >99% for tetracyclines, 50%e80% for sulfonamides, 42% for trimethoprim, 30%e40% for fluoroquinolones, and less than 10% for macrolides, respectively, were observed after a single day’s contact time, and complete removals were achieved after 10 days (Gibs et al., 2007). At the higher free chlorine concentration of 3.5e3.8 mg L1, removals of 90% to >99% were achieved for sulfamethoxazole, trimethoprim and erythromycin in river water after 24 h contact time (Westerhoff et al., 2005). Accordingly, although some antibiotics may be more resistant to chlorination than others, they appear to gradually degrade over time in the presence of free chlorine. In addition, the optimum dosage and contact time may increase with increased concentration of solids and organic matter in the water. Reactivity of sulfonamides with HOCl has been observed to be in the following descending order: sulfadiamethoxine > sulfathiazole > sulfamethazine > sulfamerazine > sulfamethoxazole > sulfamethizole (Chamberlain and Adams, 2006). Alkaline pH (>8) inhibits the removal of sulfonamides by chlorine oxidation (Chamberlain and Adams, 2006; Gibs et al., 2007). Removals of sulfonamides and macrolides using chloramination were minimal under typical drinking water conditions (Chamberlain and Adams, 2006). Huber et al. (2005) observed fast reactions between ClO2 oxidation and antibiotics, such as sulfonamides and macrolides, in natural water. Roxithromycin and sulfamethoxazole exhibited strong pH dependence in the oxidation with ClO2 (higher reactivity at pH > 7). Similar to HClO, ClO2 oxidises roxithromycin and sulfamethoxazole at specific functional groups with high electron densities, such as neutral tertiary amines and aniline. Based on the laboratory studies, rapid and substantial transformation of trimethoprim to a wide range of chlorinated and hydroxylated products is expected to occur under typical conditions of wastewater and drinking water chlorination (Dodd and Huang, 2007). Although a number of studies have demonstrated the ability of chlorination to reduce antibiotics concentrations in drinking water, it is still too early to confirm the relative importance of chlorine and its derivative products in degrading antibiotics in water and wastewater treatment. Indeed, the major concern for treating pharmaceuticals via chlorination is the formation of chlorinated byproducts since these may be more harmful than their parent compounds (von Gunten et al., 2006). In past decades, the use of chlorine for water treatment has attracted concern since the reaction of chlorine with natural organic matter is well known to produce harmful chlorinated byproducts. Data and research regarding the ultimate fate of these pharmaceuticals treated by chlorination processes and whether they are degraded to harmless metabolites or transformed into potentially more toxic contaminants is needed (Glassmeyer and Shoemaker, 2005).
4.3.2.
4315
Ozonation
Several previous studies have reported the effective treatment of ozonation for removal of antibiotics in water and wastewater effluents (Adams et al., 2002; Huber et al., 2005; Ternes et al., 2003). The study by Adams et al. (2002) showed that ozonation removed more than 95% of several sulfonamides and trimethoprim from river water within 1.3 min contact time at an ozone dose of 7.1 mg L1 Huber et al. (2005) also observed that using ozonation at doses >2 mg L1 oxidised 90% e >99% of sulfonamides and macrolides in secondary wastewater effluents. Second order rate constants for 14 fluoroquinolones, sulfonamides, b-lactams, macrolides and trimethoprim by ozone and hydroxyl radicals have been reported (Dodd et al., 2009). As the result of this study, the authors suggested that 99% removal of antibiotics from river water and wastewater effluents should be achievable at typical ozone doses used for disinfection (i.e. 5e10 mg L1 ozone with DOC of 5e23 mg L1). During ozone treatment, oxidative degradation of organic chemicals can occur either by direct reaction with molecular ozone (O3) or indirectly via hydroxyl radicals (Staehelin and Hoigne, 1985). During wastewater ozonation, many antibiotics, including sulfonamides, macrolides, fluoroquinolones and tetracyclines, have been shown to be predominantly transformed via direct reaction with ozone (Dodd et al., 2006) whereas cephalexin, penicillin, and N4-acetyl sulfamethoxazole were transformed to a large extent by hydroxyl radicals (Dodd et al., 2006). The relative dominance of the actual oxidative pathway will depend on the ratio of molecular ozone and hydroxyl radicals, the corresponding reaction kinetics, and presence of organic matter (Elovitz et al., 2000, von Gunten, 2003). Ozone and/or hydroxyl radicals deactivate bactericidal properties of antibiotics by attacking or modulating their pharmaceutically active functional groups, such as N-etheroxime and dimethylamino groups of macrolides (Dodd et al., 2009; Lange et al., 2006), aniline moieties of sulfonamides (Huber et al., 2005), thioether groups of penicillins, unsaturated bonds of cephalosporin and the phenol ring of trimethoprim (Dodd et al., 2009). The good removal (>90%) by ozonation was observed for those compounds with electron-rich aromatic systems, such as hydroxyl, amino (e.g. sulfamethoxazoles), acylamino, alkoxy and alkyl aromatic compounds, as well as those compounds with deprotonated amine (e.g. erythromycin, ofloxacin and trimethoprim) and nonaromatic alkene groups since these key structural moieties are highly amendable to oxidative attack (Dickenson et al., 2009). However, the major concern for the use of ozone for antibiotic oxidation is the potential transformation to products that remain biologically active and resistant to further ozonation. Dodd et al. (2009) reported that the ozonation products of b-lactam antibiotics (i.e. penicillin and cephalexin) are still biologically active after primary oxidation reactions, but could be further deactivated by hydroxyl radicals or ozone if the concentration of residual ozone is sufficient. However, in the case of roxithromycin, the primary ozonation products have the bactericidal dimethylamino groups preserved and are quite persistent to further degradation at very high ozone doses (Radjenovic et al.,
4316
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
2009). Desethylene ciprofloxacin is an ozone byproduct of ciprofloxacin, for which formation can be significantly affected by solution pH (De Witte et al., 2009).
4.3.3.
Ultraviolet irradiation
Ultraviolet (UV) irradiation can be used to degrade some organic chemicals in water (Rosenfeldt and Linden, 2004). Degradation is governed by UV energy absorption and quantum yield of that compound (Kim et al., 2009). The DOC concentration, UV dose and contact time are also important factors governing the removal efficiency. Negligible removal of antibiotics from WWTP secondary effluents through UV disinfection process with a typical dose of 30e80 mJ cm2 has been often reported (Batt et al., 2007; Drewes et al., 2008; Le-Minh et al., 2010). Although sulfonamide antibiotics such as sulfamethoxazole are more prone to photolytic degradation (Boreen et al., 2004), the poor removal (25e50%) of the compound is still observed during UV disinfection process (Drewes et al., 2008), due to the presence of DOC in the treated effluent which highly competes for the limited UV radiation energy at typical disinfection dose. The degradation of antibiotics tends to be effective only at very high UV radiation dose, about 20e100 times higher than the typical disinfection dose for wastewater effluent (Adams et al., 2002; Kim et al., 2009). It has been recently reported that at high UV doses of nearly 3000 mJ/cm2 and DOC of 2.5e4 mg L1, a 5 min contact time was required to achieve >90% removal for sulfamethoxazole and norfloxacin, while 15 min contact time was required for tetracycline (Kim et al., 2009).
antibiotics were removed from tertiary treated wastewater by the AOP system of 7 mg/L ozone, 3.5 mg/L H2O2 and 2 min contact time. The degradation of metronidazole using photochemical oxidations including UV, UV/H2O2, H2O2/Fe2þ (Fenton), and UV/H2O2/Fe2þ (photo-Fenton) has been investigated (Shemer et al., 2006). This study showed that while the removal of metronidazole was negligible solely by UV irradiation (a dose of 600 mJ/cm2 and 5 min retention), the addition of ferrous ions significantly enhanced the removal efficiency. However, all samples in this study were spiked deionised water samples; and the effectiveness of photochemical oxidations for treating metronidazole in wastewater is expected to be reduced because of the increased turbidity and interference of matrix compounds. The overall extent of oxidation for any AOP is dependent on the contact time and the concentration of scavengers in the water (i.e., non-target oxidisable species). Typically, dissolved organic carbon (DOC) and carbonate/bicarbonate are the most important scavengers in drinking waters. High concentrations of DOC and carbonate/bicarbonate can render mineralisation of micropollutants quite inefficient (von Gunten, 2003). The presence of natural organic material can initiate the formation of hydroxyl radical whereas the humic acid and bicarbonate can scavenge the radicals (Drewes et al., 2008). However, pre-treatment processes, such as GAC or RO, significantly reduce DOC concentrations, thus enhancing oxidation efficiency.
5. 4.3.4.
Conclusion
Advanced oxidation processes
Oxidative degradation can occur either by direct reaction with the applied oxidant, or via the production of highly reactive secondary species, most commonly, hydroxyl radicals (OH), which is one of the most powerful oxidants known (Metcalf and Eddy, 2007). Processes that promote the enhanced formation of hydroxyl radicals are generally referred to as advanced oxidation processes (AOPs). UV radiation is commonly used to promote the formation of hydroxyl radicals. This can be achieved by a number of methods including photocatalysis with titanium dioxide (TiO2) (Egerton et al., 2006; Murray and Parsons, 2006) or by direct reaction of hydrogen peroxide (H2O2) (Rosenfeldt and Linden, 2004). Other systems, such as Ozone/H2O2 and UV/Ozone are also considered AOPs since they promote the formation of hydroxyl radicals. Significant research has been conducted in the past to investigate the performance of AOPs for removing pharmaceutical contaminants in treated wastewater (Drewes et al., 2008; Rosenfeldt and Linden, 2004; Westerhoff et al., 2005). From these studies, AOPs were found to be effective treatment processes for removing the selected pharmaceutical contaminants and to provide the improved removal efficiency compared to UV radiation or ozonation alone. Although few studies on the performance of AOPs for oxidising antibiotics are available, AOPs also appear to be effective for oxidising antimicrobial contaminants. The study using lab-scale and full-scale ozone/H2O2 treatment systems showed that the concentrations of erythromycin, ofloxacin, sulfamethoxazole and trimethoprim in tertiary treated wastewater were significantly reduced (Drewes et al., 2008). More than 90% of these
Antibiotic pharmaceuticals, administered for human medication, represent a broad range of chemical classes including b-lactams, sulfonamides, macrolides, fluoroquinolones, tetracyclines, and nitroimidazoles, as well as the important compound, trimethoprim. This diversity of chemical classes is equally represented by variability of physical, chemical and biochemical properties, resulting in variable susceptibility to physical, chemical and biological treatment processes. b-lactams are characterised by the fact that they are generally very quickly degraded during conventional wastewater treatment processes. This observed behaviour is presumed to be a function of susceptibility to chemical and biochemical hydrolysis of the b-lactam ring. However, observed behaviours of other groups of antibiotics are much more difficult to characterise due to varying removal efficiencies reported from studies undertaken in different parts of the world. The most reasonable interpretation appears to be that specific design and operating conditions of individual WWTPs is an important determinative factor for antibiotic chemical removal. Unfortunately, insufficient information is currently available to thoroughly assess these differences. Inconsistencies in research objectives and focuses may also partially explain differences in reported removal efficiencies from previous studies. In fact, many previous studies often calculated removal efficiencies by the difference in an antibiotics’ concentration in treated effluent from that of the influent, but not taking into account the potential intertransformation between antibiotics and their metabolites in wastewater, particularly in the case of sulfonamide classes.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
Furthermore, the occurrence of some enzymatic degradation is known to significantly enhance the biodegradation of some antibiotics. Such enzymatic degradation is governed by several factors, such as HRT, SRT and operational temperature. Many studies did not specify, or well define, these operational factors during their investigation, rendering results difficult to compare. In spite of the variable removal of antibiotics during conventional WWTP processes, many of these chemicals are routinely observed in secondary treated effluents within the range of 10e500 ng L1. Whether these concentrations are of any public health significance is currently unknown, however the simple fact that they have been observed and reported has been a cause of scientific and public concern. Accordingly, for many water recycling applications e especially those involving a high degree of personal contact or are designed to augment potable supplies - the efficacy of advanced water treatment processes for antibiotics removal is of interest. Comparatively little information is available regarding the effectiveness of advanced treatment processes including membrane filtration, adsorptive treatment (activated carbon and ionic adsorption), or chemical and photochemical oxidation. Nonetheless, with the aid of a few basic physicalchemical molecular properties, it has been possible to qualitatively characterise various antibiotics in terms of susceptibility to some of these processes. While there appears to be no ‘silver bullet’ for the removal of all residual antibiotics under typical operational conditions, there is a strong implication that judicious selection of advanced treatment processes can be used for the effective removal of these compounds to levels unlikely to be detectable by current analytical procedures. It is recommended that more comprehensive studies are required to fill knowledge gaps in the behaviour of antibiotics under conventional sewage treatment and advanced treatment processes. Future research should include a dedicated focus on the adsorption and fate of the antibiotics into the sewage biomass, the potential formation of pharmacologically active or more toxic metabolites and degradation products during treatment processes. Furthermore, future studies should control and report all basic treatment plant operational parameters since these are essential for later comparison or assessments.
Acknowledgements This research was supported under the Australian Research Council’s Discovery Projects funding scheme (project number DP0558029).
references
Adam, M.L., Comfort, S.D., Morley, M.C., Snow, D.D., 2004. Remediating RDX-contaminated ground water with permanganate: laboratory investigations for the Pantex perched aquifer. Journal of Environmental Quality 33 (6), 2165e2173. Adams, C., Wang, Y., Loftin, K., Meyer, M., 2002. Removal of antibiotics from surface and distilled water in conventional
4317
water treatment processes. Journal of Environmental Engineering e ASCE 128 (3), 253e260. Adelowo, O.O., Fagade, O.E., Oke, A.J., 2008. Prevalence of coresistance to disinfectants and clinically relevant antibiotics in bacterial isolates from three hospital laboratory wastewaters in Southwestern Nigeria. World Journal of Microbiology & Biotechnology 24 (9), 1993e1997. Agenson, K.O., Oh, J.I., Urase, T., 2003. Retention of a wide variety of organic pollutants by different nanofiltration/reverse osmosis membranes: controlling parameters of process. Journal of Membrane Science 225 (1e2), 91e103. Aksu, Z., Tunc, O., 2005. Application of biosorption for penicillin G removal: comparison with activated carbon. Process Biochemistry 40 (2), 831e847. Alder, A.C., Bruchet, A., Carballa, M., Joss, A., Loffler, D., McArdell, C.S., Miksch, K., Omil, F., Tuhkanen, T., Ternes, T. A., 2006. In: Ternes, T.A., Joss, A. (Eds.), Human Pharmaceuticals, Hormones and Fragrances. IWA Publishing, London, pp. 15e54. Alexy, R., Ku¨mpel, T., Ku¨mmerer, K., 2004. Assessment of degradation of 18 antibiotics in the closed bottle test. Chemosphere 57, 505e512. American Chemical Society, 2009. SciFinder Scholar. Amin, M.M., Zilles, J.L., Greiner, J., Charbonneau, S., Raskin, L., Morgenroth, E., 2006. Influence of the antibiotic erythromycin on anaerobic treatment of a pharmaceutical wastewater. Environmental Science & Technology 40 (12), 3971e3977. Auerbach, E.A., Seyfried, E.E., McMahon, K.D., 2007. Tetracycline resistance genes in activated sludge wastewater treatment plants. Water Research 41 (5), 1143e1151. Balakrishnan, V.K., Terry, K.A., Toito, J., 2006. Determination of sulfonamide antibiotics in wastewater: a comparison of solid phase microextraction and solid phase extraction methods. Journal of Chromatography A 1131 (1e2), 1e10. Bansal, R.C., Goyal, M., 2005. Activated Carbon Adsorption. CRC Press, Boca Raton, USA. Baquero, F., Martinez, J.L., Canton, R., 2008. Antibiotics and antibiotic resistance in water environments. Current Opinion in Biotechnology 19 (3), 260e265. Batt, A.L., Kim, S., Aga, D.S., 2006. Enhanced biodegradation of iopromide and trimethoprim in nitrifying activated sludge. Environmental Science & Technology 40 (23), 7367e7373. Batt, A.L., Kim, S., Aga, D.S., 2007. Comparison of the occurrence of antibiotics in four full-scale wastewater treatment plants with varying designs and operations. Chemosphere 68 (3), 428e435. Baumgarten, S., Schroder, H.F., Charwath, C., Lange, M., Beier, S., Pinnekamp, J., 2007. Evaluation of advanced treatment technologies for the elimination of pharmaceutical compounds. Water Science and Technology 56 (5), 1e8. Beausse, J., 2004. Selected drugs in solid matrices: a review of environmental determination, occurrence and properties of principal substances. TrAC-Trends in Analytical Chemistry 23 (10e11), 753e761. Belden, J.B., Maul, J.D., Lydy, M.J., 2007. Partitioning and photodegradation of ciprofloxacin in aqueous systems in the presence of organic matter. Chemosphere 66 (8), 1390e1395. Bellona, C., Drewes, J.E., Xu, P., Amy, G., 2004. Factors affecting the rejection of organic solutes during NF/RO treatmenta literature review. Water Research 38 (12), 2795e2809. Bendz, D., Paxeus, N.A., Ginn, T.R., Loge, F.J., 2005. Occurrence and fate of pharmaceutically active compounds in the environment, a case study: Hoje River in Sweden. Journal of Hazardous Materials 122 (3), 195e204. Bhal, S.K., Kassam, K., Peirson, I.G., Pearl, G.M., 2007. The rule of five revisited: applying log D in place of log P in drug-likeness filters. Molecular Pharmaceutics 4 (4), 556e560. Boreen, A.L., Arnold, W.A., McNeill, K., 2004. Photochemical fate of sulfa drugs in the aquatic environment: sulfa drugs
4318
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
containing five-membered heterocyclic groups. Environmental Science & Technology 38 (14), 3933e3940. Botitsi, E., Frosyni, C., Tsipi, D., 2007. Determination of pharmaceuticals from different therapeutic classes in wastewaters by liquid chromatography-electrospray ionization-tandem mass spectrometry. Analytical and Bioanalytical Chemistry 387 (4), 1317e1327. Boussu, K., Kindts, C., Vandecasteele, C., Van der Bruggen, B., 2007. Surfactant fouling of nanofiltration membranes: measurements and mechanisms. Chemphyschem 8 (12), 1836e1845. Boyer, T.H., Singer, P.C., 2005. Bench-scale testing of a magnetic ion exchange resin for removal of disinfection by product precursors. Water Research 39 (7), 1265e1276. Brown, K.D., Kulis, J., Thomson, B., Chapman, T.H., Mawhinney, D.B., 2006. Occurrence of antibiotics in hospital, residential, and dairy effluent, municipal wastewater, and the Rio Grande in New Mexico. Science of the Total Environment 366 (2e3), 772e783. Cahill, J.D., Furlong, E.T., Burkhardt, M.R., Kolpin, D., Anderson, L.G., 2004. Determination of pharmaceutical compounds in surface-and ground-water samples by solidphase extraction and high-performance liquid chromatography-electrospray ionization mass spectrometry. Journal of Chromatography A 1041 (1e2), 171e180. CambridgeSoft, 2007. ChemBioOffice2008. Carballa, M., Omil, F., Lema, J.M., Llompart, M., Garcia-Jares, C., Rodriguez, I., Gomez, M., Ternes, T., 2004. Behavior of pharmaceuticals, cosmetics and hormones in a sewage treatment plant. Water Research 38 (12), 2918e2926. Carballa, M., Omil, F., Lema, J.M., Llompart, M., Garcia-Jares, C., Rodriguez, I., Gomez, M., 2005. Behavior of pharmaceuticals and personal care products in sewage treatment plant of northwest Spain. Water Science & Technology 52 (8), 29e35. Carballa, M., Omil, F., Alder, A.C., Lema, J.M., 2006. Comparison between the conventional anaerobic digestion of sewage sludge and its combination with a chemical or thermal pre-treatment concerning the removal of pharmaceuticals and personal care products. Water Science and Technology 53 (8), 109e117. Carballa, M., Fink, G., Omil, F., Lema, J.M., Ternes, T., 2008. Determination of the solid-water distribution coefficient (Kd) for pharmaceuticals, estrogens and musk fragrances in digested sludge. Water Research 42 (1e2), 287e295. Carberry, J., Englande, A., 1983. Sludge Characteristics and Behavior. Martinus Nijhoff Publishers, Boston, The Hague, Dordrecht, Lancaster. Cardoza, L.A., Knapp, C.W., Larive, C.K., Belden, J.B., Lydy, M.J., Graham, D.W., 2005. Factors affecting the fate of ciprofloxacin in aquatic field systems. Water, Air and Soil Pollution 161 (1e4), 383e398. Castiglioni, S., Bagnati, R., Calamari, D., Fanelli, R., Zuccato, E., 2005. A multiresidue analytical method using solid-phase extraction and high-pressure liquid chromatography tandem mass spectrometry to measure pharmaceuticals of different therapeutic classes in urban wastewaters. Journal of Chromatography A 1092 (2), 206e215. Cha, J.M., Yang, S., Carlson, K.H., 2005. Rapid analysis of trace levels of antibiotic polyether ionophores in surface water by solid-phase extraction and liquid chromatography with ion trap tandem mass spectrometric detection. Journal of Chromatography A 1065 (2), 187e198. Cha, J.M., Yang, S., Carlson, K.H., 2006. Trace determination of beta-lactam antibiotics in surface water and urban wastewater using liquid chromatography combined with electrospray tandem mass spectrometry. Journal of Chromatography A 1115 (1e2), 46e57. Chamberlain, E., Adams, C., 2006. Oxidation of sulfonamides, macrolides, and carbadox with free chlorine and monochloramine. Water Research 40 (13), 2517e2526.
Chang, H., Hu, J.Y., Asami, M., Kunikane, S., 2008. Simultaneous analysis of 16 sulfonamide and trimethoprim antibiotics in environmental waters by liquid chromatography-electrospray tandem mass spectrometry. Journal of Chromatography A 1190 (1e2), 390e393. ChemSW, 2005. Molecular Modelling Pro Plus. Norgwyn Montgomery Software Inc., North Wales, PA. Chen, X.G., Xiao, B.D., Liu, J.T., Fang, T., Xu, X., 2005. Kinetics of the oxidation of MCRR by potassium permanganate. Toxicon 45 (7), 911e917. Choi, K.J., Kim, S.G., Kim, C.W., Kim, S.H., 2007a. Determination of antibiotic compounds in water by on-line SPE-LC/MSD. Chemosphere 66 (6), 977e984. Choi, K.J., Son, H.J., Kim, S.H., 2007b. Ionic treatment for removal of sulfonamide and tetracycline classes of antibiotic. Science of the Total Environment 387 (1e3), 247e256. Choi, K.J., Kim, S.G., Kim, S.H., 2008. Removal of tetracycline and sulfonamide classes of antibiotic compound by powdered activated carbon. Environmental Technology 29 (3), 333e342. Clara, M., Kreuzinger, N., Strenn, B., Gans, O., Kroiss, H., 2005a. The solids retention timeea suitable design parameter to evaluate the capacity of wastewater treatment plants to remove micropollutants. Water Research 39 (1), 97e106. Clara, M., Strenn, B., Gans, O., Martinez, E., Kreuzinger, N., Kroiss, H., 2005b. Removal of selected pharmaceuticals, fragrances and endocrine disrupting compounds in a membrane bioreactor and conventional wastewater treatment plants. Water Research 39 (19), 4797e4807. Cooney, D.O., 1995. Activated Charcoal in Medical Applications. Dekker, New York. Cosar, C., Julou, L., 1959. The activity of 1-(2-hydroxyethyl)-2methyl-5-nitroimidazole (R.P. 8823) against experimental Trichomonas vaginalis infections. Ann Inst Pasteur (Paris) 96 (2), 238e241. Costanzo, S.D., Murby, J., Bates, J., 2005. Ecosystem response to antibiotics entering the aquatic environment. Marine Pollution Bulletin 51 (1e4), 218e223. Crain, F.T., Gottlieb, S., 1935. A study of the mechanism of the effect of chlorine on the biochemical oxygen demand. Transactions of the Kansas Academy of Science 1903 (38), 145e148. De Witte, B., Dewulf, J., Demeestere, K., Van Langenhove, H., 2009. Ozonation and advanced oxidation by the peroxone process of ciprofloxacin in water. Journal of Hazardous Materials 161 (2e3), 701e708. Delgado, J.N., Remers, W.A., 1991. Wilson and Gisvold’s Textbook of Organic Medicinal and Pharmaceutical Chemistry. Lippincott, Philadelphia. Deshpande, A.D., Baheti, K.G., Chatterjee, N.R., 2004. Degradation of beta-lactam antibiotics. Current Science 87 (12), 1684e1695. Diaz-Cruz, M.S., Lopez de Alda, M.J., Barcelo, D., 2003. Environmental behavior and analysis of veterinary and human drugs in soils, sediments and sludge. TrAC-Trends in Analytical Chemistry 22 (6), 340e351. Dickenson, E.R.V., Drewes, J.r.E., Sedlak, D.L., Wert, E.C., Snyder, S.A., 2009. Applying surrogates and indicators to assess removal efficiency of trace organic chemicals during chemical oxidation of wastewaters. Environmental Science & Technology 43 (16), 6242e6247. Dodd, M.C., Buffle, M.-O., von Gunten, U., 2006. Oxidation of antibacterial molecules by aqueous ozone: moiety-specific reaction kinetics and application to ozone-based wastewater treatment. Environmental Science & Technology 40 (6), 1969e1977. Dodd, M.C., Kohler, H.P.E., Von Gunten, U., 2009. Oxidation of antibacterial compounds by ozone and hydroxyl radical: elimination of biological activity during aqueous ozonation
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
processes. Environmental Science & Technology 43 (7), 2498e2504. Dodd, M.C., Huang, C.H., 2007. Aqueous chlorination of the antibacterial agent trimethoprim: reaction kinetics and pathways. Water Research 41 (3), 647e655. Dolar, D., Kosutic, K., Pavlovic, D.M., Kunst, B., 2009. Removal of emerging contaminants of industrial origin by NF/RO e a pilot scale study. Desalination and Water Treatment-Science and Engineering 6 (1e3), 197e203. Donoho, A.L., 1984. Biochemical studies of the fate of monensin in animals and in the environment. Journal of Animal Science 58 (6), 1538e1539. Drewes, J.E., 2007. Analysis, fate and removal of pharmaceuticals in the water cycle. In: Petrovic, M., Barcelo, D. (Eds.), Wilson & Wilsons. Elsevier, Amsterdam, pp. 427e446. Drewes, J.E., Bellona, C., Oedekoven, M., Xu, P., Kim, T.U., Amy, G., 2006. Rejection of wastewater-derived micropollutants in high-pressure membrane applications leading to indirect potable reuse. Environmental Progress 24 (4), 400e409. Drewes, J.E., Sedlak, D.L., Snyder, S., Dickenson, E., 2008. Development of Indicators and Surrogates for Chemical Contaminant Removal during Wastewater Treatment and Reclamation e Final Report. WaterReuse Foundation, Alexandra, VA. Drewes, J.E., Khan, S., 2010. Water reuse for drinking water augmentation. In: Edzwald, J. (Ed.), Water Quality and Treatment Handbook, Sixth ed. McGraw-Hill. Dutta, M., Dutta, N.N., Bhattacharya, K.G., 1999. Aqueous phase adsorption of certain beta-lactam antibiotics onto polymeric resins and activated carbon. Separation and Purification Technology 16 (3), 213e224. Egerton, T.A., Christensen, P.A., Kosa, S.A.M., Onoka, B., Harper, J.C., Tinlin, J.R., 2006. Photoelectrocatalysis by titanium dioxide for water treatment. International Journal of Environmental Pollution 27 (1e3), 2e19. Elovitz, M.S., Von Gunten, U., Kaiser, H.P., 2000. Hydroxyl radical/ ozone ratios during ozonation processes. II. The effect of temperature, pH, alkalinity, and DOM properties. OzoneScience & Engineering 22 (2), 123e150. Esponda, S.M., Padron, M.E.T., Ferrera, Z.S., Rodriguez, J.J.S., 2009. Solid-phase microextraction with micellar desorption and HPLC-fluorescence detection for the analysis of fluoroquinolones residues in water samples. Analytical and Bioanalytical Chemistry 394 (4), 927e935. Exall, K., 2004. A review of water reuse and recycling, with reference to Canadian practice and potential: 2. Applications. Water Quality Research Journal of Canada 39 (1), 13e28. Fatta, D., Nikolaou, A., Achilleos, A., Meric, S., 2007. Analytical methods for tracing pharmaceutical residues in water and wastewater. TrAC-Trends in Analytical Chemistry 26 (6), 515e533. Gartiser, S., Urich, E., Alexy, R., Kummerer, K., 2007. Anaerobic inhibition and biodegradation of antibiotics in ISO test schemes. Chemosphere 66 (10), 1839e1844. Gibs, J., Stackelberg, P.E., Furlong, E.T., Meyer, M., Zaugg, S.D., Lee Lippincott, R., 2007. Persistence of pharmaceuticals and other organic compounds in chlorinated drinking water as a function of time. Science of the Total Environment 373 (1), 240e249. Glassmeyer, S.T., Shoemaker, J.A., 2005. Effects of chlorination on the persistence of pharmaceuticals in the environment. Bulletin of Environmental Contamination and Toxicology 74 (1), 24e31. Gobel, A., McArdell, C.S., Suter, M.J.F., Giger, W., 2004. Trace determination of macrolide and sulfonamide antimicrobials, a human sulfonamide metabolite, and trimethoprim in wastewater using liquid chromatography coupled to
4319
electrospray tandem mass spectrometry. Analytical Chemistry 76 (16), 4756e4764. Gobel, A., Athomsen, A., McArdell, C.S., Joss, A., Giger, W., 2005a. Occurrence and sorption behavior of sulfonamides, macrolides, and trimethoprim in activated sludge treatment. Environmental Science Technology 39 (11), 3981e3989. Gobel, A., Thomsen, A., McArdell, C.S., Alder, A.C., Giger, W., Thei, N., Loffler, D., Ternes, T.A., 2005b. Extraction and determination of sulfonamides, macrolides, and trimethoprim in sewage sludge. Journal of Chromatography A 1085 (2), 179e189. Gobel, A., McArdell, C.S., Joss, A., Siegrist, H., Giger, W., 2007. Fate of sulfonamides, macrolides, and trimethoprim in different wastewater treatment technologies. Science of the Total Environment 372 (2e3), 361e371. Golet, E.M., Alder, A.C., Giger, W., 2002a. Environmental exposure and risk assessment of fluoroquinolone antibacterial agents in wastewater and river water of the Glatt Valley Watershed, Switzerland. Environmental Science & Technology 36 (17), 3645e3651. Golet, E.M., Strehler, A., Alder, A.C., Giger, W., 2002b. Determination of fluoroquinolone antibacterial agents in sewage sludge and sludge-treated soil using accelerated solvent extraction followed by solid-phase extraction. Analytical Chemistry 74 (21), 5455e5462. Golet, E., Xifra, I., Siegrist, H., Alder, A.C., Giger, W., 2003. Environmental exposure assessment of fluoroquinolone antibacterial agents from sewage to soil. Environmental Science Technology 37 (15), 3243e3249. Gros, M., Petrovic, M., Barcelo, D., 2006. Development of a multiresidue analytical methodology based on liquid chromatography-tandem mass spectrometry (LC-MS/MS) for screening and trace level determination of pharmaceuticals in surface and wastewaters. Talanta 70 (4), 678e690. Gulkowska, A., Leung, H.W., So, M.K., Taniyasu, S., Yamashita, N., Yeunq, L.W.Y., Richardson, B.J., Lei, A.P., Giesy, J.P., Lam, P.K.S., 2008. Removal of antibiotics from wastewater by sewage treatment facilities in Hong Kong and Shenzhen, China. Water Research 42 (1e2), 395e403. von Gunten, U., 2003. Ozonation of drinking water: part I. Oxidation kinetics and product formation. Water Research 37 (7), 1443e1467. von Gunten, U., Janex-Habibi, M.-L., Ternes, A.T., Weber, L., 2006. In: Ternes, T.A., Joss, A. (Eds.), Human Pharmaceutical, Hormones and Fragrances: The Challenge of Micropollutants in Urban Water Management. IWA Publishing, London, pp. 293e322. Gustavson, L.E., Kaiser, J.F., Edmonds, A.L., Locke, C.S., DeBartolo, M.L., Schneck, D.W., 1995. Effect of omeprazole on concentrations of clarithromycin in plasma and gastric tissue at steady state. Antimicrobial Agents and Chemotherapy 39 (9), 2078e2083. Halling-Sorensen, B., Lutzhoft, H.C.H., Andersen, H.R., Ingerslev, F., 2000. Environmental risk assessment of antibiotics: comparison of mecillinam, trimethoprim and ciprofloxacin. Journal of Antimicrobial Chemotherapy 46 (Suppl. 1), 53e58. Herna´ndez, F., Sancho, J.V., Iba´n˜ez, M., Guerrero, C., 2007. Antibiotic residue determination in environmental waters by LC-MS. TrAC Trends in Analytical Chemistry 26 (6), 466e485. Hirsch, R., Ternes, T.A., Haberer, K., Mehlich, A., Ballwanz, F., Kratz, K.-L., 1998. Determination of antibiotics in different water compartments via liquid chromatography-electrospray tandem mass spectrometry. Journal of Chromatography A 815 (2), 213e223. Hirsch, R., Ternes, T., Haberer, K., Kratz, K.-L., 1999. Occurrence of antibiotics in the aquatic environment. The Science of the Total Environment 225 (1e2), 109e118.
4320
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
Hou, J.P., Poole, J.W., 1971. Beta-lactam antibiotics e their physicochemical properties and biological activities in relation to structure. Journal of Pharmaceutical Sciences 60 (4), 503e532. Huber, M.M., Gobel, A., Joss, A., Hermann, N., Loffler, D., McArdell, C.S., Ried, A., Siegrist, H., Ternes, T.A., von Gunten, U., 2005. Oxidation of pharmaceuticals during ozonation of municipal wastewater effluents: a pilot study. Environmental Science & Technology 39 (11), 4290e4299. Ingerslev, F., Halling-Sørensen, B., 2000. Biodegradation properties of sulfonamides in activated sludge. Environmental Toxicology and Chemistry 19 (10), 2467e2473. Jen, J.F., Lee, H.L., Lee, B.N., 1998. Simultaneous determination of seven sulfonamide residues in swine wastewater by highperformance liquid chromatography. Journal of Chromatography A 793 (2), 378e382. Ji, L.L., Chen, W., Duan, L., Zhu, D.Q., 2009. Mechanisms for strong adsorption of tetracycline to carbon nanotubes: a comparative study using activated carbon and graphite as adsorbents. Environmental Science & Technology 43 (7), 2322e2327. Jones, O.A.H., Voulvoulis, N., Lester, J.N., 2007. The occurrence and removal of selected pharmaceutical compounds in a sewage treatment works utilising activated sludge treatment. Environmental Pollution 145 (3), 738e744. Joss, A., Keller, E., Alder, A.C., Gobel, A., McArdell, C.S., Ternes, T., Siegrist, H., 2005. Removal of pharmaceuticals and fragrances in biological wastewater treatment. Water Research 39 (14), 3139e3152. Jury, K.L., Khan, S.J., Vancov, T., Stuetz, R.M., Ashbolt, N.J. Are sewage treatment plants promoting antibiotic resistance? Critical Reviews in Environmental Science & Technology, in press. Karthikeyan, K.G., Meyer, M.T., 2006. Occurrence of antibiotics in wastewater treatment facilities in Wisconsin. USA. Science of the Total Environment 361 (1e3), 196e207. Khan, S.J., 2009. In: Van den Hoven, T., Kazner, C. (Eds.), TECHNEAU: Safe Drinking Water from Source to Tap. IWA Publishing, London, UK, pp. 445e456. Khan, S.J., Roser, D.J., Davies, C.M., Peters, G.M., Stuetz, R.M., Tucker, R., Ashbolt, N.J., 2008. Chemical contaminants in feedlot wastes: concentrations, effects and attenuation. Environment International 34 (6), 839e859. Khan, S.J., Ongerth, J.E., 2004. Modelling of pharmaceutical residues in Australian sewage by quantities of use and fugacity calculations. Chemosphere 54 (3), 355e367. Khan, S.J., Ongerth, J.E., 2005. Occurrence and removal of pharmaceuticals at an Australian sewage treatment plant. Water 32 (4), 80e85. Kim, S.-C., Carlson, K., 2005. LC-MS2 for quantifying trace amounts of pharmaceutical compounds in soil and sediment matrices. TrAC Trends in Analytical Chemistry 24 (7), 635e644. Kim, S.-C., Carlson, K., 2006. Occurrence of ionophore antibiotics in water and sediments of a mixed-landscape watershed. Water Research 40 (13), 2549e2560. Kim, S., Eichhorn, P., Jensen, J.N., Webber, A.S., Aga, D., 2005. Removal of antibiotics in wastewater: effect of hydraulic and solid retention times on the fate of tetracycline in the activated sludge process. Environmental Science Technology 39 (15), 5816e5823. Kim, S., Jensen, J.N., Aga, D.S., Weber, A.S., 2007. Tetracycline as a selector for resistant bacteria in activated sludge. Chemosphere 66 (9), 1643e1651. Kim, I., Yamashita, N., Tanaka, H., 2009. Performance of UV and UV/H2O2 processes for the removal of pharmaceuticals detected in secondary effluent of a sewage treatment plant in Japan. Journal of Hazardous Materials 166 (2e3), 1134e1140. Kimura, K., Toshima, S., Amy, G., Watanabe, Y., 2004. Rejection of neutral endocrine disrupting compounds (EDCs) and
pharmaceutical active compounds (PhACs) by RO membranes. Journal of Membrane Science 245 (1e2), 71e78. Kirst, H.A., 2002. In: Schonfeld, W., Kirst, H.A. (Eds.), Macrolide Antibiotics. Birkhauser Verlag, Basel, pp. 1e14. Kobayashi, Y., Yasojima, M., Komori, K., Suzuki, Y., Tanaka, H., 2006. Removal characteristics of human antibiotics during wastewater treatment in Japan. Water Practice & Technology 1 (3), 1e9. Kosutic, K., Dolar, D., Asperger, D., Kunst, B., 2007. Removal of antibiotics from a model wastewater by RO/NF membranes. Separation and Purification Technology 53 (3), 244e249. Kummerer, K., 2003. Significance of antibiotics in the environment. Journal of Antimicrobial Chemotherapy 52 (1), 5e7. Kummerer, K., Al-Ahmad, A., Mersch-Sundermann, V., 2000. Biodegradability of some antibiotics, elimination of the genotoxicity and affection of wastewater bacteria in a simple test. Chemosphere 40 (7), 701e710. Lange, F., Cornelissen, S., Kubac, D., Sein, M.M., von Sonntag, J., Hannich, C.B., Golloch, A., Heipieper, H.J., Moder, M., von Sonntag, C., 2006. Degradation of macrolide antibiotics by ozone: a mechanistic case study with clarithromycin. Chemosphere 65 (1), 17e23. Le-Minh, N., Coleman, H.M., Khan, S.J., van Luer, Y., Trang, T.T.T., Watkins, G., Stuetz, R.M., 2010. The application of membrane bioreactors as decentralised systems for removal of endocrine disrupting chemicals and pharmaceuticals. Water Science and Technology 61 (5), 1081e1088. Leiros, H.K.S., Kozielski-Stuhrmann, S., Kapp, U., Terradot, L., Leonard, G.A., McSweeney, S.M., 2004. Structural basis of 5nitroimidazole antibiotic resistance e the crystal structure of NimA from Deinococcus radiodurans. Journal of Biological Chemistry 279 (53), 55840e55849. Levine, A.D., Meyer, M.T., Kish, G., 2006. Evaluation of the persistence of micropollutants through pure-oxygen activated sludge nitrification and denitrification. Water Environment Research 78 (11), 2276e2285. Li, S.Z., Li, X.Y., Wang, D.Z., 2004. Membrane (RO-UF) filtration for antibiotic wastewater treatment and recovery of antibiotics. Separation and Purification Technology 34 (1e3), 109e114. Li, J.D., Cai, Y.Q., Shi, Y.L., Mou, S.F., Jiang, G.B., 2007. Determination of sulfonamide compounds in sewage and river by mixed hemimicelles solid-phase extraction prior to liquid chromatography-spectrophotometry. Journal of Chromatography A 1139 (2), 178e184. Li, D., Yang, M., Hu, J., Zhang, Y., Chang, H., Jin, F., 2008. Determination of penicillin G and its degradation products in a penicillin production wastewater treatment plant and the receiving river. Water Research 42 (1e2), 307e317. Li, B., Zhang, T., Xu, Z.Y., Fang, H.H.P., 2009. Rapid analysis of 21 antibiotics of multiple classes in municipal wastewater using ultra performance liquid chromatography-tandem mass spectrometry. Analytica Chimica Acta 645 (1e2), 64e72. Lin, A.Y.C., Yu, T.H., Lateef, S.K., 2009. Removal of pharmaceuticals in secondary wastewater treatment processes in Taiwan. Journal of Hazardous Materials 167 (1e3), 1163e1169. Lindberg, R.H., Wennberg, P., Johansson, M.I., Tysklind, M., Andersson, B.A.V., 2005. Screening of human antibiotic substances and determination of weekly mass flows in five sewage treatment plants in Sweden. Environmental Science & Technology 39 (10), 3421e3429. Lindberg, R., Olofsson, U., Rendahl, O., Tysklind, M., Andersson, B.A.V., 2006. Behavior of fluoroquinolones and trimethoprim during mechanical, chemical, and active sludge treatment of sewage water and digestion of sludge. Environmental Science Technology 40 (3), 1042e1048.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
Lo¨ffler, D., Ternes, T.A., 2003. Analytical method for the determination of the aminoglycoside gentamicin in hospital wastewater via liquid chromatography-electrospray-tandem mass spectrometry. Journal of Chromatography A 1000 (1e2), 583e588. Loftin, K.A., Adams, C.D., Meyer, M.T., Surampalli, R., 2008. Effects of ionic strength, temperature, and pH on degradation of selected antibiotics. Journal of Environmental Quality 37 (2), 378e386. Loke, M.L., Tjornelund, J., Halling-Sorensen, B., 2000. Determination of the distribution coefficient (logKd) of oxytetracycline, tylosin A, olaquindox and metronidazole in manure. Chemosphere 48 (3), 351e361. Maria, C.R., Reginald, F.C., 1993. Capillary electrophoresis separation of sulphonamides and dihydrofolate reductase inhibitors. Journal of Microcolumn Separations 5 (3), 207e215. Marzo, A., Dal Bo, L., 1998. Chromatography as an analytical tool for selected antibiotic classes: a reappraisal addressed to pharmacokinetic applications. Journal of Chromatography A 812 (1e2), 17e34. Masters, P.A., O’Bryan, T.A., Zurlo, J., Miller, D.Q., Joshi, N., 2003. Trimethoprimesulfamethoxazole revisited. Archives of Internal Medicine 163 (4), 402e410. McArdell, C.S., Molnar, E., Suter, M.J.F., Giger, W., 2003. Occurrence and fate of macrolide antibiotics in wastewater treatment plants and in the Glatt Valley watershed. Switzerland. Environmental Science Technology 37 (24), 5479e5486. Metcalf, Eddy, 2007. Water Reuse: Issues, Technologies, and Applications. McGraw Hill, New York. Miao, X.S., Bishay, F., Chen, M., Metcalfe, C.D., 2004. Occurrence of antimicrobials in the final effluents of wastewater treatment plants in Canada. Environmental Science Technology 38 (13), 3533e3541. Morse, A., Jackson, A., 2004. Fate of amoxicillin in two water reclamation systems. Water Air and Soil Pollution 157 (1e4), 117e132. Murray, C.A., Parsons, S.A., 2006. Preliminary laboratory investigation of disinfection by-product precursor removal using an advanced oxidation process. Water and Environment Journal 20 (3), 123e129. Nagulapally, S.R., Ahmad, A., Henry, A., Marchin, G.L., Zurek, L., Bhandari, A., 2009. Occurrence of ciprofloxacin-, trimethoprim-sulfamethoxazole-, and vancomycin-resistant bacteria in a municipal wastewater treatment plant. Water Environment Research 81 (1), 82e90. Neu, H.C., 1992. The crisis in antibiotic resistance. Science 257 (5073), 1064e1073. Nghiem, L.D., Schafer, A.I., Elimelech, M., 2004. Removal of natural hormones by nanofiltration membranes: measurement, modeling, and mechanisms. Environmental Science & Technology 38 (6), 1888e1896. Nghiem, L.D., Schafer, A.I., Elimelech, M., 2005. Pharmaceutical retention mechanisms by nanofiltration membranes. Environmental Science & Technology 39 (19), 7698e7705. Nghiem, L.D., Oschmann, N., Schafer, A.I., 2006. Fouling in greywater recycling by direct ultrafiltration. Desalination 187 (1e3), 283e290. Nghiem, D.L., Vogel, D., Khan, S.J., 2009. Characterising humic acid fouling of nanofiltration membranes using bisphenol A as a molecular indicator. Water Research 43 (4), 863e881. Nghiem, L.D., Hawkes, S., 2007. Effects of membrane fouling on the nanofiltration of pharmaceutically active compounds (PhACs): mechanisms and role of membrane pore size. Separation and Purification Technology 57 (1), 176e184. Nghiem, L.D., Schafer, A.I., 2006. Fouling autopsy of hollow-fibre MF membranes in wastewater reclamation. Desalination 188 (1e3), 113e121.
4321
Oschmann, N., Nghiem, L.D., Schafer, A.I., 2005. Fouling mechanisms of submerged ultrafiltration membranes in greywater recycling. Desalination 179 (1e3), 215e223. Paxeus, N., 2004. Removal of selected non-steroidal antiinflammatory drugs (NSAIDs), gemfibrozil, carbamazepine, beta-blockers, trimethoprim and triclosan in conventional wastewater treatment plants in five EU countries and their discharge to the aquatic environment. Water Science and Technology 50 (5), 253e260. Peng, X., Wang, Z., Kuang, W., Tan, J., Li, K., 2006. A preliminary study on the occurrence and behavior of sulfonamides, ofloxacin and chloramphenicol antimicrobials in wastewaters of two sewage treatment plants in Guangzhou, China. Science of the Total Environment 371 (1e3), 314e322. Peng, X., Tan, J., Tang, C., Yu, Y., Wangt, Z., 2008. Multiresidue determination of fluoroquinolone, sulfonamide, trimethoprim, and chloramphenicol antibiotics in urban waters in China. Environmental Toxicology and Chemistry 27 (1), 73e79. Perez, S., Eichhorn, P., Aga, D.S., 2005. Evaluating the biodegradability of sulfamethazine, sulfamethoxazole, sulfathiazole and trimethoprim at different stages of sewage treatment. Environmental Toxicology and Chemistry 24 (6), 1361e1367. du Pisani, P.L., 2006. Direct reclamation of potable water at Windhoek’s Goreangab reclamation plant. Desalination 188 (1e3), 79e88. Porter, M.C., 1990. Handbook of Industrial Membrane Technology. Noyes Publications, Park Ridge, New Jersey, USA. Powell, S.J., Wilmot, A.J., Macleod, I., Elsdon-Dew, R., 1966. Metronidazole in amoebic dysentery and amoebic liver abscess. The Lancet 2 (7477), 1329e1331. Putra, E.K., Pranowo, R., Sunarso, J., Indraswati, N., Ismadji, S., 2009. Performance of activated carbon and bentonite for adsorption of amoxicillin from wastewater: mechanisms, isotherms and kinetics. Water Research 43 (9), 2419e2430. Quiang, Z., Adams, C., 2004. Potentiometric determination of acid dissociation constants (pK(a)) for human and veterinary antibiotics. Water Research 38 (12), 2874e2890. Rabolle, M., Spliid, N.H., 2000. Sorption and mobility of metronidazole, olaquindox, oxytetracycline and tylosin in soil. Chemosphere 40 (7), 715e722. Radjenovic, J., Godehardt, M., Petrovic, M., Hein, A., Farre, M., Jekel, M., Barcelo, D., 2009. Evidencing generation of persistent ozonation products of antibiotics roxithromycin and trimethoprim. Environmental Science & Technology 43 (17), 6808e6815. Reif, R., Suarez, S., Omil, F., Lema, J.M., 2008. Fate of pharmaceuticals and cosmetic ingredients during the operation of a MBR treating sewage. Desalination 221 (1e3), 511e517. Renew, J.E., Huang, C.H., 2004. Simultaneous determination of fluoroquinolone, sulfonamide, and trimethoprim antibiotics in wastewater using tandem solid phase extraction and liquid chromatography-electrospray mass spectrometry. Journal of Chromatography A 1042 (1e2), 113e121. Rivera-Utrilla, J., Prados-Joya, G., Sanchez-Polo, M., FerroGarcia, M.A., Bautista-Toledo, I., 2009. Removal of nitroimidazole antibiotics from aqueous solution by adsorption/bioadsorption on activated carbon. Journal of Hazardous Materials 170 (1), 298e305. Robberson, K.A., Waghe, A.B., Sabatini, D.A., Butler, E.C., 2006. Adsorption of the quinolone antibiotic nalidixic acid onto anion-exchange and neutral polymers. Chemosphere 63 (6), 934e941. Roberts, P.H., Thomas, K.V., 2006. The occurrence of selected pharmaceuticals in wastewater effluent and surface waters of
4322
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
the lower Tyne catchment. Science of the Total Environment 356 (1e3), 143e153. Rodriguez, C., Van Buynder, P., Lugg, R., Blair, P., Devine, B., Cook, A., Weinstein, P., 2009. Indirect potable reuse: a sustainable water supply alternative. International Journal of Environmental Research and Public Health 6 (3), 1174e1209. Rosenfeldt, E.J., Linden, K.G., 2004. Degradation of endocrine disrupting chemicals bisphenol A, ethinyl estradiol, and estradiol during UV photolysis and advanced oxidation processes. Environmental Science & Technology 38 (20), 5476e5483. Salisbury, C.D.C., 1995. In: Oka, H., Nakazawa, H., Harada, K.E., MacNeil, J.D. (Eds.), Chemical Analysis for Antibiotics Used in Agriculture. AOAC International, Toronto. Schaep, J., Vandecasteele, C., Mohammed, A.W., Bowen, W.R., 2001. Modelling the retention of ionic components for different nanofiltration membranes. Separation and Purification Technology 22e23, 169e179. Schafer, A.I., Kalinowski, J., Puhler, A., 1994. Increased fertility of Corynebacterium glutamicum recipients in intergeneric matings with Escherichia coli after stress exposure. Applied and Environmental Microbiology 60 (2), 756e759. Schafer, A.I., Nghiem, L.D., Waite, T.D., 2003. Removal of the natural hormone estrone from aqueous solutions using nanofiltration and reverse osmosis. Environmental Science and Technology 37 (1), 182e188. Schlu¨sener, M.P., Bester, K., Spiteller, M., 2003. Determination of antibiotics such as macrolides, ionophores and tiamulin in liquid manure by HPLC-MS/MS. Analytical and Bioanalytical Chemistry 375 (7), 942e947. Schneider, J., 1961. Treatment of giardiasis (lambliasis) by metronidazole. Bulletin de la Sociiti. de Pathologie Exotique 54, 84e95. Segura, P.A., Garcia-Ac, A., Lajeunesse, A., Ghosh, D., Gagnon, C., Sauve, S., 2007. Determination of six anti-infectives in wastewater using tandem solidphase extraction and liquid chromatography-tandem mass spectrometry. Journal of Environmental Monitoring 9 (4), 307e313. Shemer, H., Kunukcu, Y.K., Linden, K.G., 2006. Degradation of the pharmaceutical metronidazole via UV, Fenton and photoFenton processes. Chemosphere 63 (2), 269e276. Shi, L., Zhou, X.F., Zhang, Y.L., Gu, G.W., 2009. Simultaneous determination of 8 fluoroquinolone antibiotics in sewage treatment plants by solid-phase extraction and liquid chromatography with fluorescence detection. Water Science and Technology 59 (4), 805e813. Simon, A., Nghiem, L.D., Le-Clech, P., Khan, S.J., Drewes, J.E., 2009. Effects of membrane degradation on the removal of pharmaceutically active compounds (PhACs) by NF/RO filtration processes. Journal of Membrane Science 340 (1e2), 16e25. Skold, O., 2001. Resistance to trimethoprim and sulfonamides. Veterinary Research 32 (3e4), 261e273. Sneader, W., 2005. Drug Discovery: a History. John Wiley & Sons, West Sussex, England. Snyder, S.A., Westerhoff, P., Yoon, Y., Sedlak, D.L., 2003. Pharmaceuticals, personal care products, and endocrine disruptors in water: implications for the water industry. Environmental Engineering Science 20 (5), 449e469. Sponza, D.T., Demirden, P., 2007. Treatability of sulfamerazine in sequential upflow anaerobic sludge blanket reactor (UASB)/ completely stirred tank reactor (CSTR) processes. Separation and Purification Technology 56 (1), 108e117. Staehelin, J., Hoigne, J., 1985. Decomposition of ozone in water in the presence of organic solutes acting as promoters and inhibitors of radical chain reactions. Environmental Science & Technology 19 (12), 1206e1213. Stoob, K., Singer, H.P., Goetz, C.W., Ruff, M., Mueller, S.R., 2005. Fully automated online solid phase extraction coupled directly
to liquid chromatography-tandem mass spectrometry: quantification of sulfonamide antibiotics, neutral and acidic pesticides at low concentrations in surface waters. Journal of Chromatography A 1097 (1e2), 138e147. Szczepanowski, R., Linke, B., Krahn, I., Gartemann, K.H., Gutzkow, T., Eichler, W., Puhler, A., Schluter, A., 2009. Detection of 140 clinically relevant antibiotic-resistance genes in the plasmid metagenome of wastewater treatment plant bacteria showing reduced susceptibility to selected antibiotics. Microbiology-Sgm 155, 2306e2319. Tadkaew, N., Sivakumar, M., Khan, S.J., McDonald, J.A., Nghiem, L.D., 2010. Effect of mixed liquor pH on the removal of trace organic contaminants in a membrane bioreactor. Bioresource Technology 101 (5), 1494e1500. Tauxe-Wuersch, A., De Alencastro, L.F., Grandjean, D., Tarradellas, J., 2005. Occurrence of several acidic drugs in sewage treatment plants in Switzerland and risk assessment. Water Research 39 (9), 1761e1772. Ternes, T.A., Stuber, J., Herrmann, N., McDowell, D., Ried, A., Kampmann, M., Teiser, B., 2003. Ozonation: a tool for removal of pharmaceuticals, contrast media and musk fragrances from wastewater? Water Research 37 (8), 1976e1982. Theron, M.M., Janse van Rensburg, M.N., Chalkley, L.J., 2004. Nitroimidazole resistance genes (nimB) in anaerobic Grampositive cocci (previously Peptostreptococcus spp.). Journal Antimicrobial Agents Chemotherapy 54 (1), 240e242. Todar, K., 2002. In: Todar, K. (Ed.), Todars Online Textbook of Bacteriology. Tolls, J., 2001. Sorption of veterinary pharmaceuticals in soils: a review. Environmental Science and Technology 35 (17), 3397e3406. Turiel, E., Bordin, G., Rodrı´guez, A.R., 2003. Trace enrichment of (fluoro)quinolone antibiotics in surface waters by solid-phase extraction and their determination by liquid chromatographyultraviolet detection. Journal of Chromatography A 1008 (2), 145e155. Upcroft, J.A., Campbell, R.W., Benakli, K., Upcroft, P., Vanelle, P., 1999. Efficacy of new 5-nitroimidazoles against metronidazole-susceptible and -resistant Giardia, Trichomonas, and Entamoeba spp. Antimicrobial Agents and Chemotherapy 43 (1), 73e76. Verliefde, A.R.D., Cornelissen, E.R., Heijman, S.G.J., Verberk, J.Q.J.C., Amy, G.L., Van der Bruggen, B., van Dijk, J.C., 2008. The role of electrostatic interactions on the rejection of organic solutes in aqueous solutions with nanofiltration. Journal of Membrane Science 322 (1), 52e66. Vieno, N., Tuhkanen, T., Kronberg, L., 2007a. Elimination of pharmaceuticals in sewage treatment plants in Finland. Water Research 41 (5), 1001e1012. Vieno, N.M., Harkki, H., Tuhkanen, T., Kronberg, L., 2007b. Occurrence of pharmaceuticals in river water and their elimination a pilot-scale drinking water treatment plant. Environmental Science & Technology 41 (14), 5077e5084. Vigneswaran, S., Sundaravadivel, M., 2004. In: Saravanamuthu, V. (Ed.), Wastewater Recycle, Reuse and Reclamation. Eolss Publishers, Oxford, UK. Watkinson, A.J., Murbyc, E.J., Costanzo, S.D., 2007. Removal of antibiotics in conventional and advanced wastewater treatment: implications for environmental discharge and wastewater recycling. Water Research 41 (18), 4164e4176. Watkinson, A.J., Murby, E.J., Kolpin, D.W., Costanzo, S.D., 2009. The occurrence of antibiotics in an urban watershed: from wastewater to drinking water. Science of the Total Environment 407 (8), 2711e2723. Westerhoff, P., Yoon, Y., Snyder, S.A., Wert, E.C., 2005. Fate of endocrine-disruptor, pharmaceutical, and personal care product chemicals during simulated drinking water treatment processes. Environmental Science & Technology 39 (17), 6649e6663.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 2 9 5 e4 3 2 3
Wintgens, T., Melin, T., Schafer, A., Khan, S., Muston, M., Bixio, D., Thoeye, C., 2005. The role of membrane processes in municipal wastewater reclamation and reuse. Desalination 17 (178), 1e11. Xiao, Y., Chang, H., Jia, A., Hu, J.Y., 2008. Trace analysis of quinolone and fluoroquinolone antibiotics from wastewaters by liquid chromatography-electrospray tandem mass spectrometry. Journal of Chromatography A 1214 (1e2), 100e108. Xu, P., Drewes, J.E., Kim, T.U., Bellona, C., Amy, G., 2006. Effect of membrane fouling on transport of organic contaminants in NF/RO membrane applications. Journal of Membrane Science 279 (1e2), 165e175. Xu, W.H., Zhang, G., Li, X.D., Zou, S.C., Li, P., Hu, Z.H., Li, J., 2007. Occurrence and elimination of antibiotics at four sewage treatment plants in the Pearl River Delta (PRD), South China. Water Research 41 (19), 4526e4534. Yang, S., Carlson, K.H., 2004. Solid-phase extraction-highperformance liquid chromatography-ion trap mass spectrometry for analysis of trace concentrations of macrolide
4323
antibiotics in natural and waste water matrices. Journal of Chromatography A 1038 (1e2), 141e155. Yang, S., Cha, J., Carlson, K., 2005. Simultaneous extraction and analysis of 11 tetracycline and sulfonamide antibiotics in influent and effluent domestic wastewater by solid-phase extraction and liquid chromatography-electrospray ionization tandem mass spectrometry. Journal of Chromatography A 1097 (1e2), 40e53. Yasojima, M., Nakada, N., Komori, K., Suzuki, Y., Tanaka, H., 2006. Occurrence of levofloxacin, clarithromycin and azithromycin in wastewater treatment plant in Japan. Water Science & Technology 53 (11), 227e233. Zorita, S., Martensson, L., Mathiasson, L., 2009. Occurrence and removal of pharmaceuticals in a municipal sewage treatment system in the south of Sweden. Science of the Total Environment 407 (8), 2760e2770. Zuccato, E., Castiglioni, S., Fanelli, R., 2005. Identification of the pharmaceuticals for human use contaminating the Italian aquatic environment. Journal of Hazardous Materials 122 (3), 205e209.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 2 3 e4 6 2 9
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Formation of methyl iodide on a natural manganese oxide Se´bastien Allard a, Herve´ Gallard a,*, Claude Fontaine b, Jean-Philippe Croue´ a a Laboratoire de Chimie et Microbiologie de l’Eau UMR CNRS 6008, ESIP, Universite´ de Poitiers, 40 avenue du recteur Pineau, 86 022 Poitiers Cedex, France b Laboratoire Hydroge´ologie, Argiles, Sols et Alte´rations FRE CNRS 3114, Faculte´ des Sciences Fondamentales et Applique´es, Universite´ de Poitiers, 40 avenue du recteur Pineau, 86 022 Poitiers Cedex, France
article info
abstract
Article history:
This paper demonstrates that manganese oxides can initiate the formation of methyl
Received 12 February 2010
iodide, a volatile compound that participates to the input of iodine into the atmosphere.
Received in revised form
The formation of methyl iodide was investigated using a natural manganese oxide in batch
13 May 2010
experiments for different conditions and concentrations of iodide, natural organic matter
Accepted 3 June 2010
(NOM) and manganese oxide. Methyl iodide was formed at concentrations 1 mg L1 for
Available online 11 June 2010
initial iodide concentrations ranging from 0.8 to 38.0 mg L1. The production of methyl iodide increased with increasing initial concentrations of iodide ion and Mn sand and
Keywords:
when pH decreased from 7 to 5. The hydrophilic NOM isolate exhibited the lowest yield of
Manganese oxide
methyl iodide whereas hydrophobic NOM isolates such as Suwannee River HPOA fraction
Natural organic matter
produced the highest concentration of methyl iodide. The formation of methyl iodide could
Iodide
take place through the oxidation of NOM on manganese dioxide in the presence of iodide.
Methyl iodide
However, the implication of elemental iodine cannot be excluded at acidic pH. Manganese oxides can then participate with ferric oxides to the formation of methyl iodide in soils and sediments. The formation of methyl iodide is unlikely in technical systems such as drinking water treatment i.e. for ppt levels of iodide and low contact times with manganese oxides. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Methyl iodide (CH3I) was first analysed in seawater and air by Lovelock et al. (1973). In the atmosphere, methyl iodide is photolysed to produce methyl radical (CH3) and iodine atom (I), which contributes to the natural iodine cycle. Methyl radical reacts with ozone and thus may influence the ozone budget. The oceans would be the main source of methyl iodide through biotic reactions mediated by microorganisms such as algae (Gschwend et al., 1985). Laboratory studies provided evidence that a wide variety of bacteria including terrestrial and marine bacteria are capable of methylating low amount of iodide in the environment (Amachi et al., 2001).
Only a few abiotic mechanisms have been proposed. An abiotic mechanism for the formation of methyl iodide and other iodoalkanes (e.g. iodoethane, iodopropane, .) has been demonstrated in soils and sediments containing natural organic matter (NOM), iodide (I) and Fe(III) as electron acceptor (Keppler et al., 2000). Experiments conducted with model compounds gave evidence that methoxy group could be responsible for the formation of methyl iodide. The mechanism involves the oxidation of NOM by ferric iron followed almost simultaneously by the nucleophilic substitution of the methyl group by iodide. The production of methyl iodide and other iodoalkanes increases with the increase of iodide, NOM and ferric iron concentrations and when pH decreases.
* Corresponding author. Tel.: þ33 5 49 45 44 31. E-mail address:
[email protected] (H. Gallard). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.008
4624
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 2 3 e4 6 2 9
Manganese oxides are widely distributed in the environment and participate with iron oxides to the adsorption and oxidation of NOM (Stone and Morgan, 1984; Chorover and Amistadi, 2001). Manganese oxides are also known to oxidise phenols used as model compound of complex NOM structures (Stone and Morgan, 1984). The oxidation of NOM by manganese oxides results in the production of low molecular weight organic compounds e.g. formaldehyde, acetaldehyde and pyruvate through adsorption and electron transfer (Sunda and Kieber, 1994). In ground waters, manganese is present as a form of soluble manganous ion species MnII at ppb to ppm levels. Iodine can also be encountered in ground waters as iodide at ppt levels (about 0e100 mg L1) with increasing concentrations near sea coast (Fuge and Johnson, 1986). Manganese and iodine are essential trace elements for human and small amounts from food or water is required to stay healthy. However, manganese has to be removed from drinking water because its oxidation products cause taste, odours and coloured water problems (Wong, 1984; Jaudon et al., 1989). Recent studies have reported links between excessive manganese exposure and neurologic disorders in children (Bouchard et al., 2007). The WHO guideline for this element in drinking water is 400 mg L1 (WHO, 2006) and a maximum concentration level of 50 mg L1 in the European Union. Manganous ion oxidation by oxygen is a slow process at neutral pH (Stumm and Morgan, 1996). Strong oxidants e.g. chlorine or ozone can be used but catalytic materials such as natural manganese oxides or manganese oxide coated sands are preferred in some circumstances especially for mineral waters when strong oxidants are prohibited. Removal of iron (White and Asfar-Siddique, 1997) and arsenic (Bissen and Frimmel, 2003; Ouvrard et al., 2002) can also be performed with manganese oxides due to their relatively high redox potential. Manganese oxides have also been proposed as an adsorbant for natural organic matter (NOM) (Bernard et al., 1997) and thus for the removal of DBP in drinking water production (Colthurst and Singer, 1982). Manganese dioxide was recently shown to oxidise iodide to iodine and iodate (Fox et al., 2009; Allard et al., 2009) as well as initiating the formation of iodinated organic compounds such as iodoform (Gallard et al., 2009). The use of manganese oxides for the treatment of iodide-containing waters could also be responsible for the formation of iodoalkanes such as methyl iodide in drinking waters. Even though methyl iodide has been recently approved as a soil fumigant by the USEPA to control soil-borne diseases and pests, it is neurotoxic for humans and mutagenic to bacteria and rats (IARC, 1999 and references therein). The International Agency for Research on Cancer (IARC) classified methyl iodide in the group 3 (i.e. not classifiable as to its carcinogenicity to humans) because there is no epidemiological data relevant to its carcinogenicity and there is limited evidence for the carcinogenicity in experimental animals (IARC, 1999). However, methyl iodide is considered to be a human carcinogen in Germany and California. To our knowledge, methyl iodide has never been reported in drinking waters and no international guideline exists for this compound in water. The goal of this study was to demonstrate the possible formation of iodoalkanes mediated by manganese dioxides. Because this study is of interest for both environmental and
technical systems, the experiments were carried out with a natural manganese oxide used in water treatment. The kinetics formation of methyl iodide were followed using a variety of conditions to evaluate the effect of different parameters e.g. pH, concentrations of reactants, nature of NOM.
2.
Material and methods
2.1.
Manganese oxide
The catalytic material was a natural manganese oxide (79% MnO2) approved for drinking water treatment. The material was rinsed with ultra pure water to eliminate the fines before experiments. The selected size fraction was in the range of 300e700 mm. The specific density was 3.88. Specific surface area determined by the BET method using N2 as adsorbate was 26 m2 g1. Manganese sand was dry ground in an agate mortar before X-ray analysis. X-ray diffraction was performed using a PANalytical Xpert Pro diffractometer operated under ˚. reflexion using a Cu Ka radiation at a wavelength of 1.54060 A Three Mn minerals were identified: birnessite, lithiophorite and cryptomelane (see X-ray diffractogram in Figure S1 in supplementary material).
2.2.
Natural organic matters (NOM)
Six NOM isolates (Table 1) extracted and characterized from previous studies were used for adsorption on manganese sand and formation of iodoalkanes. Details of the fractionation procedure have been presented elsewhere (Leenheer, 1981; Croue´ et al., 1999). The colloidal fraction was isolated using 3500 Da dialysis bag against HCl 0.1 N and then HF 0.2 N. The hydrophobic (HPO) fractions refer to the NOM fractions recovered from the XAD-8 resin by elution with an acetonitrile/water mixture; the hydrophobic acid (HPOA) and the transphilic (TPI) fractions refer to the NOM fraction recovered by elution with NaOH from the XAD-8 and XAD-4 resins, respectively. The hydrophilic fraction (HPI) refers to NOM fraction not adsorbed on both XAD-8 and XAD-4 resins. Solid samples of the NOM fractions were obtained by freezeedrying. The NOM fractions were characterized by determination of their Specific UV Absorbance values at 254 nm (SUVA254 ¼ UV254/mg/L of dissolved organic carbon (DOC)) and the % aromatic carbon content determined using solid state 13C nuclear magnetic resonance (NMR) spectroscopy. 13C NMR spectra of Jau River extracts were not determined but the origin and the specific UV absorbance values of these extracts suggest that the aromatic carbon content was similar to the Suwannee River HPOA fraction. The NOM extracts were chosen to describe a large range of properties from humic like (hydrophobic with high aromatic carbon content e.g. Suwannee HPOA fraction) to non-humic like NOM (more hydrophilic with low aromatic carbon content e.g. Loire River HPI fraction).
2.3.
Experimental procedures
The experiments were conducted at room temperature (20 2 C) with solutions prepared with ultra pure water
4625
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 2 3 e4 6 2 9
Table 1 e Characteristics of NOM isolates and methyl iodide formation (conditions: Mn sand 15 g LL1; iodide 10 mg LL1, DOC 5 mgC LL1, pH 5, 10 mM perchlorate, reaction time 24 h). Fraction
Specific UV absorbance@254 nm (L mgC1 m1)
C arom. %
CH3I formation (mg L1)
Hydrophobic HPO Hydrophobic acid HPOA Transphilic TPI TPI Colloid Hydrophilic HPI
3.9 4.6 2.6 2.0 3.9 1.6
ND 26 >11 8 ND 6
0.84 0.84 0.72 0.67 0.58 0.44
Name Source
Rio Jau, Brazil Suwannee River, USA Vienne River, France Loire River, France Rio Jau, Brazil Loire River, France ND not determined.
produced with a Milli-Q water purification system (Millipore). All chemicals used were reagent grade. The reaction was initiated by the addition of a diluted solution containing NOM and/or iodide in 40 mL EPA vials sealed with PFTE-faced silicone septa containing a given mass of manganese sand. Vials were filled without headspace to prevent the loss of volatile iodinated organic compounds and were set on a rotary tumbler for agitation at 25 C in a thermostated room. At given time intervals, a vial was taken and the solution was withdrawn with a 50 mL gas syringe and filtered with 0.45 mm membrane filter (Minisart, diameter 25 mm) before analysis. The use of the same filtration procedure on standard solutions did not cause significant loss of iodoalkanes. Experiments were performed in carbonate buffer at pH 7 and in perchlorate media at pH 5.
2.4.
Analytical procedures
The identification of alkyl iodides was performed by using the headspace GCeMS technique on a Varian CP3800 gas chromatograph coupled with a Varian 1200L mass spectrometer. The separation was carried out on a VF5HS 30 m 0.25 mm capillary column. Initial oven temperature was set at 30 C for 5 min, then temperature increased to 150 C at 20 C min1. The vials were shaken at 80 C with an incubation period of 10 min before injection. Volatile alkyl iodides were analysed using gas chromatography (model Varian 3300) with headspace injection and electron capture detection. Separation of alkyl iodides was carried out on a J&W/DB 624 30 m 0.53 mm column. Nitrogen was used as carrier gas. The oven temperature was set constant at 35 C for 20 min. Detector and injector temperatures were 300 C and 80 C, respectively. The headspace vials were equilibrated for 4 h at 50 C before injection. Quantification of methyl iodide was performed using external calibration standards. Stock standard solutions of methyl iodide were prepared in methanol by introducing 100 mL of analyte into a 40 mL EPA vial sealed with PTFE-faced silicone septa and diluting to volume. Solutions were stored at 20 C in the dark. Standard solutions were prepared in ultra pure water. Detection limit for methyl iodide was 10 ng L1 and the relative standard deviations for five replicates were in the range of 1e5%. Iodide ion analyses were carried out with ion chromatography and conductometric detection after chemical ion
suppression (Dionex AS3000). A Dionex AS19 column (internal diameter: 4 mm; length: 250 mm) and a Dionex AG19 guard column (internal diameter: 4 mm; length: 50 mm) was used with 50 mM NaOH as mobile phase at 30 C. The injection volume was 500 mL. The detection limit was 5 mg L1. Dissolved organic carbon was analyzed using a Shimadzu TOC Vcsh analyser. The detection limit was about 0.1 mgC L1.
3.
Results and discussion
3.1.
Identification of volatile iodoalkanes
Both methyl iodide and ethyl iodide were identified by GC/MS when manganese sand was added to solutions containing both NOM and iodide (see: Figure S2). However, ethyl iodide was only detected at trace levels and was not quantified. Keppler et al. (2003) also reported that methyl iodide was the main iodoalkanes produced from humic acids and the formation of high molecular weight iodoalkanes such as propyl and butyl iodide was significant for soil organic matter only. The different nature of organic matter (i.e. solubility, molecular weight, aromatic content and concentration of methoxy aryl group) between soils and surface waters is probably responsible for this finding. Keppler et al. (2000) also observed the formation of methyl bromide and methyl chloride in presence of ferrihydrite. The formation of these compounds was not investigated in the present study but it can be expected that manganese oxides could also catalyse the production of these compounds.
3.2. Influence of NOM, manganese sand and iodide concentrations at pH 7 The Fig. 1 illustrates the influence of NOM, manganese oxide and iodide concentrations on the formation kinetics of methyl iodide for reaction times up to 120 h. Concentrations range of methyl iodide varied between a few ng L1 and about 1 mg L1. The rates of formation increased with increasing the initial concentrations of manganese sand and iodide. However, for reaction times higher than 60 h, increasing Mn sand concentrations did not necessarily lead to an increase in methyl iodide formation (Fig. 1a). Further experiments conducted with aqueous standard solutions showed that methyl iodide slowly disappears on Mn sand, which could explain the
4626
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 2 3 e4 6 2 9
0.8
a
Effect of MnO2 (Iodide 10 mg L-1, NOM 5mgC L-1) 5g L-1 15g L-1 30g L-1 50g L-1
CH3I (µg L-1)
0.6
0.4
0.2
0.0 1.4
b
-1
-1
Effect of iodide (MnO2 30 g L , NOM 5 mgC L )
1.2 37.2 mg L-1 6.5 mg L-1 3.8 mg L-1 0.8 mg L-1
-1
CH3I (µg L )
1.0 0.8
dP Vmax ½DOCK ¼ dt ½DOCK þ 1
0.6 0.4 0.2 0.0 1.0
c
-1
-1
Effect of NOM (MnO2 15 g L , Iodide 10 mg L ) -1
0.5mgC L 5 mgC L-1 20 mgC L-1 50 mgC L-1 no NOM added
CH3I (µg L-1)
0.8
0.6
to 20 mgC L1 but lower final yield was observed for high DOC of 50 mgC L1. Sunda and Kieber (1994) showed that a maximum in production rate of low molecular weight compounds by manganese oxides was also achieved when all of the adsorption sites are saturated. A decrease in production rate could even be expected at very high NOM concentrations through direct competition for active sites and pore blockage. These data support the model of Keppler et al. (2000) for the formation of iodoalkanes through the oxidation of NOM by an electron acceptor. The initial rates of methyl iodide production calculated from the first data point were found to be linearly proportional to the concentrations of iodide and Mn sand (Fig. 2). The overall production rate for 5 mgC L1 of NOM was 8.62 104 nmol h1 g1 Mn sand mg1 I. A similar approach was used for the experiments performed with different NOM concentrations. As observed by Sunda and Kieber (1994), the initial rates of production followed a Monod kinetics equation:
0.4
0.2
(1)
where [DOC] is the concentration of dissolved organic carbon after filtration, K is a pseudo-equilibrium adsorption constant and Vmax is the maximum reaction rate. The initial rates were found similar in the absence of NOM and in the presence of NOM at a concentration of 0.5 mg L1 DOC. The isotherm was extrapolated to the origin. Values of 0.49 (0.03) nM h1 for Vmax and 0.073 (0.012) L mg1 for K were obtained from the nonlinear least-square regression of our data (Fig. 3). The Vmax value was 1e2 orders of magnitude lower than the values obtained for the formation of acetaldehyde and pyruvate on synthetic Mn oxides (Sunda and Kieber, 1994). A lower value for methyl iodide production was expected because the formation is also limited by the concentration of iodide. The value of K is in the same range as those determined in Sunda and Kieber (1994) and by Waite et al. (1988) for the rates of reduction of Mn oxides by Suwannee River Fulvic acid in
0.0 0
20
40
60
80
100
120 1.6
Fig. 1 e Effect of manganese sand, iodide and NOM concentrations on kinetic formation of methyl iodide at pH 7.0 (carbonate buffer 10 mM).
behaviour observed over a long time scale for high concentrations of Mn sand. Methyl iodide is a good methylating agent of carboxylic acids in alkaline solution where carboxylate acts as the nucleophile in the SN2 substitution reactions (AvilaZarraga and Martinez, 2001). Same reactions occur with phenols. These reactions were identified as the predominant pathway through which methyl iodide was degraded in presence of organic matter in soils (Gan and Yates, 1996). The DOC analysed after filtration in the absence of SR HPOA was 2.05 0.37 mgC L1 i.e. 0.14 0.02 mgC g1 MnO2, which explain the significant formation of methyl iodide observed for this condition (Fig. 1c). The formation of methyl iodide increased with increasing the concentrations of NOM from 0.5
CH3I production rate (nM h-1)
Time (hours)
1.4 1.2 1.0
MnO2
0.8
Iodide
0.6 0.4 0.2 0.0 0
10
20
30
40
50
60
MnO2 (g L-1) or iodide (mg L-1)
Fig. 2 e Effect of Mn sand (closed symbols; iodide 10 mg LL1) and iodide (open symbols; Mn sand 30 g LL1) concentrations on the initial rate of methyl iodide production (NOM 5 mgC LL1, pH 7.0, 10 mM carbonate buffer).
4627
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 2 3 e4 6 2 9
0.3
0.2
0.1
0.0 0
10
20
30
40
NOM (mgC/L)
Fig. 3 e Influence of organic carbon concentration on the initial rate of methyl iodide production (Mn sand 15 g LL1, iodide 10 mg LL1, pH 7.0, 10 mM carbonate buffer).
50 mM NaCl (i.e. 0.16 and 0.023 L mg1 at pH 4.0 and 7.1, respectively). For all experiments, the evolution of iodide was also analysed by ion chromatography. At pH 7.0, the concentrations of iodide slowly decreased for 100 h (see Fig. 4), which can be attributed either by diffusion-limited adsorption or slow oxidation on the manganese sand. Diffusion-limited adsorption of inorganic anions was also observed for similar Mn product (Ouvrard et al., 2002). Results in Fig. 4 show that the rate and extent of iodide disappearance decreased significantly when NOM concentrations increase from 0 to 50 mgC L1, which can be explained by competition mechanisms between negatively charged iodide and NOM.
3.3.
Influence of electrolyte composition and pH
with the SR HPOA fraction (5 mgC L1) and iodide (10 mg L1). The alkalinity varied from 1.3 to 6.6 mM and the concentrations of calcium and sulphate varied from 0.3 to 13.7 mM and 0 to 16 mM, respectively. The pH values of the solutions after the 24-h reaction time were in the range of 6.7e7.3. Linear negative correlations were only found with both bicarbonate concentrations (r2 ¼ 0.847, n ¼ 10) and final pH (r2 ¼ 0.668, n ¼ 10). No clear conclusion can be proposed because these two parameters are directly linked. Lower final pH of 6.7 was observed for low alkalinity of 1.3 mM. The role of pH is more likely to be of more significant importance because adsorption of bicarbonate is very weak. To study the effect of pH, the formation of methyl iodide was compared at pH 7.0 (10 mM carbonate buffer) and pH 5.0 (10 mM perchlorate media). Both the rates of methyl iodide production (Fig. 5a) and iodide decay (Fig. 5b) increased by a factor of 5 from pH 7.0 to 5.0 conditions, which confirmed the role of pH in the formation of methyl iodide. According to the model of Keppler et al. (2003), the higher formation of iodoalkanes at acidic pH can be attributed to the higher rate of NOM oxidation by metal oxides when pH decreases (Stone and Morgan, 1984). The analysis of thermodynamic data of Mn and iodine species (Truesdale et al., 2001) and recent studies on iodide oxidation by MnO2 (Allard et al., 2009; Fox et al., 2009) demonstrate that iodide is readily oxidised to
1.0
a
0.8 CH3I (µg L-1)
CH3I Production rate (nM h-1)
0.4
0.6 0.4 pH 5 perchlorate media pH 7 carbonate buffer
0.2 To evaluate the influence of electrolyte composition, 15 g L1 of Mn sand was added to ten different mineral waters spiked
0.0 1.0
b
1.2
Iodide (C/Co)
1.0
Iodide (C/Co)
pH 5 perchlorate media pH 7 carbonate buffer
0.9
0.8 0.6 no NOM added 0.5 mgC L-1 5 mgC L-1 20 mgC L-1 50 mgC L-1
0.4 0.2
0.8
0.7
0.6 0
0.0 0
20
40
60
80
100
Time (Hours)
Fig. 4 e Effect of NOM on loss of iodide (Mn sand 30 g LL1, iodide 10 mg LL1, pH 7.0, 10 mM carbonate buffer).
50
100
150
200
Time (hours) Fig. 5 e Effect of pH on the formation of methyl iodide and iodide loss (Mn sand 15 g LL1, SR HPOA 5 mgC LL1, iodide 10 mg LL1).
4628
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 2 3 e4 6 2 9
iodate (reaction (2)) by manganese oxides for pH < 6.5 and reactive iodine species (e.g. I2) are produced as intermediate species (reaction (3)). I þ 3MnO2 þ 6Hþ /3Mn2þ þ IO 3 þ 3H2 O
(2)
2I þ MnO2 þ 4Hþ /Mn2þ þ I2 þ 2H2 O
(3)
The oxidation of iodide to iodine and iodate could then explain the quick disappearance of iodide at pH 5.0 and the increase of the formation rate of methyl iodide through electrophilic iodination of natural organic matter. However, assuming that methyl iodide could also be formed through the monoiodination of a terminal methyl group, this reaction is not thermodynamically favoured because iodine is a poor electrophilic oxidant and methyl iodide is not a good leaving group. Manganese oxide might act as a catalyst of the reaction as it was suggested for iodoform formation through the activation of the iodine molecule (Gallard et al., 2009).
3.3.1.
Effect of the nature of NOM
The kinetics of methyl iodide formation was studied for six different NOM isolates (colloid, hydrophobic, transphilic and hydrophilic fractions) at pH 5.0 in 10 mM perchlorate solution (5 mgC L1 DOC, 10 mg L1 iodide and 15 g L1 Mn sand). Table 1 gives the concentrations of methyl iodide analysed after a reaction time of 24 h. The lowest formation of methyl iodide (i.e. 0.44 mg L1) was obtained for the Loire River hydrophilic fraction. Higher formation of methyl iodide (i.e. 0.84 mg L1) was observed for the Suwannee and Jau River hydrophobic NOM fractions. This result is in agreement with the study of Keppler et al. (2003) where the highest formation of methyl iodide was obtained with organic matter extracted from peaty soil. These lignin-derived organic matters are enriched in aryl methoxy groups that are responsible of the formation of methyl iodide according to the mechanism proposed by Keppler et al. (2000). Even though higher formation was observed for the hydrophobic fractions, no direct correlation could be obtained between methyl iodide formation and global parameters such as the SUVA values or the aromatic content of NOM. This can be explained by the fact that methyl iodide is produced at very low concentrations from very specific sites within NOM. Further investigations and characterization of NOM are needed to fully explain the role of the chemical composition of NOM.
4.
Conclusion
This study shows that natural manganese sand can initiate the formation of methyl iodide in the presence of iodide and natural organic matter. B Low amount of methyl iodide (<1 mg L1) was formed at pH 7 in carbonate buffer for concentrations of iodide up to 38 mg L1. The initial rate of formation linearly increased with the concentrations of iodide and Mn sand. The production of methyl iodide reached a plateau when NOM concentration increased, which corresponds probably to the saturation of the Mn sand. The formation of methyl
iodide was lower for the hydrophilic fraction of NOM compared to hydrophobic (i.e. humics) fractions. Decreasing the pH from 7.0 to 5.0 caused a strong increase in both rates of methyl iodide formation and iodide disappearance. The formation of methyl iodide through the nucleophilic substitution of methyl group by iodide or the electrophilic substitution of iodine could not be distinguished in this study. Further experiments are needed to conclude about the mechanism controlling the formation of methyl iodide by manganese oxide. B Even though manganese oxides are less abundant than ferric oxides in the environment, these results suggest that they can also contribute to the formation of methyl iodide in soils and sediments. The experiments were carried out with high concentrations of iodide and NOM, which allowed the formation of significant amount of methyl iodide. Much lower formation is expected in the environment or in technical systems i.e. for lower iodide and NOM concentrations. Also, only traces of methyl iodide are probably produced during the filtration of iodide-containing waters on MnO2 bed filters because the hydraulic retention times are limited to 10e15 min compared to several hours for this study.
Appendix. Supplementary material Supplementary data associated with article can be found in online version, at 10.1016/j.watres.2010.06.008.
references
Allard, S., von Gunten, U., Sahli, E., Nicolau, R., Gallard, H., 2009. Oxidation of iodide and iodine on birnessite (d-MnO2) in the pH range 4e8. Water Research 43 (14), 3417e3426. Amachi, S., Kamagata, Y., Kanagawa, T., Muramatsu, Y., 2001. Bacteria mediate methylation of iodine in marine and terrestrial environments. Applied Environmental Microbiology 67 (6), 2718e2722. Avila-Zarraga, J.G., Martinez, R., 2001. Efficient methylation of carboxylic acids with potassium hydroxide/methyl sulfoxide and iodomethane. Synthetic Communications 31 (14), 2177e2183. Bernard, S., Chazal, P., Mazet, M., 1997. Removal of organic compounds by adsorption on pyrolusite (bMnO2). Water Research 31, 1216e1222. Bissen, M., Frimmel, F.H., 2003. Arsenic e a review. Part II: oxidation of arsenic and its removal in water treatment. Acta Hydrochimica et Hydrobiologica 31, 97e107. Bouchard, M., Laforest, F., Vandelac, L., Bellinger, D., Mergler, D., 2007. Hair manganese and hyperactive behaviors: pilot study of school-age children exposed through tap water. Environmental Health Perspectives 115, 122e127. Chorover, J., Amistadi, M.K., 2001. Reaction of forest floor organic matter at goethite, birnessite and smectite surfaces. Geochimica et Cosmochimica Acta 65, 95e109. Colthurst, J.M., Singer, P.C., 1982. Removing trihalomethane precursors by permanganate oxidation and manganese dioxide adsorption. Journal of American Water Works Association 74, 78e83.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 2 3 e4 6 2 9
Croue´, J.P., Debroux, J.F., Amy, G.L., Aiken, G.R., Leenheer, J.A., 1999. Natural organic matter: structural characteristics and reactive properties. In: Singer, P. (Ed.), Formation and Control of Disinfection By-Products in Drinking Water. American Water Works Association, Denver, CO, pp. 65e93. Fox, P.M., Davis, J.A., Luther III, G.W., 2009. The kinetics of iodide oxidation by the manganese oxide mineral birnessite. Geochimica et Cosmochimica Acta 73 (10), 2850e2861. Fuge, R., Johnson, C.C., 1986. The geochemistry of iodine e a review. Environmental Geochemistry and Health 8, 31e54. Gan, J., Yates, S.R., 1996. Degradation and phase partition of methyl iodide in soils. Journal of Agricultural and Food Chemistry 44, 4001e4008. Gallard, H., Allard, S., Nicolau, R., von Gunten, U., Croue´, J.P., 2009. Formation of iodinated organic compounds by oxidation of iodide-containing waters with manganese dioxide. Environmental Science and Technology 43 (18), 7003e7009. Gschwend, P.M., Mac Farlane, J.K., Newman, K.A., 1985. Volatile halogenated organic compounds released to seawater from temperate marine macroalgae. Science 227, 1033e1035. IARC, 1999. Monographs on the Evaluation of Carcinogenic Risks to humans. Methyl iodide. 71 III, pp. 1503e1510. Jaudon, P., Massiani, C., Galea, J., Rey, J., Vacelet, E., 1989. Groundwater pollution by manganese. Manganese speciation: application to the selection and discussion of an in situ groundwater treatment. Science of the Total Environment 84, 169e183. Keppler, F., Borchers, R., Elsner, P., Fahimi, I., Pracht, J., Scho¨ler, H. F., 2003. Formation of volatile iodinated alkanes in soil: results from laboratory studies. Chemosphere 53, 477e483. Keppler, F., Eiden, R., Niedan, V., Pracht, J., Scho¨ler, H.F., 2000. Halocarbons produced by natural oxidation processes during degradation of organic matter. Nature 403, 298e301. Leenheer, J.A., 1981. Comprehensive approach to preparative isolation and fractionation of dissolved organic carbon from
4629
natural waters and wastewaters. Environmental Science and Technology 15, 578e587. Lovelock, J.E., Maggs, R.J., Wade, R.J., 1973. Halogenated hydrocarbons in and over the Atlantic. Nature 241, 194e196. Ouvrard, S., Simonnot, O., de Donato, P., Sardin, M., 2002. Diffusion-controlled adsorption of arsenate on a natural manganese oxide. Industrial and Engineering Chemistry Research 41, 6194e6199. Stone, A.T., Morgan, J.J., 1984. Reduction and dissolution of manganese(III) and manganese(IV) oxides by organics: 2. Survey of the reactivity of organics. Environmental Science and Technology 18, 617e624. Stumm, W., Morgan, J.J., 1996. Aquatic Chemistry, third ed. WileyInterscience, John Wiley & Sons, Inc. Sunda, W.G., Kieber, D.J., 1994. Oxidation of humic substances by manganese oxides yields low-molecular-weight organic substrates. Nature 367, 62e64. Truesdale, V.W., Watts, S.F., Rendell, A.R., 2001. On the possibility of iodide oxidation in the near-surface of the Black Sea and its implication to iodine in the general ocean. Deep Sea Research 48, 2397e2412. Waite, T.D., Wrigley, I.C., Szymczak, R., 1988. Photoassisted dissolution of a colloidal manganese oxide in the presence of fulvic acid. Environmental Science and Technology 22, 778e785. White, D.A., Asfar-Siddique, A., 1997. Removal of manganese and iron from drinking water using hydrous manganese dioxide. Solvent Extraction and Ion Exchange 15, 1133e1145. Wong, J.M., 1984. Chlorination-filtration for iron and manganese removal. Journal of American Water Works Association 76, 76e79. World Health Organization, 2006. Guidelines for drinking-water Quality Incorporating first Addendum, vol. 1, Recommendations. e third ed.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 1 7 e4 5 2 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
In situ biomonitoring of freshwater quality using the New Zealand mudsnail Potamopyrgus antipodarum (Gray) exposed to waste water treatment plant (WWTP) effluent discharges Marion Gust a,b, Thierry Buronfosse c, Olivier Geffard a, Raphael Mons a, Herve Queau a, Jacques Mouthon d, Jeanne Garric a,* a
Cemagref, UR MALY, Laboratoire d’e´cotoxicologie, 3b quai Chauveau, 69009 Lyon, France AgroParisTech ENGREF, 19 Avenue du Maine, 75732 Paris, France c Universite´ de Lyon. Laboratoire d’endocrinologie, Ecole Nationale Ve´te´rinaire de Lyon, Avenue Bourgelat, 69280 Marcy l’Etoile, France d Cemagref, UR MALY, Laboratoire DYNAM, 3b quai Chauveau 69009 Lyon, France b
article info
abstract
Article history:
Mollusk species have been shown to be sensitive to various endocrine disrupting
Received 5 November 2009
compounds (EDC) at environmentally relevant concentrations. Waste water treatment
Received in revised form
plant (WWTP) effluents are a major source of potential or known EDC in the aquatic
1 June 2010
environment. The aim of this study was to develop an in situ exposure method using the
Accepted 8 June 2010
New Zealand mudsnail Potamopyrgus antipodarum (Molluska, Hydrobiidea) to assess the
Available online 15 June 2010
impact of water quality on the life traits of this species, by focusing on its reproduction. The impact of three WWTP discharges on three different receiving rivers was studied. The effects of WWTP effluent on adult survival, weight, reproduction and vertebrate-like sex-
Keywords: Waste
water
treatment
plant
steroid levels in snails were monitored for three to four weeks. Although the physico-
effluent
chemical and hydrological parameters varied greatly between the rivers, the caging
Potamopyrgus antipodarum
experiments allowed us to detect significant impairment of the life traits of snails when
Reproduction
exposed downstream of the WWTPs discharge. While adult survival was not affected by
Steroids hormones
exposure, reproduction was significantly impacted downstream from the WWTP effluent
Caging experiment
discharges (60e70% decrease of embryos without shells after three to four weeks exposure) independently of the river. Modulations of steroid levels proved to be an informative parameter with an increase of testosterone downstream of the discharges, and increases and decreases of 17b-estradiol levels according to site. The endpoints used proved to be an adapted method for field exposures and allowed the discrimination between upstream and downstream sites. ª 2010 Elsevier Ltd. All rights reserved.
* Corresponding author. Present address: Cemagref, UR MALY, Laboratoire d’e´cotoxicologie, 3bis quai Chauveau, CP 220, 69336 Lyon cedex 09, France. Tel.: þ33 472208739; fax: þ33 478477875. E-mail addresses:
[email protected] (M. Gust),
[email protected] (T. Buronfosse),
[email protected] (O. Geffard),
[email protected] (R. Mons),
[email protected] (H. Queau),
[email protected] (J. Mouthon),
[email protected] (J. Garric). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.019
4518
1.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 1 7 e4 5 2 8
Introduction
In situ approaches are an improvement in comparison to current monitoring techniques because of their low cost and ease of application. They also allow the translocation of organisms to investigate selected sites even in the absence of native organisms. In addition, they provide a time-integrated assessment of environmental quality, reflecting exposure over several weeks (Regoli et al., 2006). Gammarus or bivalves are the main aquatic species used in field caging experiments (see for example Dedourge-Geffard et al., 2009; Gagne´ et al., 2001, 2004; Gagne et al., 2002; Blaise et al., 2003; Quinn et al., 2004). Additional species must be introduced to replace or complement the species currently used in ecotoxicology to take better account of species diversity and improve environmental risk assessment (Breitholtz et al., 2006). The addition of a gastropod species such as Potamopyrgus antipodarum (Molluska, Hydrobiidea) in a battery of experimental methodologies could help ensuring better representativeness of the biological sensitivity of aquatic macroinvertebrates, particularly when reprotoxic effects are assessed. The freshwater mudsnail P. antipodarum has been proposed as a suitable test species to assess the impact of EDC on invertebrates and has been recommended for use in the development of reproduction tests within the OECD guideline “Ad hoc Expert Group on Invertebrate Testing” (Duft et al., 2007). Also, it is currently considered to be the most promising species for use in partial lifecycle testing (Matthiessen, 2008). Indeed, known endocrine disrupting compounds (EDC) have been shown to impair mollusk reproduction at environmentally relevant concentrations (Duft et al., 2003; Oehlmann et al., 2006) although certain results are contradictory (Aufderheide et al., 2006; Selck et al., 2006), due to the high variability of results obtained from the endpoints measured, and lack of knowledge of their endocrinology. P. antipodarum is an invasive parthenogenetic ovoviviparous freshwater mudsnail now common in European countries. It thrives in running water from small creeks to streams, lakes and estuaries, in mud and sand, on rocks, gravel and aquatic plants (Winterbourn, 1970) and it a sensitive test organism already used in laboratory tests and experimental streams. It displays comparable response and relative sensitivity when compared to fish species exposed to WWTP effluent (Jobling et al., 2003). In situ exposure of caged P. antipodarum has been successfully performed to measure the effects of pollutant discharges on its survival (Brown, 1980). The effects of WWTP effluent on gastropod reproduction have been demonstrated by several studies using experimental streams (Watton and Hawkes, 1984; Jobling et al., 2003; Clarke et al., 2009), and shown a significant increase of their fecundity when exposed to an estrogenic effluent. However no in situ studies have yet described effects at the individual level in mollusks. The effects of point source pollution on mollusks in the field are mainly related to biochemical effects, through the use of different biomarkers on bivalves (metallothionein, vitellin, DNA damage, metabolic activities, stress biomarkers, etc.), either measured on autochthonous (Gagne´ et al., 2007) or caged organisms (Gagne´ et al., 2001, 2004; Gagne et al., 2002; Blaise et al., 2003; Quinn et al., 2004). However, few of these
biomarkers are potentially indicative of endocrine disruption. Although debate continues on the biological relevance of steroids debated, many observations support a physiological role of vertebrate-like sex steroids in mollusk reproduction (reviewed in Matthiessen, 2008), which could make their measurement a useful tool in endocrine disruption assessment. Vertebrate-like sex steroids have been identified in P. antipodarum by using both radioimmunoassay and LC-MS analysis (Gust et al., 2009). Our study aimed to develop an efficient in situ biotest to assess the impact of point source pollution on P. antipodarum life traits and biomarkers. To achieve this objective we exposed organisms in three rivers receiving WWTP effluent discharges of varying quantities and characteristics, since WWTP effluents are a major source of pollution in aquatic environments and contain thousands of chemicals including many potential or known EDC (natural hormones produced by animals and humans, synthetic steroids found in contraceptive pills, pesticides, surfactants, plasticizers, etc. Kolpin et al., 2002). The decline of certain freshwater mollusk populations in North America is to a certain extent due to effluent discharges, although numerous other factors exist (Lydeard et al., 2004). The specific purposes of this study were threefold: (i) to determine the suitability of P. antipodarum to assess the impact of point source WWTP effluent on field exposure; (ii) to determine the effects of effluent discharges on its survival, weight reproduction and sex-steroid levels; (iii) and finally to determine if similar response patterns occurred in the three rivers tested.
2.
Material and methods
Adult P. antipodarum organisms were obtained from natural populations collected in a canal close to the Rhone River one month before the beginning of the experiments. The juveniles were obtained from long-term cultures raised in our laboratory (Cemagref, Lyon, France). These cultures were initiated with snails that originated from the previous natural populations.
2.1.
In situ exposure of snails
The organisms were exposed upstream and downstream of three WWTP effluent discharges. These waste water treatment plants were located on three rivers characterised by quite different watersheds, about 30 000, 750 and 220 km2 for the Saone, Bourbre and Ardiere Rivers, respectively. The Saone and Bourbre Rivers drain densely populated urban and industrial areas, while the downstream Ardiere watershed is mainly characterised by large vineyards (Fig. 1). The three WWTPs treat 30 000 (Saone), 78 000 (Bourbre) and 7900 (Ardiere) equivalent inhabitants (see Table S1: Table S2 and Fig. S1 in Supplementary Information, SI, for more data). Adult snails were placed in rigid plastic containers with both extremities replaced by mesh. These containers were placed in cages specifically designed for exposure in rivers. One cage per site was exposed and contained six replicates. The remaining space in the cages was loaded with rocks to
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 1 7 e4 5 2 8
4519
Fig. 1 e Site localisation (the star represents the WWTP effluent discharge localisation and the arrows the sampling site localisations).
ballast the cages and rocks covered with biofilm collected from a reference site (Morcille, Rhone, France) were introduced in the containers to provide food. Temperature monitoring probes were fixed to one cage at each site. Conductivity was measured at least weekly in parallel with the biological measurements. The daily river and effluent flows were provided by the water body authorities and WWTP administrators (http://www.hydro.eaufrance.fr/) and are displayed in Fig. S1 (SI). The size (4.2 0.2 mm) and weight (9.5 1.2 mg) of the adults were measured one day before putting them in the exposure chambers. Experimental designs were similar in the Ardiere and Saone Rivers. One cage was placed in an easily accessible upstream site with characteristics similar to the downstream site (about 20 m in the Saone River and about 7 km in the Ardiere River) and about 50 m downstream of the WWTP effluent discharge (Fig. 1). Six replicates per site were deployed, each with 50 adults and 10 juveniles. The exposures lasted 28 days in the Saone River, from 10 June to 8 July 2008, and 21 days in the Ardiere River, from 10 to 30 June 2008, due to weather conditions. In the Bourbre River exposure also lasted 28 days, from 3 to 30 September 2008. It was performed on two upstream and two downstream sites. One of the upstream sites (Bion upstream) was located in a small tributary of the main stream into which the effluent is discharged. The other (Bourbre upstream) was located in the Bourbre River about 100 m of the confluence of the Bourbre River with the mixed effluent e Bion stream, where the first downstream site was situated (Confluence downstream). The last cage was immerged farther downstream (Bourbre downstream), several hundred meters from the confluence (Fig. 1). Six replicates of 100 adults and three replicates of 30 juveniles were deployed at each site.
2.2.
Biological measurements
Mortalities of adults were counted every week directly in the field. The cages were opened and the organisms were observed. Fecundity was assessed on days 0, 7, 21 and 28 on 30 adults randomly taken from the 6 replicates (5 per replicate) and preserved in the river water until the measurements were performed. The embryos in the brood pouch were counted at the laboratory using a technique adapted from Duft et al. (2003): the number of shelled and unshelled embryos (new embryos) was counted under a binocular microscope. Adult weight was measured again after 21 days exposure.
2.3.
Steroids measurements
The total levels of total testosterone (T), progesterone (P) and 17b-estradiol (E2) using a radioimmunoassay (RIA) technique were measured on days 0, 7, 21 and 28 on pools of 2 P. antipodarum (5e10 pools of 2 snails according to exposure) randomly taken from the containers and preserved in the water river until they were frozen. We used a previously published technique (Gust et al., 2010b). Briefly, pools of 2 snails were homogenized in 1% (v/v) trifluoro acetic acid and extracted twice with methylene chloride. After saponification, steroids were extracted twice with methylene chloride and resuspended in 80 mL 1% steroid free Bovine Serum Albumin (BSA). The samples were assayed using commercial RIA kits for E2 (Orion Diagnostic, Espoo, Finland) and T and P (Diagnostic Systems Laboratories, Texas, USA).
2.4.
Statistical analysis
Normality distribution and homogeneity of variance of the data were tested and one-way ANOVAs were performed.
4520
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 1 7 e4 5 2 8
Significance was set at p 0.05. In the case of rejection of H0, an a posteriori LSD (Least Significant Difference) test was applied. When necessary, a KruskaleWallis test followed by a ManneWhitney U-test were performed. All statistical analyses were performed using Statistica 7.1 for Windows (StatSoft Inc., Tulsa, OK, USA).
3.
Results
3.1.
Saone River
After 28 days of exposure, adult mortality was low (5%) and not significantly different between any upstream and downstream site. Adult weight was similar between upstream and downstream sites and did not differ from day 0 (Table 1). On day 7, no significant effects ( p > 0.05) were detected on the number of embryos in adult snails between upstream and downstream sites. On the contrary, on day 21 the shelled, unshelled and total number of embryos fell ( p < 0.05) downstream versus upstream (decreases of 60%, 48% and 54% respectively). The decrease of the unshelled embryos was sharper on day 28 compared to day 21 (60% decrease downstream versus upstream) and the total number of embryos and the shelled embryos continued to fall (48% and 37%, Table 1). At the upstream site, the total number of embryos did not
vary significantly ( p > 0.05) between the beginning and end of exposure, nevertheless the unshelled embryos significantly increased ( p < 0.05) on day 21 and 28 compared to day 0 and 7 (Fig. 2A). On both upstream and downstream sites, depending on the steroids, we measured significant ( p < 0.05) modifications of their levels during exposure with an increase of T and P levels compared to day 0. Moreover steroid levels in the snails were clearly impacted by the WWTP effluent discharges, with an increase of E2 levels downstream of the effluent discharge versus the upstream site, but it was only significant ( p < 0.05) on day 21 (Fig. 3A). Testosterone levels were also significantly higher ( p < 0.05) downstream versus upstream on days 21 and 28 (Fig. 4A). No significant differences between upstream and downstream P levels in the snails were measured (Fig. 5A).
3.2.
Ardiere River
After 21 days of exposure, 5% and 12% of the adult snails were dead in the upstream and downstream site respectively. Adults exposed at the downstream site for 21 days weighed more than those of the upstream site and day 0 (Table 1). The experiment was stopped at this point to assess the effect on adult fecundity. Once again no significant effect ( p > 0.05) of exposure was detected on the fecundity of the snails located upstream of the discharge during the period. The total
Table 1 e Effects of in situ exposures upstream and downstream of WWTP effluent discharges on P. antipodarum life traits (mean values and standard deviation). Bold characters indicate significant difference (P < 0.05, Dunnett-t test or ManneWithney U-test) from the upstream site. Saone River
Ardiere River
Upstream Downstream Upstream Downstream Adult mortality
5%
Reproduction Day “New” 7 embryos Shelled embryos Total embryos Day “New” 21 embryos Shelled embryos Total embryos Day “New” 28 embryos Shelled embryos Total embryos E2/T
Weight mg/ ad
5%
5%
12%
Bourbre River Bion Bourbre upstream upstream 2%
Confluence Bourbre downstream downstream
1%
2%
2%
2.72 (3.24)
2.85 (3.10)
5.57 (3.56)
5.96 (3.43)
7.46 (2.99) 6.19 (3.33)
3.50 (2.83)
5.29 (2.87)
5.75 (4.68)
7.26 (4.38)
9.83 (6.66)
6.55 (3.79)
7.04 (3.81) 5.19 (2.69)
4.89 (3.60)
3.10 (2.52)
9.52 (6.43)
10.11 (5.67)
15.40 (8.85) 12.24 (6.29)
14.50 (5.52) 11.37 (4.43) 8.39 (4.75)
8.20 (3.44)
6.20 (2.81)
3.28 (2.58)
4.54 (2.98)
1.26 (1.73)
9.63 (4.97) 8.20 (4.85)
4.37 (5.34)
5.17 (5.34)
8.24 (4.57)
3.50 (5.03)
6.82 (4.03)
5.29 (4.41)
7.32 (4.75) 6.67 (5.57)
3.80 (3.74)
7.24 (4.73)
14.44 (5.51) 6.78 (6.68)
11.36 (4.91) 6.55 (4.66)
16.44 (5.80) 14.79 (6.50) 10.60 (7.10)
12.41 (7.62)
5.03 (3.53)
2.03 (2.18)
10.03 (6.12) 10.27 (3.95) 3.07 (5.10)
5.27 (4.15)
5.97 (5.15)
3.77 (3.85)
7.73 (4.45) 5.73 (3.38)
5.53 (3.65)
5.97 (4.05)
11 (6.76)
5.81 (5.14)
17.77 (6.89) 16.00 (5.93) 8.60 (7.11)
11.23 (6.93)
Day 7 Day 21 Day 28
0.57 0.25 0.29
0.55 0.31 0.22
0.18 0.21
0.15 0.17
0.19 0.31
0.14 0.21
Day 0 Day 21
8.37 8.21
8.39
10.67 12.53
12.60
12.90
12.97
0.29 0.29
0.12 0.06
8.37 8.48
9.13
4521
Embryos (in % of Day 0)
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 1 7 e4 5 2 8
300
A
5.03
2.03
Embryos (in % of Day 0)
Saône upstream D28 Saône downstream D28
250
11.00 5.81
200 ***
150
***
***
100 50 0 Embryos without shell 300 250
B
4.54
Embryos with shell
Total embryos
6.82 5.29
Ardière upstream D21 Ardière downstream D21 11.36 6.55
1.26
200 150
*** ***
100 50 0
Embryos without shell
140 Embryos (in % of Day 0)
5.97 3.77
120 100
C 10.0
Embryos with shell
Bion D28 Bourbre upstream D28 Confluence D28 Bourbre downstream D28
7.73 10.2
Total embryos
5.73 5.53
3.07 5.27
80
***
5.97
***
17.7
16.0 8.60
***
***
11.2 ***
60 40 20 0 Embryos without shell
Embryos with shell
Total embryos
Fig. 2 e Number of embryos in the brood pouch per adult P. antipodarum at the end of the exposures in percentage of day 0 (mean value and standard deviation). Underlined values: mean absolute number of embryos. * indicates significant differences from the upstream site (***P < 0.01, Dunnett-t test). A. Saone River. B. Ardiere River. C. Bourbre River. N [ 6 pools of 5 snails.
number of embryos varied from 10.5 (day 0) to 11.2 (day 21), with an increase of the proportion of the unshelled embryos at the end of exposure compared to day 0 (Fig. 2B). On the contrary, the shelled embryos and the total number of embryos were significantly lower ( p < 0.05) downstream compared to upstream on day 7. On day 21, the numbers of unshelled embryos and total embryos in the downstream site were also significantly affected ( p < 0.05) compared to those recorded upstream, with falls of respectively 72% and 44% compared to the upstream snails (Table 1). In both upstream and downstream sites E2 levels in the snails were not significantly different ( p < 0.05) during the exposure compared to day 0. E2 levels were lower in the downstream snails compared to upstream, particularly on day 21 (43% decrease, Fig. 3B). On the contrary T levels were significantly higher ( p < 0.05) compared to day 0 throughout the whole exposure and on both sites except in the upstream snails on day 21. Once again, T levels were significantly higher ( p < 0.05) in downstream snails compared to those upstream on days 7 (216% upstream) and 21 (290% upstream, Fig. 4B). Progesterone levels in the snails increased during the
exposure, but no significant differences were noted between upstream and downstream snails. The highest levels were observed on day 7 (Fig. 5B).
3.3.
Bourbre River
Adult mortality was low (less than 5%) and not significantly different between upstream and downstream sites after 28 days of exposure. Adult weight after 21 days was slightly higher in downstream snails, but clearly increased for all snails compared to day 0 (Table 1). Adult fecundity was quite similar on the two upstream sites, with a reduction of about 30% of the total embryos after 28 days exposure compared to day 0 (Fig. 1C). Nevertheless we measured a significant ( p < 0.05) decrease of unshelled embryos at the Confluence (49%) and Bourbre downstream (33%) sites compared to the upstream sites on day 7. On day 21 unshelled embryos were lower ( p < 0.05) at both downstream sites (51% and 42%) compared to the upstream sites, whereas the number of shelled embryos ( p < 0.05) was lower only at the Confluence site (46%). As in the Saone River, the effects
4522
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 1 7 e4 5 2 8
A
14
Saone upstream Saone downstream
bc
pg/ ind estradiol
12 ab
10
b
8
ab a
6 4
a
ab
a
2 0 D0
B
6
D7
Ardiere downstream
a
a a
4 3
D28
Ardiere upstream
5 pg/ind estradiol
D21
a
a
a
2 1 0 D0
C 18 16 pg/ind estradiol
14
D7
D21
Bion Bourbre upstream Confluence Bourbre downstream
d c c
12
bc
10
bc
ab
8
bc
b
6 4
a a
a
a
2 0 D0
D21
D28
Fig. 3 e 17b-estradiol levels (mean value and standard deviation) in the snails exposed upstream and downstream of WWTP effluent discharge at day 0, 7, 21 and 28 (a < b < c < d; P < 0.05, ManneWithney U-tests). A. Saone River (N [ 6 pools of 2 snails). B. Ardiere River (N [ 6 pools of 2 snails). C. Bourbre River (N [ 10 pools of 2 snails).
were more pronounced on day 28. The decrease ( p < 0.05) of the number of unshelled embryos was higher in the snails closest to the discharge (Confluence: 70% compared to the upstream site) than in the Bourbre downstream (48%). The number of shelled and total embryos also was lower ( p < 0.05) on day 28 at both downstream sites compared to the upstream
sites (49% and 18% for the Confluence and 33% and 11% for the Bourbre downstream, Table S2). Contrary to the two other exposures, E2 levels in snails increased significantly ( p < 0.05) for all sites and days compared to day 0 (except for Bion, day 21). They were significantly ( p < 0.05) higher at the Confluence compared
4523
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 1 7 e4 5 2 8
pg/ind testosterone
A
Saône upstream
35
c
c
Saône downstream
30 25 20 15
b
ab a
a
a
a
10 5 0 D0
pg/ind testosterone
B
40 35 30 25 20 15 10 5 0
D7
pg/ind testosterone
60 50
D28
Ardière upstream d
Ardière downstream c a
a
ab
b
D0
C
D21
D7
Bion Bourbre upstream Confluence Bourbre downstream
40
cd
D21
d
d cd
d b
bc
b
30 20 10
a
a
a
a
0 D0
D21
D28
Fig. 4 e Testosterone levels (mean value and standard deviation) in the snails exposed upstream and downstream of WWTP effluent discharge at day 0, 7, 21 and 28 (a < b < c < d; P < 0.05, ManneWithney U-tests). A. Saone River (N [ 6 pools of 2 snails). B. Ardiere River (N [ 6 pools of 2 snails). C. Bourbre River (N [ 10 pools of 2 snails).
to both upstream sites on day 28. E2 levels were slightly lower at the Bourbre downstream site compared to the Confluence, but remained higher than upstream. On day 28, E2 level was highest at the Confluence site, with an increase of 726% compared to day 0 ( p < 0.05), and 219% and 286% compared to Bion ( p < 0.05) and the second upstream site ( p < 0.05; Fig. 3C). As for the other sites, T levels significantly ( p < 0.05) increased during the exposure and were significantly higher on both downstream sites compared to upstream sites on day 28 (Fig. 4C). As with the other rivers, P levels in snails increased with exposure duration but did not vary significantly between upstream and downstream sites (Fig. 5C).
4.
Discussion
4.1.
Exposure conditions and water quality
Snails were exposed in streams with different surroundings (urban, agricultural and industrial) and hydrological characteristics, and subjected to effluent treated by WWTPs with different characteristics (see SI). The conductivity between rivers ranged from 100 to 800 mS/cm. The highest temperatures were measured in the Saone River and the lowest in the Bourbre River. Whatever the river, exposure temperatures ranged from 14 to 21 C. However the physicochemical
4524
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 1 7 e4 5 2 8
pg/ind progsterone
A
15
Saône upstream Saône downstream bc
pg/ind progesterone
bc
b
a
a
5 0 D7
D21
D28 Ardière upstream
20 b
ab
Ardière downstream
15 ab
10 a
5
ab
a
0 D0
C pg/ind progesterone
bc bc
10
D0
B
c
80 60
D7
Bion Bourbre upstream Confluence Bourbre downstream
c bc bc
40 20
b
b b
b
D21
bc bc
bc
c
a
0 D0
D21
D28
Fig. 5 e Progesterone levels (mean value and standard deviation) in the snails exposed upstream and downstream of WWTP effluent discharge at day 0, 7, 21 and 28 (a < b < c < d; P < 0.05, ManneWithney U-tests). A. Saone River (N [ 6 pools of 2 snails). B. Ardiere River (N [ 6 pools of 2 snails). C. Bourbre River (N [ 10 po ols of 2 snails).
parameters were very similar between upstream and downstream sites. Chemical analysis performed simultaneously with ecotoxicological experiments showed that the three streams were heavily contaminated by beta-blocker pharmaceuticals (around 100 ng/L), estrogenic compounds (alkylphenols: 50e200 ng/L) and steroid hormones downstream from the discharge (Bourbe > Saone > Ardiere). No estrogenic hormones were found in the Ardiere River, while estriol was measured at the downstream site (10 ng/L) in the Saone River and estrone in the Bourbre upstream (2 ng/L) and downstream (15 and 5 ng/L in the immediate vicinity and farther downstream) (Jacquet et al., 2008). In addition, the Ardiere River was contaminated at the downstream site with pesticides (herbicides and fungicides) that were measured throughout the year (Chambre Agriculture Rhone, 2008), in particular tebuconazole (up to 0.47 mg/L in June 2007). Although physicochemical and hydrological parameters were rather different depending on the river, we observed high survival rates and no loss of weight in the snails exposed at the upstream sites. P. antipodarum is not sensitive to variations in
abiotic factors (Lassen and Kristensen, 1978). This supports the suitability of our exposure methodology. However, some variations of fecundity and steroid levels were observed during exposure in the upstream sites. Indeed reproduction was generally constant during exposure in the Saone and Ardiere Rivers, while it decreased in the Bourbre River. T and P levels increased in all three upstream sites during exposure, while E2 levels either increased in the Bourbre River or changed little in the Saone and Ardiere Rivers, upstream. These variations could be linked to the period of the experiment in the year as they match annual variations of fecundity and steroid levels in the natural population used for the experiments (unpublished results). Indeed the Bourbre experiment was performed in fall, while the others were performed in summer. During the summer experiments, snails were just beginning to reproduce and water temperature increased during the exposure. On the contrary, at the beginning of the Bourbre experiment, the snails had reached their highest reproduction capacities. Their fecundity was higher than in previous exposures. In September the temperature of the Bourbre River decreased throughout exposure and rapidly fell below 16 C, with fecundity decreasing
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 1 7 e4 5 2 8
in both upstream sites. As shown by Clarke et al. (2009), it is important to consider the time of exposure in the year in relation to the snails’ breeding season. The upstream and downstream snails in the same experiment were considered to have the same physiological status, thus the temporal variation of reproduction observed upstream was considered physiological. As environmental factors modulate the observed endpoints, comparisons can only be made between upstream and downstream sites of the same river, but not between exposures.
4.2. Responses of P. antipodarum’s life traits to field exposure upstream and downstream of WWTP discharges To our knowledge, this experiment is the first using caging experiments to assess the effects on water quality downstream of point source pollution (WWTP effluent discharges), using reproductive parameters and biomarkers in a mollusk species. Exposure to effluents did not lead to P. antipodarum adult mortality but clearly impacted their reproduction. The causes of such a sub lethal impact need further elucidation. No food was provided to the snails during the exposure, and organisms were fed only with biofilm present on rocks. Issues can be raised due to the possible starvation of the organisms leading to decreased reproduction. Nevertheless, we can hypothesize that the snails were in similar feeding conditions because the adults did not lose weight after 21 days exposure, with increased (but not significant) biomass downstream of the effluent compared to upstream. This could indicate increased food supply at the downstream sites. Moreover, no significant mortality was observed downstream of the discharges. These results allow us to interpret the effects of the exposures on the snails’ fecundity independently of the food resources in the same way as Clarke et al. (2009). Moreover as P. antipodarum feeds on detritus-decomposing bacteria, they are able to feed naturally in submerged cages (Brown, 1980). Thus the decrease of fecundity observed can be reasonably linked to the effluents discharged in the rivers. P. antipodarum fecundity was a very sensitive parameter with a significant reduction of embryo production downstream of all three of the WWTPs discharges. P. antipodarum has already been used to assess the effect of effluents in experimental streams at ambient temperature (Jobling et al., 2003; Clarke et al., 2009). Such semi-field or ex situ exposures have already proved the usefulness of this species in ecotoxicology testing. Contrary to the results found in the literature (Watton and Hawkes, 1984; Jobling et al., 2003; Clarke et al., 2009) where fecundity was mostly induced by effluent exposures, we observed that reproduction decreased downstream from the discharges. We have no clear hypothesis to explain these results in contradiction to those already described in snails exposed to diluted WWTP effluent in experimental streams. In both studies where fecundity was induced, the tested effluent was estrogenic to fish and contained a variety of environmental estrogens and natural and synthetic steroid estrogens (Jobling et al., 2003; Clarke et al., 2009). These chemicals were similar to those measured in our exposure, but the concentrations were much lower for the alkylphenols in our rivers (Jacquet et al., 2008). In the experimental streams used by these authors, dilution was controlled, and the minimum concentration of the effluent was 12.5% (Jobling et al., 2003), i.e.
4525
much higher than that which occurred during our exposures (Table S1). As WWTP effluents are very complex mixtures of many chemicals in various proportions, the different natures of the effluents tested could also explain these contradictory observations. It is unlikely that our exposure conditions were stressful enough downstream of the discharges to mask any potential induction of reproduction due to the presence of known inducing compounds for snails such as alkylphenols, BPA, etc. (Duft et al., 2003; Jobling et al., 2003; Oehlmann et al., 2006), as biomass increased. However, the induction of fecundity in the presence of effluent rich in organic matter may be the result of improved food supply, as already shown in other studies using experimental streams (Watton and Hawkes, 1984). Nevertheless it would appear that the various compounds (steroids, pharmaceuticals, etc.) released into the rivers by the effluent discharges interfered with the reproductive performance of P. antipodarum. The mechanisms of the fall in reproduction observed are still unknown.
4.3.
Variations of steroid levels during exposure
This is the first report of steroid measurement and variations in the freshwater gastropod P. antipodarum exposed in a river downstream of a WWTP effluent discharge. Vertebrate-like steroids, including 17b-estradiol (E2) and testosterone (T), have been identified in several bivalve and gastropod mollusks. Their origin (exogenous or endogenous), however, remains unclear, although the main steps of their endogenous synthesis have been described (reviewed in Janer and Porte, 2007). Moreover, certain temporal variations in their levels have been linked to the reproductive cycle in bivalves (reviewed in Lafont and Mathieu, 2007) and gastropods (Sternberg et al., 2008). In prosobranch gastropods, estrogen receptors (ER) have been identified (Bannister et al., 2007; Castro et al., 2007; Sternberg et al., 2008). They are orthologues of Octopus vulgaris ER, which is a strong constitutive transcriptional activator, but does not bind E2 and is unresponsive to vertebrate steroid hormones (Keay et al., 2006). The biological relevance of steroids is still being debated, but these observations support the assumption that sex steroids play a physiological role in mollusk reproduction, although their mode of action remains unknown. Increased steroid levels can be either explained by the bioaccumulation of estrogenic, androgenic and progesteronic compounds released by WWTPs, or by an increase in their biosynthesis. Bioaccumulation of T and E2 is described in several gastropod and bivalves species. Conjugation of steroids (T, E2) with fatty acids can regulate the level of free hormones, which are assumed to be the physiologically active fraction (Gooding and LeBlanc, 2001; Labadie et al., 2007). The enzymes involved in these mechanisms can be impaired by chemicals (Janer et al., 2006). The RIA technique used in this study to measure sex steroids is specific and permitted measuring the total levels of steroids, both free and conjugated. The P levels increased in all three exposures compared to day 0, independently of the exposure site. As no strong modulation of this steroid was observed, the measurement of this synthesis intermediary does not seem very informative. On the contrary, E2 and T levels were strongly modulated by the effluent discharges. The increase in T levels observed at all
4526
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 1 7 e4 5 2 8
downstream sites were several-fold above the physiological values observed during the annual cycle of our reference population of P. antipodarum (personal data). This compound is also easily bioaccumulated in gastropods (Gooding and LeBlanc, 2001). No T measurements were performed downstream of the discharges studied, but they had already been shown to be released by WWTP effluents (Kolodziej et al., 2003). Thus it can be assumed that the increased T levels finally observed in the snails encaged in the three rivers compared to those observed at the beginning of the exposures can be attributed more due to bioaccumulation than biosynthesis. The E2 levels observed in the Bourbre and Saone downstream were also up to 15-fold above the physiological values of P. antipodarum (personal data). Moreover, the E2 levels in snails in the Bourbre River were in line with the measured estrone concentrations in the River, pointing toward the hypothesis of bioaccumulation. Indeed waterborne estrone is metabolized in E2 and bioaccumulated in Mytilus edulis (Labadie et al., 2007). In the Saone River, only estriol was measured, and estriol is not susceptible to metabolization in E2 in mollusks. This could explain the lower increase of E2 levels at the downstream site compared to the upstream sites of the Saone River (150%) than those of the Bourbre River (290%). In the Ardiere River, no bioaccumulation of E2 was observed downstream of the effluent discharge and no steroids or estrogens were detected at the downstream site. Although very high levels of T and E2 were measured in the downstream sites compared to basal levels, the reproductive effects observed downstream from the effluent discharge are difficult to link to a toxic effect of steroid hormone bioaccumulation. The consequences of steroid bioaccumulation on reproduction remain to be assessed and raise the issue of long-term exposure to steroids and the effects of their bioaccumulation on natural populations. The decrease of E2 levels combined with the sharp increase of the T levels downstream of the effluent discharge in the Ardiere River which drains vineyards could be connected to a modulation of biosynthesis as observed with certain toxicants in gastropods subjected to exposure in the laboratory (Bettin et al., 1996; Gust et al., 2010a). Contrary to the other two rivers, where the E2/T ratio was comparable between upstream and downstream sites, for the snails exposed at the Ardiere sites, the ratio was 0.29 at the site upstream and 0.12 at those downstream on day 7, and 0.29 and 0.06 on day 21. These observations are consistent with the mode of action of certain fungicides that inhibit the aromatase and metabolization of T into E2. Tebuconazole is a fungicide also known for its inhibitory action on aromatase activity (Tro¨sken et al., 2004). The reduced fecundity downstream of the discharge could be partially explained by such a mechanism. Indeed, it was previously shown that a specific aromatase inhibitor (fadrozole) could inhibit reproduction (80% decrease of unshelled embryos compared to a control) of P. antipodarum and lead to a similar decrease of E2/T (Gust et al., 2010a). In our experiments reproduction impairment and total steroid level cannot be linked, as a decrease of reproduction was observed whatever the steroid levels downstream of the discharges. Nevertheless, steroid levels measured in snails showed different patterns of response when they were exposed in sites contaminated differently. Regarding these
caging experiments it appears that an increase of both E2 and T levels above their physiological values strongly indicates global organic pollution.
5.
Conclusion
Our results confirm that P. antipodarum can be a suitable organism for assessing the impact of point source pollution on freshwater systems, as it is tolerant to abiotic factor variations (temperature, conductivity, flow). Indeed P. antipodarum is (i) capable of surviving in a wide range of systems (lakes, rivers, brackish and estuarine waters), (ii) it is responsive to toxics, (iii) it is easily maintained as a viable stock population in the laboratory, and is capable of providing suitable specimens for analysis throughout the year, (iiii) and it feeds on detritusdecomposing bacteria that develop easily in submerged cages. Reproduction is a suitable endpoint for assessing the impact of point source pollution as fecundity can be monitored easily. Our exposure methodology allowed observing discriminative effects on snail fecundity between upstream and downstream sites. Hormone ratios appeared useful for detecting the exposure of snails to steroid biosynthesis pathway disruptors and steroid measurements appear to be indicative of organic pollution. However, more work is required to better characterise the natural variability of the endpoints used in field studies. The effects of biotic and abiotic parameters such as season of exposure, suspended matter, food, etc. on its life traits must also be assessed in greater detail.
Acknowledgments The present work was partly funded by the French water agency ONEMA (Office National de l’Eau et des Milieux Aquatiques). The authors wish to thank the Endocrinology Laboratory of the National Veterinary School of Lyon (Prof. F. Garnier) for the financial grant provided for RIA analysis. Solange Couturier and Claire Plantarose are gratefully acknowledged for their technical help as is Herve´ Pella for the maps. We also thank Clothilde Billat (Grand Lyon), Pierre-Yves Bigot and Stephane Faure (Lyonnaise des Eaux) for providing the WWTP effluent flows, Banque Hydro (http://www.hydro.eaufrance.fr/) and the Compagnie Nationale du Rhoˆne for providing the river flows.
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2010.06.019.
references
Aufderheide, J., Warbritton, R., Pounds, N., File-Emperador, S., Staples, C., Caspers, N., Forbes, V., 2006. Effects of husbandry parameters on the life-history traits of the apple snail, Marisa cornuarietis: effects of temperature, photoperiod, and population density. Invertebrate Biology 125, 9e20.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 1 7 e4 5 2 8
Bannister, R., Beresford, N., May, D., Routledge, E.J., Jobling, S., Rand-Weaver, M., 2007. Novel estrogen receptor-related transcripts in Marisa cornuarietis; a freshwater snail with reported sensitivity to estrogenic chemicals. Environmental Science and Technology 41, 2643e2650. Bettin, C., Oehlmann, J., Stroben, E., 1996. TBT-induced imposex in marine neogastropods is mediated by an increasing androgen level. Helgolander Meeresuntersuchungen 50, 299e317. Blaise, C., Gagne, F., Salazar, M., Salazar, S., Trottier, S., Hansen, P.D., 2003. Experimentally-induced feminisation of freshwater mussels after long-term exposure to a municipal effluent. Fresenius Environmental Bulletin 12, 865e870. Breitholtz, M., Ruden, C., Hansson, S.O., Bengtsson, B.E., 2006. Ten challenges for improved ecotoxicological testing in environmental risk assessment. Ecotoxicology and Environmental Safety 63, 324e335. Brown, L., 1980. The use of Hydrobia jenkinsi to detect intermittent toxic discharges to a river. Water Research 14, 941e947. Castro, L.F.C., Melo, C., Guillot, R., Mendes, I., Queiros, S., Lima, D., Reis-Henriques, M.A., Santos, M.M., 2007. The estrogen receptor of the gastropod Nucella lapillus: modulation following exposure to an estrogenic effluent? Aquatic Toxicology 84, 465e468. Chambre Agriculture Rhone, C.A., 2008. Protection des Eaux en Beaujolais viticole. Caracte´risation et Suivi de la Qualite´ de L’eau. Chambre de´partementale d’Agriculture, Rhone, 50 pp. Clarke, N., Routledge, E.J., Garner, A., Casey, D., Benstead, R., Walker, D., Watermann, B., Gnass, K., Thomsen, A., Jobling, S., 2009. Exposure to treated sewage effluent disrupts reproduction and development in the seasonally breeding ramshorn snail (subclass: Pulmonata, Planorbarius corneus). Environmental Science and Technology 43, 2092e2098. Dedourge-Geffard, O., Palais, F., Biagianti-Risbourg, S., Geffard, O., Geffard, A., 2009. Effects of metals on feeding rate and digestive enzymes on Gammarus fossarum: an in situ experiment. Chemosphere 77, 1569e1575. Duft, M., Schmitt, C., Bachmann, J., Brandelik, C., SchulteOehlmann, U., Oehlmann, J., 2007. Prosobranch snails as test organisms for the assessment of endocrine active chemicals e an overview and a guideline proposal for a reproduction test with the freshwater mudsnail Potamopyrgus antipodarum. Ecotoxicology 16, 169e182. Duft, M., Schulte-Oehlmann, U., Weltje, L., Tillmann, M., Oehlmann, J., 2003. Stimulated embryo production as a parameter of estrogenic exposure via sediments in the freshwater mudsnail Potamopyrgus antipodarum. Aquatic Toxicology 64, 437e449. Gagne, F., Blaise, C., Aoyama, I., Luo, R., Gagnon, C., Couillard, Y., Campbell, P., Salazar, M., 2002. Biomarker study of a municipal effluent dispersion plume in two species of freshwater mussels. Environmental Toxicology 17, 149e159. Gagne´, F., Blaise, C., Hellou, J., 2004. Endocrine disruption and health effects of caged mussels, Elliptio complanata, placed downstream from a primary-treated municipal effluent plume for 1 year. Comparative Biochemistry and Physiology Part C: Toxicology and Pharmacology 138, 33e44. Gagne´, F., Blaise, C., Pellerin, J., Andre´, C., 2007. Neuroendocrine disruption in Mya arenaria clams during gametogenesis at sites under pollution stress. Marine Environmental Research 64, 87e107. Gagne´, F., Blaise, C., Salazar, M., Salazar, S., Hansen, P.D., 2001. Evaluation of estrogenic effects of municipal effluents to the freshwater mussel Elliptio complanata. Comparative Biochemistry and Physiology Part C: Toxicology and Pharmacology 128, 213e225. Gooding, M.P., LeBlanc, G.A., 2001. Biotransformation and disposition of testosterone in the eastern mud snail Ilyanassa
4527
obsoleta. General and Comparative Endocrinology 122, 172e180. Gust, M., Buronfosse, T., Giamberini, L., Ramil, M., Mons, R., Garric, J., 2009. Effects of fluoxetine on the reproduction of two prosobranch mollusks: Potamopyrgus antipodarum and Valvata piscinalis. Environmental Pollution 157, 423e429. Gust, M., Garric, J., Giamberini, L., Mons, R., Abbaci, K., Garnier, F., Buronfosse, T., 2010a. Sensitivity of gastropod Potamopyrgus antipodarum to a specific aromatase inhibitor. Chemosphere 79, 47e53. Gust, M., Vulliet, E., Giroud, B., Garnier, F., Couturier, S., Garric, J., Buronfosse, T., 2010b. Development, validation and comparison of LC-MS/MS and RIA methods for quantification of vertebrate-like sex-steroids in prosobranch molluscs. Journal of Chromatography B. doi:10.1016/j. jchromb.2010.03.046. Jacquet, R., Miege, C., Soulier, C., Budzinski, H., Coquery, M., 2008. Comparison of POCIS and SPMD for In Situ Sampling of Moderately Hydrophobic Molecules. European Meeting on Environmental Chemistry, Girona. Janer, G., Lyssimachou, A., Bachmann, J., Oehlmann, J., SchulteOehlmann, U., Porte, C., 2006. Sexual dimorphism in esterified steroid levels in the gastropod Marisa cornuarietis: the effect of xenoandrogenic compounds. Steroids 71, 435e444. Janer, G., Porte, C., 2007. Sex steroids and potential mechanisms of non-genomic endocrine disruption in invertebrates. Ecotoxicology 16, 145e160. Jobling, S., Casey, D., Rodgers-Gray, T., Oehlmann, J., SchulteOehlmann, U., Pawlowski, S., Baunbeck, T., Turner, A.P., Tyler, C.R., 2003. Comparative responses of molluscs and fish to environmental estrogens and an estrogenic effluent. Aquatic Toxicology 65, 205e220. Keay, J., Bridgham, J.T., Thornton, J.W., 2006. The Octopus vulgaris estrogen receptor is a constitutive transcriptional activator: evolutionary and functional implications. Endocrinology 147, 3861e3869. Kolodziej, E.P., Gray, J.L., Sedlak, D.L., 2003. Quantification of steroid hormones with pheromonal properties in municipal wastewater effluent. Environmental Toxicology and Chemistry 22, 2622e2629. Kolpin, D.W., Furlong, E.T., Meyer, M.T., Thurman, E.M., Zaugg, S.D., Barber, L.B., Buston, H.T., 2002. Pharmaceuticals, hormones and other organic wastewater contaminants in U.S. streams, 1999e2000: a national reconnaissance. Environmental Science and Technology 36, 1202e1211. Lafont, R., Mathieu, M., 2007. Steroids in aquatic invertebrates. Ecotoxicology 16, 109e130. Labadie, P., Peck, M., Minier, C., Hill, E.M., 2007. Identification of the steroid fatty acid ester conjugates formed in vivo in Mytilus edulis as a result of exposure to estrogens. Steroids 72, 41e49. Lassen, H.H., Kristensen, J.H., 1978. Tolerance to abiotic factors in mudsnails (Hydrobiidae). Natura Jutlandica 20, 243e250. Lydeard, C., Cowie, R.H., Ponder, W.F., Bogan, A.E., Bouchet, P., Clark, S.A., Cummings, K.S., Frest, T.J., Gargominy, O., Herbert, D.G., Hershler, R., Perez, K.E., Roth, B., Seddon, M., Strong, E.E., Thompson, F.G., 2004. The global decline of nonmarine mollusks. Bioscience 54, 321e330. Matthiessen, P., 2008. An assessment of endocrine disruption in mollusks and the potential for developing internationally standardized mollusk life cycle test guidelines. Integrated Environmental Assessment and Management 4 ISSN 1551e3777(print)j1551e3793(electronic). Oehlmann, J., Schulte-Oehlmann, U., Oetken, M., Bachmann, J., Lutz, I., Kloas, W., Ternes, T.A., 2006. Bisphenol A induces superfeminization in the ramshorn snail Marisa cornuarietis (Gasteropoda: Prosobranchia) at environmentally relevant
4528
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 1 7 e4 5 2 8
concentrations. Environmental Health Perspectives 114, 127e133. Quinn, B., Gagne´, F., Costello, M., McKenzie, C., Wilson, J., Mothersill, C., 2004. The endocrine disrupting effect of municipal effluent on the zebra mussel (Dreissena polymorpha). Aquatic Toxicology 66, 279e292. Regoli, F., Gorbi, S., Fattorini, D., Tedesco, S., Notti, A., Machella, N., Bocchetti, R., Benedetti, M., Piva, F., 2006. Use of the land snail Helix aspersa as sentinel organism for monitoring ecotoxicologic effects of urban pollution: an integrated approach. Environmental Health Perspectives 114, 63e69. Selck, H., Aufderheide, J., Pounds, N., Staples, C., Caspers, N., Forbes, V., 2006. Effects of food type, feeding frequency, and temperature on juvenile survival and growth of Marisa cornuarietis (Mollusca: Gastropoda). Invertebrate Biology 125, 106e116.
Sternberg, R.M., Hotchkiss, A.K., LeBlanc, G.A., 2008. The contribution of steroidal androgens and estrogens to reproductive maturation of the eastern mud snail Ilyanassa obsoleta. General and Comparative Endocrinology 156, 15e26. Tro¨sken, E.R., Scholz, K., Lutz, R.W., Vo¨lkel, W., Zarn, J.A., Lutz, W. K., 2004. Comparative assessment of the inhibition of recombinant human CYP19 (aromatase) by azoles used in agriculture and as drugs for humans. Endocrine Research 30, 387e394. Watton, A.J., Hawkes, H.A., 1984. Studies on the effects of sewage effluent on gastropod populations in experimental streams. Water Research 18, 1235e1247. Winterbourn, M., 1970. The New Zealand species of Potamopyrgus (Gasteropoda: Hydrobiidae). Malacologia 10, 283e321.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 5 9 e4 3 7 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Microbial community distribution and activity dynamics of granular biomass in a CANON reactor Jose Va´zquez-Padı´n a,*, Anuska Mosquera-Corral a, Jose Luis Campos a, Ramo´n Me´ndez a, Niels Peter Revsbech b a b
Department of Chemical Engineering, University of Santiago de Compostela, Lope Go´mez de Marzoa, s/n, E-15782, Spain Department of Microbiology, University of Aarhus, Ny Munkegade, Bldg. 540, 8000 Aarhus C, Denmark
article info
abstract
Article history:
The application of microelectrodes to measure oxygen and nitrite concentrations inside
Received 23 October 2009
granules operated at 20 C in a CANON (Complete Autotrophic Nitrogen-removal Over
Received in revised form
Nitrite) reactor and the application of the FISH (Fluorescent In Situ Hybridization) tech-
14 May 2010
nique to cryosectioned slices of these granules showed the presence of two differentiated
Accepted 27 May 2010
zones inside of them: an external nitrification zone and an internal anammox zone. The
Available online 4 June 2010
FISH analysis of these layers allowed the identification of Nitrosomonas spp. and Candidatus Kuenenia Stutgartiensis as the main populations carrying out aerobic and anaerobic
Keywords:
ammonia oxidation, respectively.
Anammox
Concentration microprofiles measured at different oxygen concentrations in the bulk
CANON
liquid (from 1.5 to 35.2 mg O2 L1) revealed that oxygen was consumed in a surface layer of
Granule
100e350 mm width. The obtained consumption rate of the most active layers was of
Autotrophic nitrogen removal
80 g O2 (Lgranule)1 d1. Anammox activity was registered between 400 and 1000 mm depth
Microbial community
inside the granules. The nitrogen removal capacity of the studied sequencing batch reactor
Microscale distribution
containing the granular biomass was of 0.5 g N L1 d1. This value is similar to the mean nitrogen removal rate obtained from calculations based on in- and outflow concentrations. Information obtained in the present work allowed the establishment of a simple control strategy based on the measurements of NHþ 4 and NO2 in the bulk liquid and acting over the
dissolved oxygen concentration in the bulk liquid and the hydraulic retention time of the reactor. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Efficient nutrients removal from wastewater is essential due to disposal legislation demanding stricter effluent concentrations. Nitrogen removal from wastewaters characterized by low organic matter content and high nitrogen concentrations is difficult by the application of conventional processes like nitrification-denitrification. In these cases, the anammox process arose as an interesting alternative since the anammox
bacteria oxidize ammonium anaerobically with nitrite as electron acceptor in the absence of organic carbon compounds. Prior to the anammox reaction, part of the ammonium has to be oxidized into nitrite. This step can be carried out separately in different kind of reactors, e.g., a Sharon (Single reactor system for High activity Ammonium Removal Over Nitrite) reactor (van Dongen et al., 2001; Mosquera-Corral et al., 2005), a granular nitrifying reactor (Va´zquez-Padı´n et al., 2009), etc. Another possibility is to
* Corresponding author. Tel.: þ34 981 563100x16739; fax: þ34 981 528050. E-mail address:
[email protected] (J. Va´zquez-Padı´n). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.05.041
4360
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 5 9 e4 3 7 0
perform the partial nitrification and the anammox processes in one single reactor. This process has been given different names: CANON (Complete Autotrophic Nitrogen-removal Over Nitrite; Third et al., 2001), OLAND (Oxygen Limited Autotrophic Nitrification-Denitrification; Kuai and Verstraete, 1998) and deammonification (Hippen et al., 1997) processes. Under microaerobic conditions, ammonia oxidizing bacteria (AOB) oxidize ammonium into nitrite consuming the dissolved oxygen (DO) and creating anoxic niches where anammox bacteria can exist and convert both ammonia and nitrite into nitrogen gas and produce small amounts of nitrate. The optimization of the performance of the autotrophic nitrogen removal in the CANON process requires the control of dissolved oxygen (DO) and NO 2 concentrations. To establish an adequate control of the DO concentration in the liquid media is necessary: 1) to avoid the inhibition of anammox bacteria caused by DO concentrations higher than 0.5% of air saturation (Strous et al., 1997), 2) to prevent the growth of nitrite oxidizing bacteria (NOB) which have lower affinity for oxygen compared to AOB. The control of NO 2 concentration is necessary since this compound inhibits anammox activity although variable ranges of concentrations are provided in the literature for this inhibitory effect. Dapena-Mora et al. (2007) 1 reported that concentrations of nitrite of 350 mg NO 2 eN L resulted in 50% inhibition of anammox bacteria. Due to the slow growth of both nitrifying and anammox bacteria involved in the CANON process, the use of good biomass retention systems is mandatory to reach significant nitrogen removal rates. In this sense, the development of granular biomass allows the accumulation of large biomass concentrations in the reactors without the need of carrier material. Moreover, the use of granular biomass allows the existence of substrate gradients, in such a way that the external part of granule can be under aerobic conditions while anoxic conditions are maintained in the core of the granule. Therefore, different biological processes can be carried out in the same granule: partial nitrification in the outer part and anammox in the inner part. The potential of the technologies based on single-reactors to carry out autotrophic nitrogen removal using granular biomass has been demonstrated (Vlaeminck et al., 2008; Va´zquez-Padı´n et al., 2009). Using these systems it is possible to treat nitrogen loads similar to systems with two different units for partial nitrification and anammox processes, respectively. Microsensors, due to their very small dimensions, can be used for the determination of substrate profiles while the distribution of bacterial populations can be determined by microbiological techniques (e.g. FISH, PCR). The combination of microbiological techniques and measurements using microelectrodes has previously been used to obtain detailed knowledge about the in situ structure and function of nitrifying biofilms. Schramm et al. (1999) and Kindaichi et al. (2006) used this combination to estimate kinetic parameters to be further used in mathematical models. de Beer et al. (1993) and Gieseke et al. (2003) studied the mass transport of substrates through biofilms or aggregates using microelectrodes and determined the limiting substrate, the size of the active zone of the biofilm and the biomass activities under different substrate concentrations. These techniques are also suitable for the determination of the spatial distribution of substrate
consumption rates in non- homogeneous biofilm reactors (Schramm et al., 1999; Kindaichi et al., 2007), for researching the environmental conditions (e.g., pH) inside the biofilm (Gieseke et al., 2006) or for observing the stratification of active biomass (Okabe et al., 1999; Kindaichi et al., 2006). The combination of microsensor measurements and FISH analysis makes it possible to gain information about the substrate concentrations to which the different layers of a biofilm/aggregate are exposed and about the microbial populations involved in the different biological processes. All the microscopic information could then be transposed to a macroscopic level, giving valuable information about the possible control strategies of CANON reactors. Since little information is known from a microscopic point of view about granules performing complete autotrophic nitrogen removal and taking into account the relevance of DO and NO 2 concentrations, the objectives of this study were: the identification of the main bacteria populations present in the granules, the determination of their distribution inside the granules and the estimation of their activities by combining the concentrations profiles measured with microelectrodes and FISH images taken from cryosectioned slices of granules. The improved insight into the functioning of the CANON aggregates should then be used to devise a control strategy for the optimization of the reactor performance.
2.
Materials and methods
2.1.
Reactor description
The growth of biomass in the form of granules performing the CANON process was described elsewhere (Va´zquez-Padı´n et al., 2009). Initially, the complete nitrification was developed and later, after the regulation of the DO concentration in the bulk liquid, the partial nitrification to nitrite was achieved. Finally, anammox bacteria were grown in the anoxic core of the granules to form the CANON granules. Five months after the appearance of the macroscopic evidence of anammox activity in the CANON granules, 4 g VSS of granular biomass from an anammox reactor operated at the University of Santiago de Compostela were inoculated into a sequencing batch reactor (SBR) with a working volume of 0.7 L at the University of Aarhus. This new reactor was operated in cycles of 3 h distributed as: 175 min of aeration and feeding, 1 min of settling and 4 min for effluent withdrawal. The hydraulic retention time (HRT) was fixed at 0.45 d and the exchange volume was fixed at 30%. The reactor was operated at room temperature which ranged between 19 and 22 C. The pH value was not controlled and ranged from 7.0 to 8.1 with a mean value of 7.6. Air was supplied through a diffuser at the bottom of the reactor to promote the transfer of oxygen into the bulk liquid and to reach a suitable mixing. The average DO concentration in the reactor was 6.6 mg O2 L1 and this represented the main difference as compared to the operational conditions of the reactor operated at the laboratory in Santiago de Compostela where the DO was kept around 3.5 mg O2 L1. The CANON SBR was fed with the supernatant from the anaerobic sludge digester of the WWTP of Aarhus (Denmark)
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 5 9 e4 3 7 0
which was stored in a cold room (4 C). The main properties of the supernatant were: pH of 7.6e8.3; ammonium concentration of 256e666 mg N L1; inorganic carbon (IC) of 232e374 mg IC L1 and total organic carbon (TOC) of 103e150 mg TOC L1 which was poorly biodegradable.
2.2.
kLa measurement
An experimental estimation of the oxygen gas-liquid transfer coefficient (kLa) was carried out by means of a dynamic method, registering the increments of DO concentrations in the SBR after the reestablishment of the aeration (without biomass in the reactor). The value of the kLa for dissolved oxygen obtained in the reactor was of 1 min1 which was similar value to the value obtained in the case of the SBR installed in Santiago de Compostela.
2.3.
Microscale experiments
Microsensors were used to measure the concentration profiles of nitrite and dissolved oxygen (DO) inside the granules performing the CANON process. The granules were collected directly from the CANON SBR and fastened with a needle to a metal grid inside the experimental chamber. In order to simulate the hydrodynamic conditions from the reactor, the aeration flow in the chamber was regulated to maintain a kLa value of 1 min1 which was the same value as in the SBR. All the microprofiles were measured in granules with average diameters of 5 mm. The mean temperature of the aerated chamber was 20 1 C. Granules were kept for 1 h inside the chamber as a pre-incubation period to create pseudo steady state conditions. Concentration profiles were recorded by introducing the sensors into the granules at different depth positions using a manual micromanipulator. A dissection microscope was used to visually estimate the position of the granule/water interface by visual observation. For each granule and experimental condition tested several microprofiles were measured (the number of microprofiles performed is indicated in the caption of figures as n). Microprofiles of DO were performed by measuring its concentration at depth intervals of 25 mm while in the case of the microprofiles of NO 2 the measurements were performed at 50 or 100 mm intervals due to a slower response time of this sensor. The sensor signal was continuously recorded on a strip-chart recorder. The liquid medium inside the chamber was the anaerobic digester supernatant diluted with tap water. The concentration of ammonium was in all experiments maintained constant at 140 mg N L1 in order to avoid ammonium limitation. The nitrite concentrations were varied from 0.7 to 1 by NaNO2 addition. The microsensors were 42 mg NO 2 eN L calibrated prior to each experiment using the same basic medium with various added nitrite concentrations. The dissolved oxygen concentration in the bulk liquid was varied between 1.5 and 35.2 mg O2 L1. These different DO concentrations were achieved by flushing a mixture of pure O2 and air for DO values in the aeration chamber above air saturation, or a mixture of air and N2 for DO values under air saturation. In order to maintain a constant kLa, in all the experiments the total gas flow (N2/air/O2) was kept constant.
2.4.
4361
NO 2 and O2 microsensors
Microsensors were used to determine concentrations of the measured compounds inside the granules. A Clark-type O2 microsensor equipped with a guard cathode was used for microscale analysis of DO (Revsbech, 1989). The sensor was made with a tip diameter of 10 mm and it had a 90% response time lower than 2 s. The oxygen microsensor was calibrated at two concentration points: the zero value was established by inserting the microsensor into an anoxic alkaline ascorbic acid solution and the other calibration point was obtained by means of an air saturated solution or a pure oxygen saturated solution, depending on the DO range to be tested. The NO 2 microsensor was a biosensor containing an immobilized pure culture of Stenotrophomonas nitritireducens that reduced NO 2 eN2O, which was quantified by a built-in Clark-type N2O microsensor (Nielsen et al., 2005). The 90% response time was 30e50 s. Due to the broad range of NO 2 concentrations in the bulk liquid varying from 0.7 to 42 mg 1 it was necessary to use Electrophoretic Sensitivity NO 2 eN L Control (ESC) to adjust the sensitivity of the sensor to the relevant concentration range (Kjær et al., 1999).
2.5.
Analytical methods
The pH and the concentrations of DO, ammonium, nitrite, nitrate, volatile suspended solids (VSS), settling velocity and sludge volumetric index (SVI) were determined according to the Standard Methods (APHA-AWWA-WPCF, 1998). Concentrations of TOC and IC were measured with a Shimadzu analyser (TOC-5000). Density of the granules was measured using the dextran blue method described by Beun et al. (2001). The morphology and size distribution of the granules were measured regularly by using an image analysis procedure with a stereomicroscope (Stemi 2000-C, Zeiss) provided with a digital camera (Coolsnap, Roper Sicientific Photometrics). For the digital image analysis the programme Image Pro Plus was used.
2.6.
Description of the FISH protocol
In order to identify bacterial populations of AOB, NOB and anammox bacteria, granules from the reactor were collected, kept in their aggregated form or disaggregated, and fixed according to Amann et al. (1995) with 4% paraformaldehyde solution. Entire granules were embedded in OCT reagent (Tissue-Tek; Miles, Ind.) prior to their cryosectioning at 35 C. Slides with a thickness of 14 mm were cut at 16 C and each single section was placed on the surface of poly-L-lysine coated microscope slides. Hybridization was performed at 46 C for 90 minutes adjusting formamide concentrations at the percentages shown in Table 1. The used probes for in situ hybridization were 50 labelled with the fluorochromes FITC, Cy3 or Cy5. A TCS-SP2 confocal laser scanning microscope (Leica, Germany), equipped with a HeNe laser for detection of Cy3 and Cy5 and one Ar ion laser for detection of FITC, was used with the sliced samples. The method to quantify bacterial populations was based on the one published by Crocetti et al. (2002). The digital image
4362
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 5 9 e4 3 7 0
Table 1 e Targeted organisms and the corresponding formamide (FA) percentages for the used oligonucleotide probes. Probea
Probe sequence (50 /30 )
EUB338 EUB338 II EUB338 III Nso190 NEU653b
Ntspa712b NIT3
b
AMX820
Kst157 Ban162
% FA
Targeted organisms
GCT GCC TCC CGT AGG AGT GCA GCC ACC CGT AGG TGT GCT GCC ACC CGT AGG TGT CGA TCC CCT GCT TTT CTC C CCC CTC TGC TGC ACT CTA
0e50
Domain bacteria
0e50
Planctomycetales
0e50
Verrucomicrobiales
55
CGC CTT CGC CAC CGG CCT TCC CCT GTG CTC CAT GCT CCG AAA ACC CCT CTA CTT AGT GCC C
50
Ammonia-oxidizing b-Proteobacteria Most of the halophilic and halotolerant Nitrosomonas spp. Most members of the phylum Nitrospira Nitrobacter spp.
GTT CCG ATT GCT CGA AAC CGG TAG CCC CAA TTG CTT
40
40 40
25 40
Anaerobic ammonium-oxidizing bacteria Candidatus Brocadia anammoxidans and Candidatus Kuenenia stuttgartiensis Candidatus Kuenenia stuttgartiensis Candidatus Brocadia anammoxidans
a Details on oligonucleotide probes are available at probeBase (Loy et al., 2007). b Used with an equimolar amount of corresponding unlabeled competitor oligonucleotide probe.
analysis program used was Image ProPlus. For the quantification of bacteria populations, cryosectioned slices of granules were used to perform a triple hybridization with EUBmix (a mixture of EUB338, EUB338 II and EUB338 III) labelled with Cy5, AMx820 labelled with Cy3 and NEU653 labelled with FITC. Several pictures were taken subsequently from the granule surface throughout the active layers with the confocal microscope at a magnification of 630 times. From each image three different colour components corresponding to each fluorochrome were separated generating three different images. The area corresponding to the fluorescence of each FISH probe was obtained as the area of all pixels above one value manually determined corresponding to the minimum pixel intensity. In order to be able to compare the areas occupied by the different populations, the maximal area occupied in a discretized picture was taken as reference and the values of fluorescence obtained with the different probes in the different layers were obtained as normalized area values.
3.
Calculations
3.1. Estimation of oxygen and nitrite consumption and production rates Dissolved oxygen and nitrite mass balances were calculated for the granules using a one-dimensional diffusion model. The
activities of the nitrogen removal processes occurring inside the granules were restricted to an external layer of 1 mm (as it will be further discussed) which was smaller than the mean radius of the granules used to record the microprofiles (mean radius of 2.5 mm). In this calculation procedure the assumption was made that the diffusional transport into the granules can be modelled as a one-dimensional transport, as the diameter of granules was large (5 mm) compared to the studied surface layer (1 mm). The concentration changes with time for a diffusible substance in a matrix with a flat geometry can be written according to Eq. 1. It is assumed that the compound mass transfer is carried out only by diffusion and that the granule has a homogeneous structure so that the diffusion coefficient (D) can be considered constant (Lorenzen et al., 1998). vCðz; tÞ v2 Cðz; tÞ þ AðzÞ ¼D vt vz2
(1)
where C is the concentration of the compound (g L1), z the depth coordinate in the granule (dm), t the time (d ) and A the reaction rate (g (Lgranule)1 d1). Assuming steady state conditions Eq. 2 is obtained. D
v2 CðzÞ ¼ AðzÞ vz2
(2)
using Euler’s formula for numeric integration, Eq. 3 is obtained. vC vC An ¼ þh D vznþ1 vzn
(3)
where h represents the step size used for numerical integration. This discretization parameter was of 25 mm for the oxygen profiles and of 50 mm for nitrite profiles with concen1 and trations in the bulk liquid smaller than 2.8 mg NO 2 eN L concentrations. After further inteof 100 mm for higher NO 2 gration Eq. 4 is obtained. Cnþ1 ¼ Cn þ h
vC vzn
(4)
Substituting Eq. 3 in Eq. 4, Eq. 5 is obtained. Cnþ1 ¼ Cn þ h
vC An1 þh D vzn1
(5)
Using the Solver tool (available in Microsoft Excel software) the values of A (i.e., the local volumetric consumption rate) were iterated in order to minimize the error between the concentration calculated with Eq. 2 and that one measured with the microsensor. The values of the diffusion coefficients 4 and of NO 2 and O2 in water at 20 C were chosen as 1.5 10 4 2 1 1.7 10 m d , respectively (Picioreanu et al., 1997). To calculate the fluxes of substrates (J, g N m2 d1) through the diffusive boundary layer (DBL) which separates the surface of the granule and the bulk liquid, Fick’s first law of diffusion was used (Eq. 6). J ¼ DW
Cb Cs dh
(6)
being Dw the molecular diffusion coefficient in water (m2 d1), Cb is the bulk liquid concentration (g m3), CS is the concentration (g m3) at the surface of the granule, and dh is the
4363
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 5 9 e4 3 7 0
3.2.
Nitrogen removal rates
Ammonia oxidation rates (AOR) and nitrogen removal rates by anammox bacteria (ANR) of the CANON granular reactor were estimated as g N L1 d1 based on nitrogen balances and the stoichiometry of the anammox process (1.02 moles of dinitrogen gas produced per mole of ammonium reacted). NHþ DN ¼ NHþ 4 Ninf 4 Neff þ NO2 Neff þ NO 3 Neff DN þ NHþ 4 Ninf NH4 Neff 2:04 AOR ¼ HRT
(7)
ANR ¼
DN HRT
(8)
(9)
Where DN is the difference between total nitrogen concentration in the influent and effluent (g N L1), NHþ 4 eNinf is the ammonium concentration in the influent (g N L1) and NHþ 4 eNeff, NO2 eNeff, NO3 eNeff are the ammonium, nitrite and nitrate concentrations in the effluent (g N L1), respectively.
3.3.
Estimation of the number of granules
The number of granules in the reactor was calculated as follows: Vgranule ¼ nT ¼
VR $XR rgranule
(10)
Vgranule 4 $p$R3m 3
(11)
Where Vgranule is the volume of granules (L), VR is the reactor volume (L), XR is the biomass concentration in the reactor (g VSS L1), rgranule is the granules density (g VSS (Lgranule)1), nT is the number of granules, and Rm is the average radius of the granules (dm). In order to estimate either the AOR or the ANR from the microscopic observations, Eq. 12 was used to describe the zone were AOB or anammox bacteria were located.
Rate ¼
z¼1
4 AðzÞ$ $p$ R3z R3z1 3
VR
(12)
Where the rate is calculated in terms of g (Lreactor)1 d1, A is the local reaction rate (g (Lgranule)1 d1), VR is the reactor volume (L), and r corresponds to the granule radius (dm).
4.
Results and discussion
4.1.
Operation of the CANON SBR
Identification of bacteria populations by FISH
The stratification of the AOB and anammox bacteria in depth inside the granule can be observed in Fig. 2. In the outermost 200 mm layer of the granule almost all the biomass consisted of Nitrosomonas spp. which gave positive signals to probes NEU653 and Nso190. Bacteria belonging to the genus Nitrosomonas spp. were still present at a depth of 600 mm but their proportion decreased when the depth increased (Fig. 2). Nitrosomonas spp. was identified as the main AOB community as it was expected due to operational conditions in the reactor with ammonium concentration in excess (Schramm et al., 1998). Significant fluorescence signal was detected neither with probe NIT3 specific for Nitrobacter spp. nor with probe Ntspa712 specific for the Nitrospira phylum. Therefore, no NOB activity was expected in the granules despite the high DO concentration in the liquid bulk. Anammox bacteria were mainly located between 400 and 1000 mm in depth inside the granule where dissolved oxygen was absent during normal reactor operation. Bacteria belonging to the genus Candidatus Kuenenia stuttgartiensis were identified as the main anammox bacteria in the reactor through positive results with probe Kst157. No positive results with probe Ban162 were obtained indicating absence of Candidatus Brocardia anammoxidans. Thus, in the depth interval between 400 and 600 mm the AOB and anammox bacteria coexisted.
-1
n X
4.2.
1.5
-1
nT $
concentration in the effluent was 25 mg N L1. The total nitrogen removal rate (Fig. 1) ranged between 0.35 and 0.91 g N L1 d1. Those values are among the highest ones registered for autotrophic nitrogen removal in one reactor despite the low temperature of operation. According to Dosta et al. (2008) there is a strong dependence between the temperature and the anammox activity in such a way that the specific activity of anammox biomass at 37 C is 3.6 times higher than at 20 C. The biomass concentration inside the reactor remained almost constant at 7.5 g VSS L1 during the 60 days of operation. The average diameter and density of the granules were 5 mm and 36 g VSS (Lgranule)1, respectively. With these data, the number of granules was estimated, using Eq. 11, as 2230 granules. The value of the SVI was 25 mL (g VSS)1 and the settling velocity of the granular sludge was of 110 m h1.
NLR, AOR, ANR (g N L d )
hypothetical (also called effective) thickness of the diffusive boundary layer (m) which is defined by extrapolating the radial oxygen gradient at the granule-water interface to the bulk water phase concentration (Ploug et al., 1997).
The SBR reactor was operated at DO concentrations around 6.6 mg O2 L1 and at a temperature of 20 C. The mean NO 2
1.2 0.9 0.6 0.3 0.0 0
10
20
30 40 Time (d)
50
Fig. 1 e Applied nitrogen loading rate (e), ammonia oxidation rate (B), nitrogen removal rate ( ).
60
4364
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 5 9 e4 3 7 0
Fig. 2 e a) Depth distribution of AOB populations (hybridized with probe NEU653), anammox bacteria (hybridized with probe AMX820) and all bacteria (hybridized with EUBmix) inside the granule. The value of depth equal to 0 mm corresponds to the granule surface. b) Image of a cryosectioned slice of a granule with a triple hybridization of FISH probes targeting: AOB (probe: NEU653; fluorochrome: FITC; colour: light green); anammox bacteria (probe: AMX820; fluorochrome: Cy3; colour: pink) and all bacteria (probe: EUBmix; fluorochrome: Cy5; colour: blue). The right part of the picture corresponds to the surface of the granule (the bar corresponds to 75 mm).
It is also interesting to point out that the area corresponding to EUBmix probe which represent all bacteria decreased with the increase in depth inside the granule and that the activity was mainly located in the external 1000 mm of the granule.
4.3.
Microprofiles measurements
4.3.1.
Oxygen microprofiles in the partial nitrification zone
In order to determine the oxygen consumption kinetics, microprofiles were measured along the granule varying the DO concentration in the bulk liquid over a wide range, from 1.5 to 35.2 mg O2 L1 (Fig. 3) with an initial ammonium concentration of 140 mg N L1. The trend of the concentration profiles was similar in all cases illustrating the various steps of oxygen transport and consumption: first the diffusion of DO through the external diffusive boundary layer (DBL) and later the internal diffusion together with the biological reaction which resulted in a curve profile. The external mass transfer resistance was significant as a large decrease of DO concentration could be observed between the bulk liquid and the granule surface. This significant decrease of DO concentration within the DBL working with highly active biofilms has also been reported by other authors (Jørgensen and Revsbech, 1985; de Beer et al., 1993; Rasmussen and Lewandowski, 1998; Wilen et al., 2004). This demonstrates that the external mass transfer resistance plays an important role specially when working with granules with a mean diameter of 5 mm. In the present study the width of the external DBL was around 100e120 mm.
Regarding the DO microprofiles inside the granules, it was observed that the oxygen penetration depth increased from 100 to 350 mm when increasing the DO in the bulk liquid from 1.5 to 35.2 mg O2 L1 (Fig. 3). The microprofiles revealed that even when the DO concentration in the bulk liquid was kept at 8 mg O2 L1 (close to 100% of air saturation), the maximal oxygen consumption rate was only attained in the outer part of the granule (corresponding to a depth of around 30 mm). Moreover, a fast decrease in the DO concentration in depth inside the granule was registered in all the cases demonstrating that the oxygen mass transfer rate strongly limited the ammonia oxidation process. The high oxygen demand in the surface layer allowed an anoxic zone to be created in the internal part of the granule where the anammox process could take place. An estimation of the affinity constant of the AOB was performed using the profiles obtained with the highest DO concentrations in the bulk liquid. The value obtained was 0.6 mg O2 L1, which is in agreement with the values previously published for AOB at 20 C (Wiesmann, 1994; Guisasola et al., 2005). According to Harremoes and Henze (2002) the reaction rate can be modelled as a function of the concentration of substrate outside a biofilm using 3 kinetic orders: 1; 1/2 and 0. A kinetic order of 1 is a good approximation when the substrate concentrations in the bulk liquid are lower than 2$KS (KS is the half saturation constant for the substrate S), and a kinetic order of 0 is obtained when the biofilm is fully penetrated by the substrate (assuming a low Km value). Half order kinetics represents the transition between 1 and 0 order and it is caused by the progressively deeper substrate penetration into the biofilm with the increasing bulk concentration. In Fig. 4, the AOR is represented versus the square root of DO concentration at the surface of the granules. The AOR in the SBR was estimated using Eq. 12 considering the stoichiometry requirements for ammonium oxidation: 3.4 g O2 1 (g NHþ 4 eN) . As it was expected according to Harremoes and Henze (2002), the tendency of these values is linear, i.e. 1/2 order kinetic, and even when working with pure oxygen in the bulk liquid, the granule was oxygen limited. Wilen et al. (2004) estimated that the DO required to obtain a 0 order kinetic was higher than 20 mg O2 L1 in aggregates with a diameter of 0.50e0.69 mm, and it is then expected that the experiments performed with much larger aggregates exhibited half order kinetics These results illustrate the key role of the mass transfer limitations on the overall biomass activity of the granular sludge. The reactor was usually operated at DO concentrations of 6.6 mg O2 L1 meaning that the oxygen is available only in the first 200 mm of the granule. When the DO concentration in the bulk liquid was increased up to pure DO saturation, progressively deeper layers became active which indicates that nitrifying bacteria might be able to remain alive in the absence of substrates and very quickly be activated as soon as NHþ 4 and DO are supplied. Such long term survival of the nitrifying bacteria under ammonium-starving or anoxic conditions has previously been reported (Wilhelm et al., 1998; Freitag and Prosser, 2003). Oxygen profiles along the nitrifying granules/biofilms reported in literature are shown in Table 2. The different
4365
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 5 9 e4 3 7 0
Oxygen (mg O2 L-1)
Oxygen (mg O2 L-1) 0.5
1.0
1.5
2.0
0.0
1.2
1.8
2.4
0.3
0.2
0.2
0.2
0.1
0.1
0.1
0.0 -0.1
-0.3 -0.4
Depth (mm)
0.3
-0.2
0.0 -0.1
25
50
75
-0.3
-0.3
2.5
5.0
7.5
25
50
75
4.5
6.0
0
25
50
75
100
100 -1
-1
DO cons. rate (g O2 (Lgranule) d )
Oxygen (mg O2 L-1) 0.0
3.0
-0.4 0
DO cons. rate (g O2 (Lgranule)-1 d-1)
1.5
-0.1 -0.2
100
0.0
0.0
-0.2
-0.4
0
DO cons. rate (g O2 (Lgranule)-1 d-1)
Oxygen (mg O2 L-1) 10.0
0
6
12
18
Oxygen (mg O2 L-1) 24
0.3
0.3
0.2
0.2
0.2
0.1
0.1
0.1
0.0 -0.1 -0.2 -0.3 -0.4 0
25
50
75
Depth (mm)
0.3
Depth (mm)
Depth (mm)
0.6
0.3
Depth (mm)
Depth (mm)
0.0
Oxygen (mg O2 L-1)
0.0 -0.1
20
30
40
0
25
50
75
100
-0.1 -0.2
-0.3
-0.3 -0.4 0
DO cons. rate (g O2 (Lgranule)-1 d-1)
10
0.0
-0.2
-0.4
100
0
25
50
75
100
DO cons. rate (g O2 (Lgranule)-1 d-1)
DO cons. rate (g O2 (Lgranule)-1 d-1)
Fig. 3 e Dissolved oxygen concentration profiles (C, bars indicate standard deviation) and local consumption rates ( ) under different DO concentrations in the bulk liquid (number of microprofiles performed (n): n [ 3 for all cases except for DO [ 8 mg O2 LL1 where n [ 20). Note the different concentration scales.
-1
-1
AOR (g N Lrea ctor d )
biofilms or granules analyzed came from different reactors with different hydrodynamics characteristics: rotating disk reactors (Okabe et al., 1999; Kindaichi et al., 2006); fluidized bed reactors (de Beer et al., 1993; Schramm et al., 1999); sequencing batch biofilm reactor (Gieseke et al., 2003) and granular SBR (Wilen et al., 2004; this study). The thickness of the DBL ranged between 91 and 140 mm and a considerable DO 2.0
drop through the diffusive boundary layer was always observed which shows the importance of the external mass transfer resistance. The nitrifying activity was mainly concentrated in the first 100 mm resulting in a steep gradient of DO concentration. From the obtained values it can be inferred that the oxygen flux is mainly influenced by the DO in the bulk liquid and by the temperature of operation since this affects the biomass activity (Jubany et al., 2008).
1.5
4.3.2.
1.0 0.5 0.0 0
1
2 S 1/2
(CO 2 )
3
4
5
-1 1/2
(mg O2 L )
Fig. 4 e AOR estimation from oxygen microprofiles vs. square root of DO concentration at the surface of the granule (with trend line).
Nitrite microprofiles in the partial nitrification zone
NO 2 microprofiles were determined under conditions of excess of ammonium (initial NHþ 4 concentration in the bulk liquid was fixed at 140 mg N L1), starting with low concentration of nitrite in the bulk liquid and at different DO concentration. NO 2 microprofiles corresponding to the three higher DO concentrations tested are represented in Fig. 5. Low NO 2 concentrations in the bulk liquid were used in order to obtain a clearly visible nitrite peak in the nitrification zone allowing the calculation of the nitrification rate profiles. The NO 2 concentration at the surface of the granules was
4366
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 5 9 e4 3 7 0
Table 2 e Comparison of DO microprofiles in autotrophic nitrifying biofilms. DObulk (mg O2 L1)
DOsurface (mg O2 L1)
DO100 mm (mg O2 L1)
T ( C)
Biofilm width (mm)
DBL dh (mm)
JO2 (g O2 m2 d1)
Refd
2.9 4.5 2.6 3.7 3.0 4.8 0.5 0.8
0.2 1.3 0.2 1.3 0.6 1.4 0.0 0.1
30 30 25 20 20 20 25e27 20
1000e1250c 500e1500c 200e500 200 350 2500c 400 2500c
120 100a 100 140 100 108 91a 120
10.5a 10.6a 5.9b 5.0a 5.2 5.1 2.8b 1.8
[1] [2] [3] [4] [5] T.S. [6] T.S.
8.0 7.4 8.0 7.4 6.1 8.0 2.0 2.0
a Values published by the authors in the indicated reference. The value published by Kindaichi et al. (2006) was the ammonium flux of 1.1 g N m2 d1 and it was converted to oxygen flux with the stoichiometric coefficient (4.57 g O2 per g NHþ 4 eN for complete nitrification). b Values estimated considering an oxygen diffusivity at 25 C of 1.9 104 m2 d1 (Gieseke et al., 2003). c Radius of the granules. d [1] Schramm et al. (1999): Values corresponding to the port A of the fluidized bed reactor [2] de Beer et al. (1993) [3] Gieseke et al. (2003) [4] Kindaichi et al. (2006) [5] Okabe et al. (1999) [6] Wilen et al. (2004): Values corresponding to the upstream microprofile of the reference. T.S. ¼ this study.
significantly higher than the NO 2 concentration in the bulk liquid. Two well defined zones could be differentiated: 1) the nitrification zone, corresponding to the external layers in contact with the bulk liquid where nitrite was produced by AOB and 2) the anammox zone where NO 2 was consumed together with NHþ 4 under anoxic conditions. From the obtained microprofiles it is observed that the nitrite peak and the nitrite production rate became higher by increasing the DO concentration in the bulk liquid. Due to the produced concentrations gradient in the nitrification zone, nitrite diffused to both bulk liquid and inner layers of the granule. These concentrations gradients were caused by the nitrite generation in the nitrifying zone, its diffusional transfer to the bulk liquid during continuous operation, and the NO 2 consumption in the inner layers by anammox bacteria. However, the steepness of the NO 2 gradient in the DBL is an artefact caused by the low NO 2 concentration in the bulk liquid, and inside the reactor on the average there should be a considerably lower net flux to the liquid phase than that suggested by the present model example.
a
-
0.0
b
-1
NO2 (mg N L ) 1.5 3.0 4.5
0.0
6.0
NO2- (mg N L-1) 1.5 3.0 4.5
c 0.0
6.0
0.4
0.4
0.4
0.2
0.2
0.2
-0.6
Depth (mm)
-0.4
Depth (mm)
-0.2
-0.2 -0.4 -0.6
-0.4 -0.6
-0.8
-0.8
-1.0
-1.0
-1.0
-1.2
-1.2
-1.2 -12
-
6.0
-0.2
-0.8
-12 -6 0 6 12 NO2- cons. rate (g N (Lgranule)-1 d-1)
NO2- (mg N L-1) 1.5 3.0 4.5
0.0
0.0
0.0 Depth (mm)
The zone of NO 2 production determined with the NO2 microsensor fitted well with the one obtained using the DO microsensor. A lower level of details compared to the DO microprofiles was obtained due to the bigger discretization step used (50 mm instead of 25 mm) and the Monod kinetic was also well reflected with higher activities in the external layers. Similar as in the case of DO consumption, using pure DO saturation, the width of the outer 200 mm layer of the granules was working at maximal activity. The oxygen consumption rates were five times higher than the nitrite production rates. This ratio is superior to the stoichiometrically required one (according to the stoichiometry of ammonia oxidation to nitrite, the ratio would be 3.4 g O2 (g N)1). The different obtained values can be attributed to a mismatch of the NO 2 diffusivity. The NO2 diffusivity might be enhanced/decreased by the production of anions and cations in the aggregate which would cause the diffusivity NO 2 /diffusivity O2 ratio in the nitrification zone to have a different value than the ratio in pure water. The diffusional transport of negative and positive ions through the aggregates
-6
0
6
-12
12 -1
-1
NO2 cons. rate (g N (Lgranule) d )
-
-6
0
6
12 -1
NO2 cons. rate (g N (Lgranule) d-1)
Fig. 5 e Nitrite profiles (C, bars indicate standard deviation, n [ 3) and local consumption rates (bar) under different O2 concentrations in the bulk liquid: a) DO [ 8.0 mg O2 LL1 b) DO [ 20.5 mg O2 LL1 c) DO [ 35.2 mg O2 LL1. The NOL 2 concentration in the aqueous phase was kept at low values around 1 mg LL1.
4367
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 5 9 e4 3 7 0
and the bulk liquid must be in equilibrium. This may affect the ion diffusivities which may be different from the self-diffusion coefficients (Reimer and Harremoes, 1978). In fact, the actual diffusion coefficients might even be more different than the 10% suggested by literature (Li and Gregory, 1974; Picioreanu et al., 1997). The flux of NO 2 out of the nitrification zone, the total flux (i. e., the NO 2 production), and the flux to the core of the granule are shown in Table 3. As it was expected, the flux of nitrite towards the bulk liquid and towards the anammox zone increased when the DO concentration in the bulk liquid increased.
4.3.3.
Kindaichi et al. (2007) obtained a similar maximal nitrite removal capacity with a maximal volumetric rate of 2 1 d 5.0 g N (Lbiofilm)1 d1 and a similar NO 2 flux of 2.2 g N m performing microprofiles in the biofilm of an anaerobic fixed bed column operated at 37 C. However in their case, the maximal anammox reaction width was over 1300 mm of the biofilm, whereas in the present study, anammox bacteria were located in the range of depths between 400 and 1000 mm. Using the information of the NO 2 microprofiles, an estimation of the nitrogen removal rate of the SBR using Eq. 12 was performed. With these data and taking into account the stoichiometry of the anammox reaction, an ANR of 0.5 g N L1 d1 was estimated in the SBR. This value was close to the mean ANR value obtained in the reactor from the macroscopic measurements (Fig. 1). Our microscopic analysis of the granules was thus in agreement with macroscopic observations.
Anammox zone
In order to obtain the maximal anammox activity inside the granule, neither ammonium nor nitrite concentrations should be limiting. Therefore, high concentrations of ammonium 1 (initial concentration of 140 mg NHþ 4 eN L ) and nitrite (from 1 8.4 to 42.0 mg NO2 eN L ) were applied in the bulk liquid while the DO concentration was kept at air saturation (8 mg O2 L1). The CANON reactors are normally operated under nitrite limitation since it is necessary to keep the anammox potential higher than the nitrification potential in order to avoid the occurrence of nitrite build-up which is usually followed by irreversible nitrite inhibition of the anammox biomass (Nielsen et al., 2005). Even in those experiments performed with pure oxygen saturation, the NO 2 concentration in the bulk liquid of 1.3 mg N L1 was not sufficient to attain maximal anammox activity since NO 2 concentration became zero in the deeper layers where anammox bacteria were still present. The strategy followed to estimate the maximal anammox activity was to maintain the DO concentration at 8 mg O2 L1 (the conditions of the granular SBR) and to increase the NO 2 concentration in the bulk liquid. In the experiment carried out at a nitrite concentration in the bulk liquid of 9 mg N L1 (Fig. 6), the nitrite concentration obtained in the inner core of the granule was 0.7 mg N L1. Since nitrite half saturation constant of anammox bacteria was reported to be less than 0.1 mg N L1 (Strous et al., 1999), maximal anammox activity in the granule was then ensured. From the nitrite profiles obtained in Fig. 5 and Fig. 6 it can be inferred that, by keeping the DO concentration in the bulk liquid constant, the increase of the NO 2 concentration in the bulk liquid increases the flux of nitrite to the anammox zone up to a maximum value of 1.0 g N m2 d1 (Table 3). Such high fluxes of nitrite to the anoxic layers of the granule are expected during the SBR operation since the mean nitrite concentration in the bulk liquid ranged between 12 and 42 mg N L1.
4.3.4.
Causes of error
In order to better explain the obtained results, the different assumptions made during the performance of the experiments with the microsensors and the corresponding calculations have to be analyzed. As it was indicated the calculations were based on the assumption that the granules are described as flat surfaces, whereas the granules were sphere shaped and thus this fact could influence the flow pattern. Besides, it was assumed that the granules were homogeneous in structure. It has been widely demonstrated that biofilms are heterogeneous and therefore diffusivities should be locally defined. However, it was beyond the scope of this article to study in depth the heterogeneity of the bacteria populations and the approximations done are satisfactory to give important information about the processes carried out into the granule. Finally, the surface layer of the granule was somewhat fluffy, where a visual definition of “surface” within 50 mm was difficult. By performing several repetitions, the high values obtained for the standard deviations confirmed the high variability of the results, but, as it will be further explained, the approximations were satisfactory since the microscopic results predicted to a high extent the macroscopic ones.
4.4. From microscale results to granular CANON operation A successful strategy to start-up a CANON reactor is the promotion of the growth of the AOB population in the form of granular biomass. AOB produce nitrite and consume oxygen to provide anoxic conditions in the inner core of the granules. In this anoxic zone, ammonium (left from the AOB activity) and nitrite (from partial nitrification) have to be present in order to allow the growth of anammox bacteria (Va´zquez-Padı´n et al.,
L Table 3 e DO and NOL 2 fluxes in and out of the nitrifying zone at different DO concentrations in the bulk liquid. The NO2 concentration in the bulk liquid was kept at about 1 mg N LL1.
DObulk (mg O2 L1) 35.2 20.5 8.1
JDBL O2 (g O2 m2 d1)
DBL Amx JTotal NO2 ¼ JNO2 þ JNO2 (g N m2 d1)
JAmx NO2 (g N m2 d1)
Total 1 JDBL O2 ðJNO2 Þ (g O2 (g N)1)
20.2 11.9 5.1
2.7 2.3 1.3
1.0 0.7 0.4
7.5 5.2 4.0
4368
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 5 9 e4 3 7 0
NO2- (mg N L-1) 0.4
0
3
6
9
12
-15
0
15
30
Depth (mm)
0.2 0.0 -0.2 -0.4 -0.6 -0.8 -1.0 -30
NO2- cons. rate (g N (Lgranule)-1 d-1) Fig. 6 e Nitrite concentration profile (C) and average local consumption rate ( ; bars indicate standard deviation: n [ 6) by incubation with a NOL 2 concentration of 9 mg N LL1 in the bulk liquid.
2009). The width of the external layer mainly composed by AOB should be thick enough to protect anammox bacteria from the penetration of dissolved oxygen, but its activity must be controlled to avoid inhibition of anammox bacteria by high nitrite concentrations. From the results obtained by application of microsensors it is possible to extract knowledge about both optimal and safe conditions for the operation of the CANON system, in terms of ammonia and nitrite concentrations in the bulk liquid. At this point a control strategy could be defined to regulate the concentration of both compounds by using on-line ammonia and nitrite analyzers. The variables of control would be the DO concentration and the HRT value using the air flow rate and the inlet flow rate as actuation elements.
Different scenarios for the different ranges of ammonium and nitrite concentrations are possible (Fig. 7a). Low NHþ 4 and concentrations would limit the activities of AOB and/or NO 2 anammox bacteria. AOB activity would be limited by the ammonia concentration when the ratio DO/NHþ 4 (RON) in the bulk liquid is lower than 3.4 g O2 (g N)1 DNH4/DO2 (Harremoes and Henze, 2002). The limitation of AOB by ammonia would expose the reactor to a risk of failure since the granules would be fully penetrated by oxygen which would temporally inhibit anammox bacteria and enhance the undesired growth of NOB (Sliekers et al., 2005). On the other hand, high NHþ 4 and/or NO2 concentrations have the following disadvantages: both substrates can inhibit AOB and anammox bacteria with a consequent detriment of the produced effluent quality. Moreover, maintaining concentrations of NO 2 higher than the optimal ones involves higher costs of aeration. The control strategy corresponding to the different defined zones to restore the optimal conditions will be: a) in case of nitrite concentration out of the optimal range to act over the DO concentration (increasing or decreasing its value); b) whereas to optimize the NHþ 4 concentration the value of the HRT must be modified as the control parameter; and in the case where NHþ 4 and NO2 concentrations were not in the optimal range, both, HRT value and DO concentration would be changed according to Fig. 7b. Making use of the previous qualitative analysis the optimal values for NHþ 4 and NO2 concentrations can be defined for a specific case. These parameters will be determined specifically for the biomass present in each CANON reactor. From the microprofiles measured into the granules, the minimum nitrite concentration required to avoid substrate limitation for anammox bacteria can be estimated. Then, from these values and taking into account the stoichiometry of the partial nitrification and anammox processes, a minimum ammonium concentration in the bulk liquid can also be calculated. The fact that a minimal concentration of both ammonium and nitrite are necessary in the bulk liquid to
L Fig. 7 e a) Zones defining the different operational conditions in a CANON reactor fed with different NHD 4 and NO2 concentrations in the bulk liquid. b) Control strategy to return the CANON reactor to optimal conditions depending on NHD 4 and NOL 2 concentrations in the bulk liquid. ðRON [3:4 DNH4 =DO2 Þ.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 5 9 e4 3 7 0
ensure the maximal activity of anammox bacteria can represent a drawback if the effluent of the CANON reactor has to be released to a natural media but has not a significant impact if the effluent is returned to the head of the WWTP.
5.
Conclusions
Ammonia oxidation to nitrite is highly dependent on the oxygen mass transfer. Oxygen limitation regulates the amount of nitrite produced and as a consequence influences the anammox activity. Nitrite microprofiles revealed that a minimum nitrite concentration around 9 mg N L1 was necessary to ensure that the anammox biomass was working at maximal activity. Oxygen and nitrite microprofiles determined by application of microsensors correlated to the distributions of AOB and anammox bacteria inside the granules. Estimated bacterial activities from microscale analysis (0.5 g N L1 d1) gave similar results to those obtained from macroscopic measurements in the granular SBR. Accurate regulation of the dissolved oxygen concentration in the bulk liquid and the HRT value are crucial to control the nitrogen removal process in a CANON system avoiding either limitations or inhibitory phenomena.
Acknowledgement This work was funded by the Spanish Government (Togransys and NOVEDAR_Consolider project CSD2007-00055). We gratefully acknowledge Preben Sorensen for technical assistance, Andreas Schramm and Daniel Aagren Nielsen for assistance with cryosectioning and FISH.
references
Amann, R., Ludwig, W., Schleifer, K.H., 1995. Phylogenetic identification and in-situ detection of individual microbial-cells without cultivation. Microbiological Reviews 59 (1), 143e169. Beun, J.J., Heijnen, J.J., van Loosdrecht, M.C.M., 2001. N-removal in a granular sludge sequencing batch airlift reactor. Biotechnology and Bioengineering 75 (1), 82e92. Crocetti, G.R., Banfield, J.F., Keller, J., Bond, P.L., Blackall, L.L., 2002. Glycogen-accumulating organisms in laboratory-scale and full-scale wastewater treatment processes. Microbiology 148 (11), 3353e3364. Dapena-Mora, A., Fernandez, I., Campos, J.L., Mosquera-Corral, A. , Mendez, R., Jetten, M.S.M., 2007. Evaluation of activity and inhibition effects on anammox process by batch tests based on the nitrogen gas production. Enzyme and Microbial Technology 40 (4), 859e865. de Beer, D., van den Heuvel, J.C., Ottengraf, S.P.P., 1993. Microelectrode measurements of the activity distribution in nitrifying bacterial aggregates. Applied and Environmental Microbiology 59 (2), 573e579. Dosta, J., Fernandez, I., Vazquez-Padin, J.R., Mosquera-Corral, A., Campos, J.L., Mata-Alvarez, J., Mendez, R., 2008. Short- and long-term effects of temperature on the anammox process. Journal of Hazardous Materials 154 (1e3), 688e693.
4369
Freitag, T.E., Prosser, J.I., 2003. Community structure of ammoniaoxidizing bacteria within anoxic marine sediments. Applied and Environmental Microbiology 69 (3), 1359e1371. Gieseke, A., Bjerrum, L., Wagner, M., Amann, R., 2003. Structure and activity of multiple nitrifying bacterial populations co-existing in a biofilm. Environmental Microbiology 5 (5), 355e369. Gieseke, A., Tarre, S., Green, M., de Beer, D., 2006. Nitrification in a biofilm at low pH values: role of in situ microenvironments and acid tolerance. Applied and Environmental Microbiology 72 (6), 4283e4292. Guisasola, A., Jubany, I., Baeza, J.A., Carrera, J., Lafuente, J., 2005. Respirometric estimation of the oxygen affinity constants for biological ammonium and nitrite oxidation. Journal of Chemical Technology and Biotechnology 80 (4), 388e396. Harremoes, P., Henze, M., 2002. Wastewater Treatment. In: Henze, M., Harremoes, P., La Cour Jansen, J., Arvin, E. (Eds.). Springer, pp. 157e208. Hippen, A., Rosenwinkel, K.H., Baumgarten, G., Seyfried, C.F., 1997. Aerobic deammonification: a new experience in the treatment of wastewaters. Water Science and Technology 35 (10), 111e120. Jørgensen, B.B., Revsbech, N.P., 1985. Diffusive boundary layers and the oxygen uptake of sediments and detritus. Limnology and Oceanography 30, 11e21. Jubany, I., Carrera, J., Lafuente, J., Baeza, J.A., 2008. Start-up of a nitrification system with automatic control to treat highly concentrated ammonium wastewater: experimental results and modeling. Chemical Engineering Journal 144 (3), 407e419. Kindaichi, T., Kawano, Y., Ito, T., Satoh, H., Okabe, S., 2006. Population dynamics and in situ kinetics of nitrifying bacteria in autotrophic nitrifying biofilms as determined by real-time quantitative PCR. Biotechnology and Bioengineering 94 (6), 1111e1121. Kindaichi, T., Tsushima, I., Ogasawara, Y., Shimokawa, M., Ozaki, N., Satoh, H., Okabe, S., 2007. In situ activity and spatial organization of anaerobic ammonium-oxidizing (anammox) bacteria in biofilms. Applied and Environmental Microbiology 73 (15), 4931e4939. Kjær, T., Larsen, L.H., Revsbech, N.P., 1999. Sensitivity control of ion-selective biosensors by electrophoretically mediated analyte transport. Analytica chimica acta 391 (1), 57e63. Kuai, L.P., Verstraete, W., 1998. Ammonium removal by the oxygen-limited autotrophic nitrification-denitrification system. Applied and Environmental Microbiology 64 (11), 4500e4506. Li, Y.H., Gregory, S., 1974. Diffusion of ions in sea-water and in deep-sea sediments. Geochimica et Cosmochimica Acta 38 (5), 703e714. Lorenzen, J., Larsen, L.H., Kjaer, T., Revsbech, N.P., 1998. Biosensor determination of the microscale distribution of nitrate, nitrate assimilation, nitrification, and denitrification in a diatom-inhabited freshwater sediment. Applied and Environmental Microbiology 64 (9), 3264e3269. Loy, A., Maixner, F., Wagner, M., Horn, M., 2007. probeBase e an online resource for rRNA-targeted oligonucleotide probes: new features 2007. Nucleic Acids Research 35 (Suppl. 1), D800eD804. Mosquera-Corral, A., Gonzalez, F., Campos, J.L., Mendez, R., 2005. Partial nitrification in a SHARON reactor in the presence of salts and organic carbon compounds. Process Biochemistry 40 (9), 3109e3118. Nielsen, M., Bollmann, A., Sliekers, O., Jetten, M., Schmid, M., Strous, M., Schmidt, I., Larsen, L.H., Nielsen, L.P., Revsbech, N. P., 2005. Kinetics, diffusional limitation and microscale distribution of chemistry and organisms in a CANON reactor. FEMS Microbiology Ecology 51 (2), 247e256. Okabe, S., Satoh, H., Watanabe, Y., 1999. In situ analysis of nitrifying biofilms as determined by in situ hybridization and
4370
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 5 9 e4 3 7 0
the use of microelectrodes. Applied and Environmental Microbiology 65 (7), 3182e3191. Picioreanu, C., van Loosdrecht, M.C.M., Heijnen, J.J., 1997. Modelling the effect of oxygen concentration on nitrite accumulation in a biofilm airlift suspension reactor. Water Science and Technology 36 (1), 147e156. Ploug, H., Kuhl, M., BuchholzCleven, B., Jorgensen, B.B., 1997. Anoxic aggregates e an ephemeral phenomenon in the pelagic environment? Aquatic Microbial Ecology 13 (3), 285e294. Rasmussen, K., Lewandowski, Z., 1998. Microelectrode measurements of local mass transport rates in heterogeneous biofilms. Biotechnology and Bioengineering 59 (3), 302e309. Reimer, M., Harremoes, P., 1978. Multi-component diffusion in denitrifying biofilm. Progress in Water Technology 10, 149e165. Revsbech, N.P., 1989. An oxygen microsensor with a guard cathode. Limnology and Oceanography 34 (2), 474e478. Schramm, A., de Beer, D., Wagner, M., Amann, R., 1998. Identification and activities in situ of Nitrosospira and Nitrospira spp. as dominant populations in a nitrifying fluidized bed reactor. Applied and Environmental Microbiology 64 (9), 3480e3485. Schramm, A., de Beer, D., van den Heuvel, J.C., Ottengraf, S., Amann, R., 1999. Microscale distribution of populations and activities of Nitrosospira and Nitrospira spp. along a macroscale gradient in a nitrifying bioreactor: quantification by in situ hybridization and the use of microsensors. Applied and Environmental Microbiology 65 (8), 3690e3696. Sliekers, A.O., Haaijer, S.C.M., Stafsnes, M.H., Kuenen, J.G., Jetten, M.S.M., 2005. Competition and coexistence of aerobic ammonium- and nitrite-oxidizing bacteria at low oxygen concentrations. Applied Microbiology and Biotechnology 68 (6), 808e817. Strous, M., VanGerven, E., Zheng, P., Kuenen, J.G., Jetten, M.S.M., 1997. Ammonium removal from concentrated waste streams
with the anaerobic ammonium oxidation (anammox) process in different reactor configurations. Water Research 31 (8), 1955e1962. Strous, M., Kuenen, J.G., Jetten, M.S.M., 1999. Key physiology of anaerobic ammonium oxidation. Applied and Environmental Microbiology 65 (7), 3248e3250. Third, K.A., Sliekers, A.O., Kuenen, J.G., Jetten, M.S.M., 2001. The CANON system (completely autotrophic nitrogen-removal over nitrite) under ammonium limitation: Interaction and competition between three groups of bacteria. Systematic and Applied Microbiology 24 (4), 588e596. van Dongen, U., Jetten, M.S.M., van Loosdrecht, M.C.M., 2001. The SHARON-Anammox process for treatment of ammonium rich wastewater. Water Science and Technology 44 (1), 153e160. Va´zquez-Padı´n, J.R., Ferna´ndez, I., Figueroa, M., MosqueraCorral, A., Campos, J.L., Me´ndez, R., 2009. Applications of anammox based processes to treat anaerobic digester supernatant at room temperature. Bioresource Technology 100, 2988e2994. Vlaeminck, S.E., Cloetens, L.F.F., Carballa, M., Boon, N., Verstraete, W., 2008. Granular biomass capable of partial nitritation and anammox. Water Science and Technology 58 (5), 1113e1120. Wiesmann, U., 1994. Advances in Biochemical Engineering/ Biotechnology, Heidelberg, pp. 113e154. Wilen, B.M., Gapes, D., Keller, J., 2004. Determination of external and internal mass transfer limitation in nitrifying microbial aggregates. Biotechnology and Bioengineering 86 (4), 445e457. Wilhelm, R., Abeliovich, A., Nejidat, A., 1998. Effect of long-term ammonia starvation on the oxidation of ammonia and hydroxylamine by Nitrosomonas europaea. Journal of Biochemistry 124 (4), 811e815.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 7 3 e4 4 8 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Modelling the population dynamics and metabolic diversity of organisms relevant in anaerobic/anoxic/aerobic enhanced biological phosphorus removal processes A. Oehmen a,*, C.M. Lopez-Vazquez b, G. Carvalho a,c, M.A.M. Reis a, M.C.M. van Loosdrecht d a
REQUIMTE/CQFB, Chemistry Department, FCT, Universidade Nova de Lisboa, 2829-516 Caparica, Portugal Department of Urban Water and Sanitation, UNESCO-IHE Institute for Water Education, Wesvest 7, 2611 AX Delft, The Netherlands c Instituto de Biologia Experimental e Tecnolo´gica (IBET), Av. da Repu´blica EAN, 2780-157 Oeiras, Portugal d Department of Biotechnology, Delft University of Technology, Julianalaan 67, 2628 BC Delft, The Netherlands b
article info
abstract
Article history:
In this study, enhanced biological phosphorus removal (EBPR) metabolic models are expanded
Received 9 March 2010
in order to incorporate the competition between polyphosphate accumulating organisms
Received in revised form
(PAOs) and glycogen accumulating organisms (GAOs) under sequential anaerobic/anoxic/
31 May 2010
aerobic conditions, which are representative of most full-scale EBPR plants. Since PAOs and
Accepted 7 June 2010
GAOs display different denitrification tendencies, which is dependent on the phylogenetic
Available online 12 June 2010
identity of the organism, the model was separated into six distinct biomass groups, constituting Accumulibacter Types I and II, as well as denitrifying and non-denitrifying Competibacter
Keywords:
and Defluviicoccus GAOs. Denitrification was modelled as a multi-step process, with nitrate
Polyphosphate accumulating
(NO3), nitrite (NO2), nitrous oxide (N2O) and di-nitrogen gas (N2) being the primary components.
organisms (PAO)
The model was calibrated and validated using literature data from enriched cultures of PAOs
Glycogen accumulating organisms
and GAOs, obtaining a good description of the observed biochemical transformations. A strong
(GAO)
correlation was observed between Accumulibacter Types I and II, and nitrate-reducing and non-
Kinetics
nitrate-reducing PAOs, respectively, where the abundance of each PAO subgroup was well
Model calibration
predicted by the model during an acclimatisation period from anaerobiceaerobic to anaero-
Candidatus Accumulibacter
biceanoxic conditions. Interestingly, a strong interdependency was observed between the
Phosphatis clades
anaerobic, anoxic and aerobic kinetic parameters of PAOs and GAOs. This could be exploited
Fluorescence in situ hybridisation
when metabolic models are calibrated, since all of these parameters should be changed by an
(FISH)
identical factor from their default value. Factors that influence these kinetic parameters include the fraction of active biomass, relative aerobic/anoxic fraction and the ratio of acetylCoA to propionyl-CoA. Employing a metabolic approach was found to be advantageous in describing the performance and population dynamics in such complex microbial ecosystems. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
In the enhanced biological phosphorus removal (EBPR) process, the group of organisms primarily responsible for
phosphorus (P) removal are known as the polyphosphate accumulating organisms (PAOs). In order to promote the development of PAO and, consequently, P removal, anaerobic followed by anoxic and/or aerobic conditions are generally
* Corresponding author. Tel.: þ351 212 948 385; fax: þ351 212 948 550. E-mail address:
[email protected] (A. Oehmen). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.017
4474
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 7 3 e4 4 8 6
employed. PAOs are able to take up carbon sources such as volatile fatty acids (VFAs) anaerobically and store them as polyhydroxyalkanoates (PHAs), providing them a selective advantage over most ordinary heterotrophs. However, glycogen accumulating organisms (GAOs) are also capable of anaerobic VFA uptake and therefore can also be enriched under similar conditions as PAOs, consuming the generally limited VFA supply, without contributing to P removal. In recent years, the competition between PAO and GAO has been studied intensively due to (a) its impact on phosphorus removal performance and efficiency, and (b) PAO-dominated systems have the potential to decrease operational costs through minimising the addition of supplemental additives (e.g. chemical precipitants, organic carbon sources) necessary to achieve sufficient P removal (Oehmen et al., 2007a). The addition of these chemicals is also undesirable since they result in higher sludge generation, increasing sludge disposal costs. Among numerous operational and environmental factors, the carbon source (Pijuan et al., 2004; Oehmen et al., 2005b; Lu et al., 2006), pH (Filipe et al., 2001b; Schuler and Jenkins, 2002; Oehmen et al., 2005a) and temperature (Panswad et al., 2003; Lopez-Vazquez et al., 2007b, 2009a) have been observed to have a profound impact on the PAOeGAO competition. Recently, Lopez-Vazquez et al. (2009b) formulated a metabolic model that incorporates the combined effects of carbon source, pH and temperature on the metabolism of key EBPR microorganisms under anaerobiceaerobic conditions: specifically, Accumulibacter (PAO), Competibacter (GAO) and Defluviicoccus (GAO). In full-scale plants the EBPR process is invariably combined with nitrogen (N) removal. Different groups of PAOs and GAOs have shown varying denitrification capacities (Zeng et al., 2003c; Carvalho et al., 2007; Wang et al., 2008) that may have an important impact on their competition. This served as the motivation for the present study. It has been postulated that denitrifying PAOs (or DPAOs) able to reduce nitrate correlate well with Type I Accumulibacter, while non-DPAOs (or simply, PAOs) that are unable to reduce nitrate but able to reduce nitrite have been correlated with Type II Accumulibacter (Carvalho et al., 2007; Flowers et al., 2009; Oehmen et al., 2010). Due to the fact that Type I and Type II Accumulibacter correlate strongly with the so-called DPAO and PAO, respectively, we have adopted the terms PAOI and PAOII to differentiate between their different denitrification tendencies in this manuscript. This was done to avoid confusion, since both organisms appear capable of nitrite reduction, while the only difference is that PAOI are capable of nitrate reduction as well. The term “PAO” seems better suited as a more general term to describe all organisms that contribute to enhanced biological phosphorus removal in activated sludge systems. Kong et al. (2006) hypothesised that the different subgroups of Competibacter also display varying denitrifying capacities: (i) capable of nitrate and nitrite reduction (subgroup 6), (ii) able to reduce nitrate only (subgroups 1, 4 and 5) and (iii) unable to denitrify (subgroups 3 and 7). Wang et al. (2008) showed that an enrichment of Defluviicoccus Cluster I was able to reduce nitrate, but not nitrite, while Burow et al. (2007) suggested that Defluviicoccus Cluster II was unable to denitrify. It is clear from these studies that the denitrification
activity of PAOs and GAOs depends on the abundance of the different subgroups enriched. By expanding the metabolic model developed by LopezVazquez et al. (2009b), the present study focuses on the calibration and validation of a metabolic model developed to describe the biochemical activity of 6 microbial groups of PAOs and GAOs, namely the nitrate-reducing and non-nitrate-reducing Accumulibacter (i.e. PAOI and PAOII, respectively), denitrifying and nondenitrifying Competibacter (DGB and GB, respectively) and denitrifying and non-denitrifying Defluviicoccus (DDEF and DEF, respectively). Strategies aimed at facilitating the calibration of metabolic models based on a small number of parameters are also addressed in this study. Since model calibration is a very important and challenging issue in activated sludge modelling, ensuring the ease of metabolic model calibration is crucial in order to increase its potential utility in practice. Further, the ability of this model to assess the population dynamics of DPAO and PAO in microbial enrichments will be illustrated.
2.
Materials and methods
2.1.
Model development
From a physiological perspective, the metabolic model developed in this study incorporates the different capabilities of PAOs and GAOs to denitrify by separating them into multiple distinct groups. For this purpose, denitrification by PAOs and GAOs was modelled as a multi-step process from nitrate to nitrite, followed by nitrite to N2O, and finally N2O to N2. The stoichiometric matrix for PAOI and PAOII is shown in Appendix A, and that for GAOs and DGAOs (including GB, DGB, DEF and DDEF) is shown in Appendix B. In summary (see Fig. 1), PAOI are assumed to be capable of NO3, NO2 and N2O reduction, while PAOII are capable of NO2 and N2O reduction only, as suggested from their metagenome (Garcia Martin et al., 2006). DGB is also considered to be capable of NO3, NO2 and N2O reduction, while DDEF is capable of NO3 reduction only and GB and DEF are not capable of denitrifying. These properties are consistent with the results obtained from the aforementioned literature studies. It should be noted that PAOs and GAOs have been found to require a brief acclimation period (e4e5 h) to induce denitrification enzymes after being exclusively exposed to anaerobic/ aerobic conditions (Kuba et al., 1996b; Zeng et al., 2003a; Wang et al., 2008). In most systems, the organisms are exposed to anaerobic/anoxic/aerobic conditions, continuously exposing the bacteria to denitrifying conditions. Since the goal of this study was to model PAO and GAO steady-state metabolism, enzyme induction was not incorporated into the model. Nitrite accumulation (in the form of free nitrous acid: SHNO2) is known to inhibit P uptake by PAOs (Saito et al., 2004; Zhou et al., 2007), and can lead to the undesirable production of N2O (a powerful greenhouse gas) (Zhou et al., 2008). These aspects have been incorporated into the kinetic equations of the model.
2.2.
Anaerobic stoichiometry
The anaerobic stoichiometry of PAOI and PAOII is modelled identically. The anaerobic processes consist of VFA uptake (as
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 7 3 e4 4 8 6
4475
Fig. 1 e Extent of denitrification for each organism modelled in this study.
well as maintenance), where both acetate (SAc) and propionate (SProp) uptake are considered, since these form the major fraction of VFA in wastewater (Oehmen et al., 2007a). VFA uptake by PAOI and PAOII is coupled with polyphosphate hydrolysis (XPP) and phosphate release (SPO4), glycogen degradation (XGly) and PHA production (XPHA). The anaerobic processes of GAOs and DGAOs are similar to PAOs, except that polyphosphate (poly-P) hydrolysis and release as phosphate (P) does not occur. The anaerobic stoichiometric parameters are shown in Appendix C and are identical to the model of Lopez-Vazquez et al. (2009b).
2.3.
Aerobic and anoxic stoichiometry
Aerobically and anoxically, PAOs degrade PHA for P uptake and poly-P production, glycogen production and biomass growth (modelled here as the resultant of the total PHA degraded minus the PHA utilised for poly-P and glycogen production, see Murnleitner et al., 1997 and Lopez-Vazquez et al., 2009b), with the sole difference being the electron acceptor utilised for ATP production. The metabolism of GAOs is similar, without the reaction describing poly-P storage. The aerobic and anoxic yield coefficients are a function of the P/O ratio (d and dN, also known as YNADH_ATP), which represents the ATP produced per NADH oxidised during oxidative phosphorylation. The ATP production is dependent on the electron acceptor, and is lower under anoxic conditions as compared to aerobic conditions (Smolders et al., 1994; Kuba et al., 1996a). The aerobic and anoxic parameters used in this study are detailed in Appendix D. The yield coefficients are also a function of the phosphate transport energy (e and eN), the ATP necessary for biomass synthesis (K1 and K2) and the
percentage of acetyl-CoA* and propionyl-CoA* (l and b) in the PHA polymer (Smolders et al., 1994; Kuba et al., 1996a; Zeng et al., 2003b). The l and b coefficients are based on the relative composition of the PHA polymer produced under anaerobic conditions, including polyhydroxybutyrate (PHB), polyhydroxyvalerate (PHV) and polyhydroxy-2-methylvalerate (PH2MV). Since these PHA fractions are determined by the fraction of acetate and propionate in the influent wastewater, the carbon source affects the anaerobic, anoxic and aerobic stoichiometric yields. Similarly to Zeng et al. (2003b), it was assumed that there was no preferential utilisation of a particular PHA fraction for any particular aerobic or anoxic metabolic reaction. The NOx-based yield coefficients were calculated from the PHA-based yield coefficients through redox balancing, as performed in Lopez-Vazquez et al. (2009b) for the oxygen-based yields (see Appendices A and B). This calculation incorporates the degree of reduction of the PHA polymer and biomass formula, where the biomass composition found by Zeng et al. (2003b) was assumed for both PAOs and GAOs. In analogy to the approach taken by Kuba et al. (1996a), the NOx-based anoxic yields are multiplied by a factor corresponding to the difference in the number of available electrons per mole of electron acceptor (see Appendices A and B): 4 e/mol of O2 reduced, 2 e/mol of N reduced from NO3 to NO2, 2 e/mol of N reduced from NO2 to N2O, and 1 e/mol of N reduced from N2O to N2.
2.4.
Kinetic model
The kinetic expressions for the anaerobic, anoxic and aerobic processes of each organism are detailed in Appendix E. The structure of these expressions is similar to previous studies
4476
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 7 3 e4 4 8 6
(Murnleitner et al., 1997; Meijer et al., 2002; Lopez-Vazquez et al., 2009b), except for the inclusion of: (1) A switching function in the acetate and propionate uptake processes (below) to limit the specific fraction of PHA storage by PAOs and GAOs to a pre-specified value, fPHA, max, in order to avoid situations where PHA may accumulate beyond realistic levels. This is similar to the approach of de Kreuk et al. (2007). f PHA;max f X;PHA f PHA;max f X;PHA þ Ks;fPHA where: fX,PHA ¼ the specific fraction of PHA stored by each organism (PAO or GAO) fPHA,max ¼ the maximum specific fraction of PHA able to be stored by each organism Ks,fPHA ¼ the half-saturation coefficient for the specific PHA fraction (2) An exponential function, e170$SHNO2 , was added to the anoxic poly-P formation expressions, in order to account for inhibition by free nitrous acid (Zhou et al., 2007). It should be noted that while nitrite/SHNO2 has also been observed to inhibit aerobic P uptake (Yoshida et al., 2009), these expressions were not amended since this was beyond the scope of the present study. (3) An exponential function, e760$SHNO2 , was added to the N2O reduction process of PHA degradation by PAOI and PAOII, in order to account for the inhibitory effect of SHNO2 on N2O reduction (Zhou et al., 2008). It should be pointed out that the mechanisms affecting N2O reduction in PAOs and GAOs require further investigation. (4) Switching functions, SAc =SAc þ SProp and SProp =SAc þ SProp were added to the anaerobic acetate and propionate uptake expressions, respectively, in order to account for the competition by PAOs and GAOs for carbon source when both are present simultaneously, and preventing the VFA uptake rate from doubling when both substrates are present concurrently. The validity of this approach was considered beyond the scope of this study, since the
bioreactors that were described by the model were fed with single substrates only. (5) A switching function, SNO2 =SNO2 þ SNO3 , was added to the nitrite reduction kinetics for PHA degradation, glycogen production and maintenance of DGBs. The purpose of this switching function is to account for the observation that DGBs have been observed to display a preference towards nitrate reduction vs. nitrite reduction (Zeng et al., 2003c). This may be due to the fact that more Competibacter subgroups have been found to be capable of nitrate reduction as compared to nitrite reduction (Kong et al., 2006). This type of switching function is a common feature of other modelling strategies incorporating multistep denitrification (Sin et al., 2008). The kinetic parameters for the expressions of Appendix E are detailed in Table 1 and Appendix F.
2.5.
Model calibration/validation
This integrated metabolic model was implemented in Aquasim (Reichert, 1994), where it was defined as a sequencing batch reactor (SBR) with alternating anaerobiceanoxiceaerobicsettling-idle stages, similar to the approach of Lopez-Vazquez et al. (2009b). The model calibration was performed by adjusting only 4 kinetic parameters: the VFA uptake rate (qVFA_PHA), glycogen production rate (qGLY), PHA degradation rate (qPHA) and poly-P formation rate (qPO4_PP). Experimental data (i.e. acetate or propionate, glycogen, PHA and phosphate) from PAO or GAO systems operated under anaerobic/anoxic conditions (Zeng et al., 2003c; Carvalho et al., 2007; Wang et al., 2008) were used for fitting these parameters. The simulation length was set to a period of 3 sludge retention times (SRT) in order to describe the steady-state behaviour of PAO and DGB (Zeng et al., 2003c; Carvalho et al., 2007). The initial poly-P fraction was set to half the maximum fraction in order to ensure that it was neither limiting, nor approaching the saturation level, while the initial glycogen and PHA levels were defined according to their initial measured values. Due to the very low anoxic activity of the DDEF biomass group (Wang et al., 2008), the operation of this bioreactor necessitated the washout of
Table 1 e Parameter estimates for DPAOs and DGAOs obtained during model calibration and those applied during model validation. Carbon source
Organism
Cycle
VFA uptake
Glycogen prod.
PHA deg.
Poly-P Form.
Experimental Data
qVFA_PHA qGLY qPHA qPO4_PP C-mol/C-mol h C-mol/C-mol h C-mol/C-mol h P-mol/C-mol h Model calibration Propionate Acetate Acetate
PAOI DGBa DDEF
AN/AX AN/AX AN/AX
0.100 0.050 0.028
0.0025 0.0190 0.0025
0.050 0.210 0.100
0.0008
Carvalho et al. (2007) Zeng et al. (2003c) Wang et al. (2008)
Model validation Acetate Acetate
PAOI & DGBa DGBa
AN/AX AN/AX
0.050 & 0.050 0.050
0.0100 & 0.0190 0.0190
0.150 & 0.210 0.210
0.0025 & 0
Zeng et al. (2003a) Zeng et al. (2003c)
a Denotes granular sludge bioreactor.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 7 3 e4 4 8 6
the VFA and NOx at the end of the anaerobic and anoxic phases, respectively, while no sludge was wasted for this system. This suggested that a steady-state under anaerobic/ anoxic conditions was not achieved for this system, which is consistent with the model simulations (see Section 3.1). Thus, the parameter estimates correspond to the first cycle. In this cycle, the total amount of acetate taken up by the culture of Wang et al. (2008) was defined as the initial acetate concentration. Similarly to the FISH quantification results from this study, the sludge was simulated as 85% DDEF and the remainder as DGB. The parameter estimates obtained from model calibration were validated through simulating the results of a mixed PAO/ DGB enrichment under anaerobic/anoxic conditions with nitrate feeding (Zeng et al., 2003a) and a DGB enrichment under anaerobic/anoxic conditions with nitrite feeding (Zeng et al., 2003c). Since Zeng et al. (2003c) showed that N2O was the main product of denitrification in their system, the N2O reduction processes were not included during these model calibration or validation studies. The N2O reduction processes were included in the other simulation studies despite the fact that N2O was not measured (Zeng et al., 2003a; Carvalho et al., 2007; Wang et al., 2008). There is a clear need to better elucidate the activation/inactivation of N2O reductase.
2.6.
Simulation studies
The estimation of the kinetic parameters qVFA_PHA, qGLY, qPHA and qPO4_PP from denitrifying PAO/GAO studies was compared with estimates from other PAO/GAO studies (Filipe et al., 2001a; Zeng et al., 2003b; Oehmen et al., 2005a, 2006; Dai et al., 2007; Lopez-Vazquez et al., 2008; Bengtsson, 2009), in order to assess any inter-relationships between these parameters. In this set of simulations, the goal was to assess the maximum specific initial rate of PHA degradation, glycogen production and poly-P formation, rather than assess steady-state performance, thus the simulations were run over one cycle only. The maximum VFA uptake rate was calculated through linear regression directly from the experimental data of each study. Further, a sensitivity analysis was performed on each of these calibrated parameters in order to assess the effect of varying each parameter by 10% and 50% (see Appendix G). This was coupled with an error analysis in order to quantify the agreement between the model predictions and the experimental data (see Appendix G). Percent error was assessed through the normalised root mean squared deviation (NRMSD), as shown in the following equation: rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Pn 2 ðxmeas;i xpred;i Þ i¼1 NRMSD ¼
n
xmeas;max xmeas;min
(6)
where xmeas and xpred represent the measured and predicted concentrations of a given variable (e.g. VFA, PHA, P, glycogen) with n data points and xmeas,min and xmeas,max represent the minimum and maximum measured concentrations in that dataset. In the sensitivity analysis presented in Appendix G, the effect of a 10% change in parameter on the propagation of error over an extended simulation period (3 SRT) is also presented and discussed.
4477
Additionally, the acclimatisation of the propionate-fed EBPR system of Carvalho et al. (2007) from anaerobic/aerobic conditions to anaerobic/anoxic conditions was simulated in accordance with the operational conditions of this study. The initial PAOI and PAOII concentrations were estimated through fluorescence in situ hybridisation (FISH) analysis using probes Acc-I-444 and Acc-II-444 (Flowers et al., 2009), which target Accumulibacter clade IA and others, and clades IIA, IIC and IID, respectively. Details of the FISH quantification procedure with these probes are specified in Oehmen et al. (2010). The correlation between the measured Type I Accumulibacter by FISH and the model-simulated PAOI, as well as Type II Accumulibacter with PAOII, was explored. The minimum anoxic and aerobic SRT were calculated according to the following equation (Brdjanovic et al., 1998) for both acetate-fed (Murnleitner et al., 1997) and propionate-fed (Carvalho et al., 2007) anaerobic/anoxic systems, as well as acetate-fed (Brdjanovic et al., 1998) and propionate-fed (Oehmen et al., 2005a) anaerobic/aerobic systems.
SRTmin
YVFA PHA t $ YVFA X 24 ¼ qPHA $t3 1 f PHA;max f PHA;max3 3
where: YVFA_PHA ¼ the anaerobic PHA yield from VFA uptake. YVFA_X ¼ the anoxic (or aerobic) biomass growth from anaerobic VFA uptake. qPHA ¼ the maximum specific PHA degradation rate. t ¼ the total aerobic or anoxic phase time (in h) per day.
3.
Results
3.1.
Model calibration
The results of the model calibration for enriched cultures of PAOs and GAOs operated under anaerobic/anoxic conditions are displayed in Fig. 2. There was a good description of the conversion processes in each reactor. The propionate-fed PAO reactor of Carvalho et al. (2007) and acetate-fed DGB reactor of Zeng et al. (2003c) were simulated for a period of 3 SRTs in order to describe the steady-state behaviour of these systems (Fig. 2a and b, respectively). In addition to the good description of the VFA, PHA, glycogen, phosphate and nitrate/nitrite data, it was found that the final biomass concentration (56.9 and 89.8 Cmmol/L, for PAO and DGB, respectively) agreed within 10% of the measured value for each system (50.9 and 89.9 C-mmol/L, for PAO and DGB, respectively). This is particularly significant considering that the biomass growth rate is not explicitly defined in the model, but is calculated from the remainder of the PHA not used for glycogen or polyphosphate production. This supports the validity of the model structure. Furthermore, it was noteworthy that the competition between nitrate and nitrite reduction observed for DGBs was well described after including the switching function SNO2 =SNO2 þ SNO3 , with no other changes necessary in the model structure or anoxic kinetic parameters. The multiple expressions for each anoxic metabolic process that was necessitated by the inclusion of multi-step denitrification in the model of PAOs and DGBs
4478
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 7 3 e4 4 8 6
a
20 X_PHA X_PHA_meas X_Gly X_Gly_ me as S_HPr S_HPr_me as S_NO3 S_NO3_meas S_PO4 S_PO4_meas
C, P, N mmol/L
16 12 8 4 0 720
722
72 4
726
728
Time (hr )
b
30 X_PHA X_PHA_meas X_Gly X_Gly_meas S_HAc S_HAc_ meas S_NO3 S_NO3_meas S_NO2 S_NO2_meas
C, N mmol/L
25 20 15 10 5 0 1434
1435
1436
1437
1438
1439
1440
Time (hr)
c
10 X_PHA X_PHA_ meas X_Gly X_Gly_me as S_HAc S_HAc_ meas S_NO3 S_NO3_ meas S_NO2 S_NO2_ meas
C, N mmol/L
8 6 4 2 0 0
1
2
3
4
5
6
Time (hr ) Fig. 2 e Model calibration description of the biomass behaviour in a) an enriched PAO (Accumulibacter) culture fed with propionate b) an enriched DGB (Competibacter) culture fed with acetate c) an enriched DDEF (Defluviicoccus) culture fed with acetate. Lines indicate model description and symbols indicate experimental measurements.
contained identical anoxic parameters for PHA degradation (qPHA), glycogen production (qGLY) and P uptake (qPO4_PP). Each of these parameters can thus be viewed as the rate that describes the overall anoxic metabolism of each organism, independent of the actual denitrification electron acceptor (e.g. NO 3 or NO2 ). Fig. 2c shows the calibrated model description of the system of Wang et al. (2008), consisting mainly of DDEF. While
a good description of the data was achieved, the parameter estimates associated with the activity of DDEF were assessed from a single cycle, and the simulation was not able to be sustained over an extended period. This is consistent with the operational conditions employed to sustain the culture of Wang et al. (2008), and likely due to the very low glycogen concentration contained in the biomass. Notably, the initial glycogen concentration of this culture was approximately one
4479
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 7 3 e4 4 8 6
order of magnitude smaller than a similarly enriched Defluviicoccus culture operated under anaerobiceaerobic conditions (Dai et al., 2007). It appears that the low anoxic glycogen production is a limiting factor towards sustaining a DDEF culture, which is consistent with previous findings for Type II Accumulibacter enrichments (Carvalho et al., 2007). The commonality between DDEF and Type II Accumulibacter is an incomplete denitrification pathway; however, it is unclear how this impacts their regulation of glycogen production. In simulations with DDEF containing a higher initial glycogen concentration, their glycogen content was observed to decrease over time (data not shown), which appears to agree with the experimental results. The slower kinetic activity of DDEF as compared to both PAO and DGB indicates that these organisms are unlikely to be competitive under anaerobic/anoxic conditions as compared to PAOs and DGBs (see Table 1). Thus, the error associated with the kinetic parameter estimations of DDEF during anoxic conditions (see Appendix G) is likely to be of minor practical relevance.
3.2.
Model validation
For model validation purposes, two additional reactor systems were simulated using the calibrated parameters. Firstly, the data from the anaerobic/anoxic acetate-fed reactor of
a
Zeng et al. (2003a) is described in Fig. 3a. The anaerobic stoichiometry suggested that both PAOs and DGAOs were present in this sludge. The sludge was estimated to contain 70% PAO and 30% DGB, according to the method of Lopez-Vazquez et al. (2007a), which is based on the P release/acetate uptake ratio (YPO4_Ac). Furthermore, Zeng et al. (2003a) observed that the specific acetate uptake rate (qAc_PHA) was much lower as compared to previous studies, which was attributed to the biomass aggregation in the form of granules. Nevertheless, it has been found that a granular sludge system can be well modelled using the same parameters as for flocculent biomass (de Kreuk et al., 2007; Xavier et al., 2007). The lower maximal activity of biomass observed in granular systems is likely due to accumulation of inert biomass in the core of the granule. We decided not to further increase the current complexity of our model by including a (also unknown) biomass inactivation parameter. We solved it by adjusting the kinetic parameters for granular biomass in the anaerobic/anoxic acetate-fed PAO system previously calibrated by Murnleitner et al. (1997) (Fig. 3a). It should be noted that the kinetic parameters of acetate-fed DGBs (Fig. 2b) were also obtained from a system containing granules with a similar size (1e3 mm) (Zeng et al., 2003c). The kinetic parameters employed for PAO and DGB in Fig. 3a are shown in Table 1. Since qAc_PHA has been observed
20
C, P, N mmol/L
16 12 8 4 0 0
1
2
3
4
5
6
X_PHA X_PHA_ meas X_Gly X_Gly_me as S_HAc S_HAc_ meas S_NO3 S_NO3_ meas S_PO4 S_PO4_ meas S_NO2 S_NO2_ meas
Time (hr )
b
30
C, N mmol/L
25
X_PHA X_PHA_meas X_Gly X_Gly_ me as S_HAc S_HAc_meas S_NO2 S_NO2_meas
20 15 10 5 0 0
1
2
3
4
5
6
Time (hr ) Fig. 3 e Model validation description of the biomass behaviour in a) a 70% PAOe30% DGB culture fed with acetate and b) an enriched DGB culture metabolising nitrite anoxically. Lines indicate model description and symbols indicate experimental measurements.
4480
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 7 3 e4 4 8 6
to be similar between PAO and GB in anaerobic/aerobic systems (Lopez-Vazquez et al., 2009b), it was assumed that this trend would also hold under anaerobic/anoxic conditions. The anoxic kinetic parameters of Murnleitner et al. (1997) were then reduced by the same factor (half) as qAc_PHA. The justification for this decision is further explored in Section 3.3. From Fig. 3a, an acceptable fit between the model predictions and experimental data was found (see also Appendix G), supporting the validity of the kinetic parameters in a sludge containing both PAO and DGB. Interestingly, nitrite was observed to accumulate in this system; however, the amount was lower than the level of nitrite accumulation observed in the DGB culture of Zeng et al. (2003c). This effect was well described by the model when incorporating both the PAO and DGB fractions and suggests that the fraction of DGB contained in this sludge was likely responsible for the nitrite accumulation. Saito et al. (2004) previously observed that nitrite accumulation in an EBPR system was linked with the proliferation of GAOs, which is consistent with this result. It is possible that enriched PAO cultures not containing GAOs are less likely to lead to nitrite accumulation under anoxic conditions, assuming that Type II Accumulibacter are also present in the sludge (see Section 3.5) and able to reduce nitrite but not nitrate. This is also significant because anoxic nitrite (SHNO2) accumulation has been linked to N2O production in both PAOs (Zhou et al., 2008) and GAOs (Zeng et al., 2003c), thus minimising nitrite accumulation is desirable towards minimising greenhouse gas emissions. While N2O production was not found from the model predictions of Fig. 3a, it should be noted that a quantifiable relationship between SHNO2 accumulation and N2O production remains to be elucidated for GAOs, whereas it was developed for PAOs (Zhou et al., 2008). Predicting N2O production by PAOs and GAOs was not one of the objectives of this study, as more experimentally-based results would first be needed in order to develop a proper model description. Furthermore, the acetate-fed DGB reactor of Zeng et al. (2003c) was simulated with nitrite instead of nitrate as the electron acceptor (Fig. 3b). The model predictions present a close fit with the experimental data, supporting the validity of the model to describe the metabolism associated with both nitrate and nitrite reduction by DGB.
1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0.00
The results described above suggest that multiplying each kinetic parameter (qVFA_PHA, qPHA, qGly, qPO4_PP) by an identical factor can be a useful means of calibrating the kinetic rates of metabolic models. The validity of this strategy was evaluated by comparing the kinetic rates of PAO and GAO enriched cultures from literature studies. Fig. 4a illustrates the relationship between the specific anaerobic acetate uptake rate (qAc_PHA) and specific aerobic or anoxic PHA consumption rate (qPHA) for enriched GAO cultures. It should be pointed out that each of these enriched cultures, fed with acetate as the sole carbon source, was dominated with either Competibacter or Defluviicoccus, and was operated under either anaerobic/aerobic (Filipe et al., 2001a; Zeng et al., 2003b; Oehmen et al., 2006; Dai et al., 2007; Lopez-Vazquez et al., 2008; Bengtsson, 2009) or anaerobic/ anoxic conditions (Zeng et al., 2003c; Wang et al., 2008) in systems with flocular or granular sludge. A linear relationship was observed between these two parameters, as was also observed between qAc_PHA and qGly (Fig. 4b). This result suggests interdependency between the kinetic parameters. The reason for these differences in kinetic parameters could be related to disparities in the overall activity of the biomass in each system. It is clear that the kinetic activity will be proportional to the quantity of GAOs enriched in the sludge. The overall specific activity will also be dependent on the quantity of inactive cells and inert particulates. Indeed, Moussa et al. (2005) showed that the active biomass fraction had a clear impact on the kinetics of nitrifying SBR systems. In addition to the flocular structure of the biomass, other factors impacting the active biomass fraction include the sludge retention time (longer SRT leads to a higher accumulation of inert material resulting in a lower fraction of active cells), the aerobic/anoxic fraction (cells are less active under anoxic vs. aerobic conditions due to their lower efficiency in respiration) and substrate availability (i.e. organic loading rate can impact the availability of the substrate to the cells), even in very highly enriched cultures (e.g. 80e90%). Similarly, the relationship between the anaerobic and aerobic/anoxic parameters of qVFA_PHA, qPHA, qGly and qPO4_PP was analysed for enriched PAO cultures (Fig. 5). In this case,
b 0.35 qGAO,Gly (C-mol/C-mol hr)
qGAO,PHA (C-mol/C-mol hr)
a
3.3. The relationship between anaerobic VFA uptake rate and aerobic/anoxic kinetics of PAOs and GAOs
y = 4.17x - 0.03 R2 = 0.98
0.30 y = 0.98x - 0.02 R2 = 0.95
0.25 0.20 0.15 0.10 0.05 0.00
0.10
0.20
0.30
qGAO,Ac_PHA (C-mol/C-mol hr)
0.40
0
0.1
0.2
0.3
0.4
qGAO,Ac_PHA (C-mol/C-mol hr)
Fig. 4 e a) Relationship between anaerobic acetate uptake and aerobic or anoxic a) PHA degradation rate and b) glycogen production rate; by enriched GAO cultures (Competibacter and Defluviicoccus). Error bars correspond to a 10% change in parameter value.
4481
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 7 3 e4 4 8 6
b
0.50 0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00
qPAO,PO4_PP (P-mol/C-mol hr)
qPAO,PHA (C-mol/C-mol hr)
a
y = 2.85x - 0.23 R2 = 0.99
0. 00
0.05
0.10
0.15
0.20
0.006 0.005 y = 0.034x - 0.002 R2 = 0.959
0.004 0.003 0.002 0.001 0.000
0.25
0
qPAO,Prop_PHA (C-mol/C-mol hr)
d µPAO (C-mol/C-mol hr)
qPAO,Gly (C-mol/C-mol hr)
0.1
0.15
0.2
0.25
qPAO,Prop_PHA (C-mol/C-mol hr)
c 0.025 0.020 y = 0.16x - 0.01 R2 = 0.95
0.015
0.05
0.010 0.005
0.016 0.014
y = 0.08x - 0.00 R2 = 0.74
0.012 0.010 0.008 0.006 0.004 0.002 0.000
0.000 0
0.05
0.1
0.15
0.2
0.25
qPAO,Prop_PHA (C-mol/C-mol hr)
0
0.05
0.1
0.15
0.2
0.25
qPAO,Prop_PHA (C-mol/C-mol hr)
Fig. 5 e a) Relationship between anaerobic propionate uptake and aerobic or anoxic a) PHA degradation rate, b) polyphosphate formation rate, c) glycogen production rate, and d) biomass growth rate; by enriched PAO cultures (Accumulibacter). Error bars correspond to a 10% change in parameter value.
the sludges were enriched with Accumulibacter using propionate as the sole carbon source, under either anaerobic/anoxic conditions (Carvalho et al., 2007) or anaerobic/aerobic conditions at a series of different pHs, which was found to alter the kinetic activity of the biomass (Oehmen et al., 2005a). A linear relationship was observed between qVFA_PHA and qPHA (Fig. 5a), qPO4_PP (Fig. 5b), and qGly (Fig. 5c). It could be noted that one point seemed to fall outside the trend in Fig. 5b and c, which corresponded to the study performed at a pH of 8.5. This deviation was more prominent when plotting the maximum biomass growth rate (mPAO) vs. qPHA (Fig. 5d) for these studies (mPAO could be calculated from the aerobic ammonia removal rate by Oehmen et al., 2005a since a nitrification inhibitor was present in the media). This result suggests that a high pH constituted a “stressful condition” that altered PAO metabolism. Furthermore, the pH of 8.5 coincided with a higher-thanexpected glycogen production at the expense of biomass growth, which is consistent with the hypothesis put forth by Murnleitner et al. (1997) that PAOs (or GAOs) favour the replenishment of storage polymers over a high biomass growth rate for their survival. It should also be noted that the carbon source fed anaerobically (i.e. acetate or propionate) not only has an impact on the anaerobic kinetics, but also the anoxic kinetics, whereby the propionate-fed systems consistently display lower anoxic kinetic rates. Indeed, the aerobic kinetic parameters describing anaerobic/aerobic propionate-fed reactors were also lower than those associated with acetate-fed anaerobic/ aerobic reactors (Lopez-Vazquez et al., 2009b). The reason is likely due to a lower conversion rate for different PHA fractions produced with each carbon source. After PHB, PHV and PH2MV are converted to acetyl-CoA and propionyl-CoA, acetyl-CoA can be directly catabolised through the TCA cycle,
while propionyl-CoA must first be converted into acetyl-CoA (via e.g. pyruvate) before proceeding through the catabolic pathway (Zeng et al., 2003b; Oehmen et al., 2007b). This extra metabolic step likely leads to slower kinetic rates, necessitating the expression of the aerobic (or anoxic) rates as a function of the acetate/propionate ratio (Lopez-Vazquez et al., 2009b). This property of acetyl-CoA vs. propionyl-CoA metabolism has also been shown for aerobic PHA producing cultures, in addition to PAOeGAO systems (Dias et al., 2008; Lopez-Vazquez et al., 2009b). The results of the sensitivity and error analysis (Appendix G) showed that an increase or decrease of 10% in the parameters displayed in Figs. 4 and 5 led to a small increase (3%) in the deviation between the experimental measurements and model predictions (as calculated by NRMSD, see Section 2.6). A change of 50% to the parameter estimates led to a more substantial increase in the NRMSD of approximately 20%. Further, a 10% change in parameter value did not lead to significant steady-state error propagation. Therefore, a change in the kinetic parameters of approximately 10% does not have a significant impact on the capability of the model to describe PAO or GAO metabolism. Despite the fact that extraordinary conditions (e.g. high pH) could alter PAO or GAO metabolism, the linear relationship found in Figs. 4 and 5 suggest that these three or four main kinetic parameters can in most situations be described through assessment of only one parameter (qVFA_PHA), and the interdependency between the anaerobic and aerobic/ anoxic parameters can be used to calculate the remaining parameters. We hypothesise that the kinetics of these organisms are constant on a per cell basis, while the apparent rates observed in each system are a function of the relative level of activity of the culture.
4482
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 7 3 e4 4 8 6
3.4. The relationship between VFA uptake rate and aerobic/anoxic fraction
3.5. Describing the population dynamics of Type I and Type II Accumulibacter
As stated above, the anaerobic VFA uptake rate is influenced by the subsequent phase (aerobic or anoxic) that was employed in previous cycles. The explanation for this result could stem from either (1) a change in the microbial population (e.g. PAO or DGAO subgroup) that became adapted to these different operational conditions or (2) a direct influence of the aerobic/anoxic time fraction on the metabolism of the same population. To clarify this point, the biomass composition in the propionate-fed bioreactor operated by Carvalho et al. (2007) was studied along the acclimatisation period from anaerobic/aerobic to anaerobic/anoxic conditions. From Table 2, it can be observed that the anaerobic VFA uptake rate steadily decreased throughout acclimatisation, as the length of the anoxic phase was increased. However, an insignificant change in the population structure of Accumulibacter was found while the aerobic phase was applied; PAOII persisted while the PAOI population did not increase. The anoxic P uptake rate during this period also did not increase, which is in agreement with the stable Accumulibacter population observed. Only when the aerobic phase was eliminated did the PAOI population and anoxic P uptake rate increase significantly (at the expense of the PAOII population). It is noteworthy that this correlation between anoxic P uptake rate and Type I Accumulibacter further supports the hypothesis that Type I Accumulibacter are nitrate-reducing PAOs and Type II Accumulibacter are non-nitrate-reducing PAOs. Since the dynamics of the microbial populations contrasted with the steady decline in propionate uptake rate throughout acclimatisation, it was concluded that the aerobic/ anoxic fraction is proportional to qVFA_PHA. Based on this result, qVFA_PHA was related to the aerobic/anoxic fraction through the following relationship:
The applicability of this model was further tested through running a set of simulations under the same operational conditions as employed in the propionate-fed EBPR reactor of Carvalho et al. (2007) during the acclimatisation from anaerobic/aerobic conditions to anaerobic/anoxic conditions. Since GAOs were found to represent a very small portion of the biomass (<5%) throughout the study, their activity was assumed negligible, and only the PAOI and PAOII processes were considered. The initial PAOI and PAOII concentrations corresponded to the abundance of Type I and Type II Accumulibacter, respectively, measured by FISH. The results are displayed in Fig. 6, and show that the changes in PAOI and PAOII abundance predicted by the model correlate very strongly with the dynamics of Type I and Type II Accumulibacter. This result supports the validity of the model to describe the population dynamics of both PAOs under anaerobic/anoxic/aerobic conditions and further supports the hypothesis that Type I Accumulibacter are nitrate-reducing DPAOs, and Type II Accumulibacter are non-nitrate-reducing PAOs. Moreover, the strong correlation between model predictions and FISH quantification highlight the utility of the FISH probes developed by Flowers et al. (2009) to distinguish between DPAO and PAO. The strong agreement between the total Type I and Type II Accumulibacter as quantified by each individual probe with the total Accumulibacter population as assessed by the PAOmix probes of Crocetti et al. (2000) shows that the Type I and Type II probes covered the entire Accumulibacter population present in this sludge. The results from Fig. 6 also illustrate that the acclimatisation phase between anaerobic/aerobic and fully anaerobic/ anoxic conditions had little impact on the relative fraction of PAOI and PAOII. The small number of PAOI present in the sludge may have been responsible for the reduction of the 5e10 mgN/L in the 1e2 h anoxic phase, where the addition of nitrate to the reactor was regulated based on the nitrate demand of the culture (i.e. the quantity that could be removed without leaking into the aerobic and subsequent anaerobic periods). Only when the aerobic phase was completely eliminated did the nitrate demand increase substantially (50 mgN/ L), and correspondingly, the PAOI population also increased. These results are in good agreement with the model predictions of Filipe and Daigger (1999), who proposed that PAO outcompete DPAO because of their higher efficiency under
qVFA
PHA
¼ qVFA
PHA;AN AX
þ qVFA
PHA;AN OX
qVFA
PHA;AN AX
$fOX (7)
where fox represents the aerobic phase length divided by the sum of the aerobic and anoxic phase lengths. Table 2 shows that the qVFA_PHA,AN_AX is approximately half that of the qVFA_PHA,AN_OX. This is consistent with the results of Kuba et al. (1996a), who also observed that the qVFA_PHA decreased by half after adapting an acetate-fed anaerobic/ aerobic system to anaerobic/anoxic conditions.
Table 2 e Change in VFA uptake rate along acclimatisation from anaerobic/aerobic to anaerobic/anoxic conditions in the bioreactor operated by Carvalho et al. (2007). Stage of acclimatisation
e I II III
Propionate uptake rate
FISH quantification (%)
Anoxic P uptake rate
C-mol/ (C-mol h)
Acc-I-444
Acc-II-444
Pmmol/ (gVSS h)
0.26 0.21 0.16 0.12
5 8 5 44
82 81 74 31
0.10 0.10 0.13 0.63
Anoxic fraction (%)
Aerobic fraction (%)
0 25 50 100
100 75 50 0
4483
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 7 3 e4 4 8 6
Step 1
Abundance/Total Bacteria (%)
100
Step 2
Step 3
90 80 70 60 50 40 30 20 10 0 0
10
20
30
40
50
60
70
80
90
TIme (d)
Accumulibacter I Accumulibacter II Accumulibacter I+II
PAOI (Model) PAOII (Model) PAOmix
Fig. 6 e FISH quantification of Accumulibacter Type I and Type II and total Accumulibacter (Types I D II, PAOmix) in comparison with model predictions along the acclimatisation from anaerobiceaerobic to anaerobiceanoxic conditions. Day 0: aerobic fraction [ 100%; Step 1: anoxic fraction [ 25%, aerobic fraction [ 75%; Step 2: anoxic fraction [ 50%, aerobic fraction [ 50%; Step 3: anoxic fraction [ 100%.
aerobic conditions as compared to anoxic conditions. Indeed, the efficiency of oxidative phosphorylation (YNADH_ATP) under anoxic conditions is approximately half the aerobic value (Murnleitner et al., 1997; see Appendix D), and the anoxic kinetics are also far lower than the aerobic kinetics (Appendix G). In an effort to estimate the critical length of the anoxic phase necessary to enrich for PAOI, the minimum SRT was calculated according to the method of Brdjanovic et al. (1998) for anaerobic/anoxic and anaerobic/aerobic systems fed with either acetate or propionate. A close agreement was observed between the calculated minimum anoxic SRT at 20 C for acetate (4.3 d) and propionate (4.7 d), as well as the minimum aerobic SRT in each case (1.6 d and 1.7 d, respectively). Since PAOI will also grow under aerobic conditions the minimum anoxic SRT for PAOI selection could be lower in an integrated anaerobic/anoxic/aerobic system. The critical anoxic phase length for PAOI enrichment over PAOII still requires further research, especially since the actual competitive difference between the two organisms is likely small and e.g. PAOII could compete with PAOI for nitrite in denitrification. Indeed, it is noteworthy that the PAOII population was not completely washed out in Fig. 6, even after 4 SRTs of bioreactor operation and model simulation. In order to assess if this effect was due to an insufficient bioreactor operation/simulation time, the model was then simulated under the same conditions for an additional 4 SRTs. It was found that the PAOI and PAOII populations remained similar, and the relative fraction achieved at steady-state was 55% PAOI and 45% PAOII. The fact that PAOs did not completely washout under anaerobic/anoxic conditions supports the hypothesis that PAOII are able to use nitrite (Garcia Martin et al., 2006), and can thus survive by scavenging the nitrite produced from nitrate reducers (e.g. PAOI). This finding could also potentially explain why nitrite was observed to accumulate in an enrichment of DGBs (Zeng et al., 2003c), but not in the case of the PAO reactor
of Carvalho et al. (2007), as both PAOI and PAOII may have been responsible for the nitrite reduction, while only the PAOI were able to perform nitrate reduction.
4.
Discussion
4.1.
Modelling the metabolic activity of PAOI and PAOII
It has been often debated in literature whether or not PAO and DPAO are different organisms, and more recently, if PAOs are separated into 2 groups, whereby one group is able to denitrify from nitrate to di-nitrogen gas, and the other from nitrite to di-nitrogen gas. The stoichiometry and kinetics of two such Accumulibacter Types were described in the metabolic model of this study, and provided an ideal platform to test this hypothesis. The activity predicted by the model for each Accumulibacter Type agreed very well with the abundance of each Type assessed by oligonucleotide probes (see Fig. 6). We propose that this model can be useful in describing the differences in Accumulibacter metabolic activity, providing a valuable tool in assessing the impact of different Accumulibacter Types on EBPR performance. This will lead to advances in our knowledge about how PAOs function, and our ability to control the process. Ekama and Wentzel (1999) have previously recognized the challenge in modelling the kinetics of DPAOs, as previous attempts to model this process always required calibration to specific case studies and were of low predictive value. This work provides evidence showing that the model developed in this study can indeed be used to predict the growth and activity of PAOI and PAOII based on the operational conditions employed, thus providing an advantage over other modelling strategies. Further, this work opens up the possibility for new design and control strategies of full-scale EBPR systems based
4484
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 7 3 e4 4 8 6
on achieving a desired microbial community. For example, in COD-limited situations, it may be more desirable to enrich PAOI as they can provide both denitrification and P removal simultaneously. Under situations with a high influent P concentration, it may be more desirable to maximise the PHA driven for aerobic P uptake, which has a higher metabolic efficiency. Future work should examine the practical applicability of enriching each PAO Type in full-scale systems.
4.2. Factors affecting the kinetic parameters of metabolic models The correlation between anaerobic and aerobic (or anoxic) kinetic parameters of PAOs and GAOs was investigated in this study. The linear relationships shown in Figs. 4 and 5 suggest that these correlations may be useful when calibrating the apparent PAO and GAO activity, since each aerobic/anoxic parameter is essentially a linear function of the anaerobic VFA uptake rate. Measuring the VFA uptake rate is experimentally much easier then e.g. the glycogen formation or PHA consumption rate. In calibrating a full-scale system the default parameters for the different kinetic constants should be changed with the same percentage as the VFA uptake rate. This study represents the first time that such a correlation between kinetic parameters of PAOs and GAOs has been reported, to the best of our knowledge. In order to confidently make use of this information, it is important to understand the factors affecting the kinetic parameters, and the reason why they are subject to change in certain situations. The relative quantity of active/inactive biomass seems to be one of the major regulatory causes, which is consistent with the findings of Moussa et al. (2005). Hao et al. (2010) demonstrated that decay processes have a stronger effect on the specific activity of PAOs and GAOs (i.e. kinetic rates lowered under famine conditions) rather than on cell death (i.e. only small decreases in the biomass concentration was observed), which further highlights the importance of the active biomass fraction when describing the activity of these systems. The fact that the anoxic/aerobic fraction affects the kinetic parameters, including the anaerobic VFA uptake rate, may also be a reflection of relative biomass activity, since cells are less efficient metabolising NOx as compared to oxygen, leading to a lower overall biomass activity under anoxic conditions. The influent carbon source composition is another factor affecting aerobic/anoxic kinetics, likely due to its influence on the PHA composition, and more directly, the lower rate of propionyl-CoA metabolism as compared to acetyl-CoA.
5.
Conclusions
A metabolic model describing the steady-state growth, activity and competition of PAOs and GAOs under anaerobiceanoxiceaerobic conditions was developed, calibrated and experimentally validated. The main conclusions from this work are: 1) The model was able to describe well the metabolic transformations occurring in enriched PAO or GAO cultures
under anaerobic/anoxic conditions, as well as mixed PAO/ GAO cultures. 2) The population dynamics of PAOI and PAOII were successfully predicted along the acclimatisation from anaerobic/aerobic to anaerobic/anoxic conditions. Indeed, a strong correlation was observed between the abundance of Accumulibacter Types I and II in the sludge and the predicted PAOI and PAOII population, suggesting clear differences in the ability of each group to denitrify. 3) Interdependency was observed between the calibrated kinetic parameters for the apparent anaerobic, anoxic and aerobic activities of PAOs and GAOs. This suggests that although the kinetics of these cells on a per cell basis are likely constant, the apparent rates are subject to change depending on the activity level of each group of organisms in the sludge. This inter-relationship could be exploited as a practical means of calibrating metabolic models to describe the apparent activity in activated sludge systems subjected to different operational conditions.
Nomenclature The notation used in this article is in accordance with the new standardised framework for wastewater treatment modelling notation (Corominas et al., 2010). Descriptions of the model parameters are provided in the Appendices.
Acknowledgements The authors would like to thank an anonymous reviewer for their helpful comments. The Fundac¸a˜o para a Cieˆncia e a Tecnologia is acknowledged for grant SFRH/BPD/30800/2006.
Appendix. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2010.06.017.
references
Bengtsson, S., 2009. The utilization of glycogen accumulating organisms for mixed culture production of polyhydroxyalkanoates. Biotechnology and Bioengineering 104 (4), 698e708. Brdjanovic, D., van Loosdrecht, M.C.M., Hooijmans, C.M., Alaerts, G.J., Heijnen, J.J., 1998. Minimal aerobic sludge retention time in biological phosphorus removal systems. Biotechnology and Bioengineering 60 (3), 326e332. Burow, L.C., Kong, Y.H., Nielsen, J.L., Blackall, L.L., Nielsen, P.H., 2007. Abundance and ecophysiology of Defluviicoccus spp., glycogen accumulating organisms in full-scale wastewater treatment processes. Microbiology 153 (1), 178e185. Carvalho, G., Lemos, P.C., Oehmen, A., Reis, M.A.M., 2007. Denitrifying phosphorus removal: linking the process performance with the microbial community structure. Water Research 41 (19), 4383e4396.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 7 3 e4 4 8 6
Corominas, L., Rieger, L., Takacs, I., Ekama, G., Hauduc, H., Vanrolleghem, P.A., Oehmen, A., Gernaey, K.V., Van Loosdrecht, M.C.M., Comeau, Y., 2010. New framework for standardized notation in wastewater treatment modelling. Water Science and Technology 61 (4), 841e857. Crocetti, G.R., Hugenholtz, P., Bond, P.L., Schuler, A., Keller, J., Jenkins, D., Blackall, L.L., 2000. Identification of polyphosphate-accumulating organisms and design of 16S rRNA-directed probes for their detection and quantitation. Applied and Environmental Microbiology 66 (3), 1175e1182. Dai, Y., Yuan, Z.G., Wang, X., Oehmen, A., Keller, J., 2007. Anaerobic metabolism of Defluviicoccus vanus related glycogen accumulating organisms (GAOs) with acetate and propionate as carbon sources. Water Research 41 (9), 1885e1896. Dias, J., Oehmen, A., Serafim, L.S., Lemos, P.C., Reis, M.A.M., Oliveira, R., 2008. Metabolic modelling of polyhydroxyalkanoate copolymers production by mixed microbial cultures. BMC Systems Biology 2, 59. Ekama, G.A., Wentzel, M.C., 1999. Difficulties and developments in biological nutrient removal technology and modelling. Water Science and Technology 39 (6), 1e11. Filipe, C.D.M., Daigger, G.T., 1999. Evaluation of the capacity of phosphorus-accumulating organisms to use nitrate and oxygen as final electron acceptors: a theoretical study on population dynamics. Water Environment Research 71 (6), 1140e1150. Filipe, C.D.M., Daigger, G.T., Grady, C.P.L., 2001a. A metabolic model for acetate uptake under anaerobic conditions by glycogen accumulating organisms: stoichiometry, kinetics, and the effect of pH. Biotechnology and Bioengineering 76 (1), 17e31. Filipe, C.D.M., Daigger, G.T., Grady, C.P.L., 2001b. pH as a key factor in the competition between glycogen-accumulating organisms and phosphorus-accumulating organisms. Water Environment Research 73 (2), 223e232. Flowers, J.J., He, S., Yilmaz, S., Noguera, D.R., McMahon, K.D., 2009. Denitrification capabilities of two biological phosphorus removal sludges dominated by different ‘Candidatus Accumulibacter’ clades. Environmental Microbiology Reports 1 (6), 583e588. Garcia Martin, H., Ivanova, N., Kunin, V., Warnecke, F., Barry, K., McHardy, A.C., Yeates, C., He, S., Salamov, A., Szeto, E., Dalin, E., Putnam, N., Shapiro, H.J., Pangilinan, J.L., Rigoutsos, I., Kyrpides, N.C., Blackall, L.L., McMahon, K.D., Hugenholtz, P., 2006. Metagenomic analysis of two enhanced biological phosphorus removal (EBPR) sludge communities. Nature Biotechnology 24 (10), 1263e1269. Hao, X.D., Wang, Q.L., Cao, Y.L., van Loosdrecht, M.C.M., 2010. Experimental evaluation of decrease in the activities of polyphosphate/glycogen-accumulating organisms due to cell death and activity decay in activated sludge. Biotechnology and Bioengineering. doi:10.1002/bit.22703. Kong, Y.H., Xia, Y., Nielsen, J.L., Nielsen, P.H., 2006. Ecophysiology of a group of uncultured gammaproteobacterial glycogenaccumulating organisms in full-scale enhanced biological phosphorus removal wastewater treatment plants. Environmental Microbiology 8 (3), 479e489. de Kreuk, M.K., Picioreanu, C., Hosseini, M., Xavier, J.B., van Loosdrecht, M.C.M., 2007. Kinetic model of a granular sludge SBR: influences on nutrient removal. Biotechnology and Bioengineering 97 (4), 801e815. Kuba, T., Murnleitner, E., vanLoosdrecht, M.C.M., Heijnen, J.J., 1996a. A metabolic model for biological phosphorus removal by denitrifying organisms. Biotechnology and Bioengineering 52 (6), 685e695. Kuba, T., vanLoosdrecht, M.C.M., Heijnen, J.J., 1996b. Effect of cyclic oxygen exposure on the activity of denitrifying phosphorus removing bacteria. Water Science and Technology 34 (1e2), 33e40.
4485
Lopez-Vazquez, C.M., Hooijmans, C.M., Brdjanovic, D., Gijzen, H.J., van Loosdrecht, M.C.M., 2007a. A practical method for quantification of phosphorus- and glycogen-accumulating organism populations in activated sludge systems. Water Environment Research 79 (13), 2487e2498. Lopez-Vazquez, C.M., Song, Y.I., Hooijmans, C.M., Brdjanovic, D., Moussa, M.S., Gijzen, H.J., van Loosdrecht, M.C.M., 2007b. Short-term temperature effects on the anaerobic metabolism of glycogen accumulating organisms. Biotechnology and Bioengineering 97 (3), 483e495. Lopez-Vazquez, C.M., Song, Y.I., Hooijmans, C.M., Brdjanovic, D., Moussa, M.S., Gijzen, H.J., van Loosdrecht, M.C.M., 2008. Temperature effects on the aerobic metabolism of glycogenaccumulating organisms. Biotechnology and Bioengineering 101 (2), 295e306. Lopez-Vazquez, C.M., Hooijmans, C.M., Brdjanovic, D., Gijzen, H.J., van Loosdrecht, M.C.M., 2009a. Temperature effects on glycogen accumulating organisms. Water Research 43 (11), 2852e2864. Lopez-Vazquez, C.M., Oehmen, A., Hooijmans, C.M., Brdjanovic, D., Gijzen, H.J., Yuan, Z.G., Van Loosdrecht, M.C.M., 2009b. Modeling the PAOeGAO competition: effects of carbon source, pH and temperature. Water Research 43 (2), 450e462. Lu, H., Oehmen, A., Virdis, B., Keller, J., Yuan, Z.G., 2006. Obtaining highly enriched cultures of Candidatus Accumulibacter Phosphatis through alternating carbon sources. Water Research 40 (20), 3838e3848. Meijer, S.C.F., van Loosdrecht, M.C.M., Heijnen, J.J., 2002. Modelling the start-up of a full-scale biological phosphorous and nitrogen removing WWTP. Water Research 36 (19), 4667e4682. Moussa, M.S., Hooijmans, C.M., Lubberding, H.J., Gijzen, H.J., van Loosdrecht, M.C.M., 2005. Modelling nitrification, heterotrophic growth and predation in activated sludge. Water Research 39 (20), 5080e5098. Murnleitner, E., Kuba, T., vanLoosdrecht, M.C.M., Heijnen, J.J., 1997. An integrated metabolic model for the aerobic and denitrifying biological phosphorus removal. Biotechnology and Bioengineering 54 (5), 434e450. Oehmen, A., Vives, M.T., Lu, H., Yuan, Z.G., Keller, J., 2005a. The effect of pH on the competition between polyphosphateaccumulating organisms and glycogen-accumulating organisms. Water Research 39 (15), 3727e3737. Oehmen, A., Yuan, Z.G., Blackall, L.L., Keller, J., 2005b. Comparison of acetate and propionate uptake by polyphosphate accumulating organisms and glycogen accumulating organisms. Biotechnology and Bioengineering 91 (2), 162e168. Oehmen, A., Saunders, A.M., Vives, M.T., Yuan, Z.G., Keller, J., 2006. Competition between polyphosphate and glycogen accumulating organisms in enhanced biological phosphorus removal systems with acetate and propionate as carbon sources. Journal of Biotechnology 123 (1), 22e32. Oehmen, A., Lemos, P.C., Carvalho, G., Yuan, Z.G., Keller, J., Blackall, L.L., Reis, M.A.M., 2007a. Advances in enhanced biological phosphorus removal: from micro to macro scale. Water Research 41 (11), 2271e2300. Oehmen, A., Zeng, R.J., Keller, J., Yuan, Z.G., 2007b. Modeling the aerobic metabolism of polyphosphate accumulating organisms enriched with propionate as a carbon source. Water Environment Research 79 (13), 2477e2486. Oehmen, A., Carvalho, G., Freitas, F., Reis, M.A.M., 2010. Assessing the abundance and activity of denitrifying polyphosphate accumulating organisms through molecular and chemical techniques. Water Science and Technology 61 (8), 2061e2068.
4486
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 7 3 e4 4 8 6
Panswad, T., Doungchai, A., Anotai, J., 2003. Temperature effect on microbial community of enhanced biological phosphorus removal system. Water Research 37 (2), 409e415. Pijuan, M., Saunders, A.M., Guisasola, A., Baeza, J.A., Casas, C., Blackall, L.L., 2004. Enhanced biological phosphorus removal in a sequencing batch reactor using propionate as the sole carbon source. Biotechnology and Bioengineering 85 (1), 56e67. Reichert, P., 1994. Aquasim e a tool for simulation and dataanalysis of aquatic systems. Water Science and Technology 30 (2), 21e30. Saito, T., Brdjanovic, D., van Loosdrecht, M.C.M., 2004. Effect of nitrite on phosphate uptake by phosphate accumulating organisms. Water Research 38 (17), 3760e3768. Schuler, A.J., Jenkins, D., 2002. Effects of pH on enhanced biological phosphorus removal metabolisms. Water Science and Technology 46 (4e5), 171e178. Sin, G., Kaelin, D., Kampschreur, M.J., Takacs, I., Wett, B., Gernaey, K.V., Rieger, L., Siegrist, H., Van Loosdrecht, M.C.M., 2008. Modelling nitrite in wastewater treatment systems: a discussion of different modelling concepts. Water Science and Technology 58 (6), 1155e1171. Smolders, G.J.F., Vandermeij, J., Vanloosdrecht, M.C.M., Heijnen, J. J., 1994. Stoichiometric model of the aerobic metabolism of the biological phosphorus removal process. Biotechnology and Bioengineering 44 (7), 837e848. Wang, X.L., Zeng, R.J., Dai, Y., Peng, Y.Z., Yuan, Z.G., 2008. The denitrification capability of cluster 1 Defluviioccus vanusrelated glycogen-accumulating organisms. Biotechnology and Bioengineering 99 (6), 1329e1336.
Xavier, J.B., De Kreuk, M.K., Picioreanu, C., Van Loosdrecht, M.C. M., 2007. Multi-scale individual-based model of microbial and bioconversion dynamics in aerobic granular sludge. Environmental Science & Technology 41 (18), 6410e6417. Yoshida, Y., Kim, Y., Saito, T., Tanaka, K., 2009. Development of the modified activated sludge model describing nitrite inhibition of aerobic phosphate uptake. Water Science and Technology 59 (4), 621e630. Zeng, R.J., Saunders, A.M., Yuan, Z., Blackall, L.L., Keller, J., 2003a. Identification and comparison of aerobic and denitrifying polyphosphate-accumulating organisms. Biotechnology and Bioengineering 83 (2), 140e148. Zeng, R.J., van Loosdrecht, M.C.M., Yuan, Z.G., Keller, J., 2003b. Metabolic model for glycogen-accumulating organisms in anaerobic/aerobic activated sludge systems. Biotechnology and Bioengineering 81 (1), 92e105. Zeng, R.J., Yuan, Z.G., Keller, J., 2003c. Enrichment of denitrifying glycogen-accumulating organisms in anaerobic/anoxic activated sludge system. Biotechnology and Bioengineering 81 (4), 397e404. Zhou, Y., Pijuan, M., Yuan, Z.G., 2007. Free nitrous acid inhibition on anoxic phosphorus uptake and denitrification by polyphosphate accumulating organisms. Biotechnology and Bioengineering 98 (4), 903e912. Zhou, Y., Pijuan, M., Zeng, R.J., Yuan, Z., 2008. Free nitrous acid inhibition on nitrous oxide reduction by a denitrifying-enhanced biological phosphorus removal sludge. Environmental Science & Technology 42 (22), 8260e8265.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Molecular detection of pathogens in water e The pros and cons of molecular techniques Rosina Girones a,*, Maria Antonia Ferru´s b, Jose´ Luis Alonso c, Jesus Rodriguez-Manzano a, Byron Calgua a, Adriana de Abreu Correˆa a,d, Ayalkibet Hundesa a, Anna Carratala a, Sı´lvia Bofill-Mas a a
Department of Microbiology, Faculty of Biology, University of Barcelona. Av. Diagonal 645, 08028 Barcelona, Spain Department of Biotechnology, Polytechnic University of Valencia, Camino de Vera 14, 46022 Valencia, Spain c Institute of Water Engineering and Environment, Polytechnic University of Valencia, Camino de Vera 14, 46022 Valencia, Spain d Department of Microbiology and Parasitology, Laboratory of Applied Virology, Federal University of Santa Catarina, 88040-900 Floriano´polis, Santa Catarina, Brazil b
article info
abstract
Article history:
Pollution of water by sewage and run-off from farms produces a serious public health problem
Received 1 February 2010
in many countries. Viruses, along with bacteria and protozoa in the intestine or in urine are
Received in revised form
shed and transported through the sewer system. Even in highly industrialized countries,
10 June 2010
pathogens, including viruses, are prevalent throughout the environment. Molecular methods
Accepted 14 June 2010
are used to monitor viral, bacterial, and protozoan pathogens, and to track pathogen- and
Available online 19 June 2010
source-specific markers in the environment. Molecular techniques, specifically polymerase chain reaction-based methods, provide sensitive, rapid, and quantitative analytical tools with
Keywords:
which to study such pathogens, including new or emerging strains. These techniques are used
Pathogen
to evaluate the microbiological quality of food and water, and to assess the efficiency of virus
Water
removal in drinking and wastewater treatment plants. The range of methods available for the
Virus
application of molecular techniques has increased, and the costs involved have fallen. These
Protozoa
developments have allowed the potential standardization and automation of certain tech-
Bacteria
niques. In some cases they facilitate the identification, genotyping, enumeration, viability
PCR
assessment, and source-tracking of human and animal contamination. Additionally, recent improvements in detection technologies have allowed the simultaneous detection of multiple targets in a single assay. However, the molecular techniques available today and those under development require further refinement in order to be standardized and applicable to a diversity of matrices. Water disinfection treatments may have an effect on the viability of pathogens and the numbers obtained by molecular techniques may overestimate the quantification of infectious microorganisms. The pros and cons of molecular techniques for the detection and quantification of pathogens in water are discussed. ª 2010 Elsevier Ltd. All rights reserved.
Abbreviations: MST, microbial source-tracking; HAdV, human adenoviruses; HAV, hepatitis A virus; HEV, hepatitis E virus; JCPyV, human polyomavirus JC; BKPyV, human polyomavirus BK; PCR, polymerase chain reaction; qPCR, quantitative PCR; qRT-PCR, quantitative reverse transcriptase PCR; NASBA, acid sequence-based amplification; CFU, colony-forming units; mPCR, multiplex PCR; IFAs, immunofluorescent assays; IMS, immunomagnetic separation; RT-PCR, reverse transcriptase PCR; mRNA, messenger RNA; PAdV, porcine adenoviruses; BPyV, bovine polyomaviruses; EMA, ethidium monoazide; PMA, propidium monoazide; VBNC, viable non-culturable; PBS, phosphate buffered saline; nPCR, nested-PCR. * Corresponding author. Tel.: þ34 93 402 1483; fax: þ34 93 403 9047. E-mail address:
[email protected] (R. Girones). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.030
4326
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
Contents 1. 2.
3.
4. 5.
6. 7. 8.
1.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pathogens in water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1. Viruses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2. Bacteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3. Protozoa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Available molecular techniques and applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Molecular techniques for the analysis of viruses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Molecular techniques for the analysis of bacteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Molecular techniques for the analysis of protozoa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Human and animal viruses as faecal indicators and MST tools . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Molecular techniques in the microbiological control of water quality: The pros and cons . . . . . . . . . . . . . . . . . . . . . . . . 5.1. The pros of molecular techniques applied to the detection of pathogens in water . . . . . . . . . . . . . . . . . . . . . . . . . 5.2. The cons of molecular techniques applied to the detection of pathogens in water . . . . . . . . . . . . . . . . . . . . . . . . . Viability assays and molecular detection methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Future developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Introduction
Significant numbers of human microbial pathogens are present in urban sewage and may be considered environmental contaminants. Although most pathogens can be removed by sewage treatment, many are discharged into the effluent and enter receiving waters. Point-source pollution enters the environment at distinct locations, through a direct route of discharge of treated or untreated sewage. Non-point sources of contamination are of significant concern with respect to the dissemination of pathogens and their indicators in the water systems. They are generally diffuse and intermittent and may be attributable to the run-off from urban and agricultural areas, leakage from sewers and septic systems, and sewer overflows (Stewart et al., 2008). Molecular methods are used to monitor viral, bacterial and protozoan pathogens, and to track pathogenand source-specific markers in the environment. Classic microbiological indicators such as faecal coliforms, E. coli and Enterococci are the indicators most commonly analyzed to evaluate the level of faecal contamination. They are also used to assess the efficiency of pathogen removal in water purification processes. However, whether these bacteria are suitable indicators of the occurrence and concentration of human viruses and protozoa cysts has been questioned (Lipp et al., 2001; Tree et al., 2003; We´ry et al., 2008). Indicator bacteria are more sensitive to inactivation through treatment processes and by sunlight than viral or protozoan pathogens (Hurst et al., 2002; Sinclair et al., 2009). Other limitations have been associated with their application: short survival compared to pathogens (McFeters et al., 1974), non-exclusive faecal source (Scott et al., 2002; Simpson et al., 2002), ability to multiply in some environments (Solo-Gabriele et al., 2000; Pote et al., 2009), inability to identify the source of faecal contamination (point and non-point) (Field et al., 2003) and low correlation with the presence of pathogens (Pina et al., 1998; Horman et al., 2004;
4326 4326 4326 4327 4327 4328 4329 4329 4330 4331 4332 4333 4333 4333 4334 4335 4335 4335
Savichtcheva and Okabe, 2006). As a result, none of the bacterial indicators currently used meets all established criteria for water quality. Thus in certain cases, such as drinking or bathing water, direct analysis of specific pathogens of concern is considered to be a more suitable alternative. Source water contamination by Cryptosporidium and Giardia presents a particular challenge to water-quality managers for several reasons. These include the ubiquity of protozoa in wastewater effluents (Carey et al., 2004), the widespread infection of domestic animals and wildlife (Fayer, 2004), the resistance of protozoans, especially Cryptosporidium, to traditional disinfection methods (Steiner et al., 1997), and the uncertain relationship between the presence of protozoans and faecal indicator bacteria typically used in water quality monitoring (Chauret et al., 1995; Cizek et al., 2008; Keeley and Faulkner, 2008). Molecular techniques, specifically nucleic acid amplification procedures, provide sensitive, rapid and quantitative analytical tools for detecting specific pathogens, including new emergent strains and indicators. They are used to evaluate the microbiological quality of food and water, the efficiency of virus removal in drinking and wastewater treatment plants, and as microbial source-tracking (MST) tools (Albinana-Gimenez et al., 2009b; Field et al., 2003; Hundesa et al., 2006).
2.
Pathogens in water
2.1.
Viruses
The list of potentially pathogenic viruses present in urban sewage includes the DNA viruses, adenovirus and polyomavirus, and RNA viruses such as enterovirus, hepatitis A and E viruses, norovirus, rotavirus and astrovirus. Human
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
adenoviruses (HAdV) and polyomaviruses exhibit a high prevalence in all geographical areas studied (Pina et al., 1998; Bofill-Mas et al., 2000). Enteroviruses, noroviruses, rotavirus, and astroviruses have been described as showing diverse prevalence levels, depending on the time of year and the presence of outbreaks in the population. The presence of hepatitis A virus (HAV) varies in different geographical areas, but it is frequently detected in urban sewage in endemic areas throughout the year. Hepatitis E virus (HEV), like HAV, is more abundant in countries where sanitation is poor. Autochthonous strains of HEV have been reported in urban sewage in several highly industrialized countries, as well as related cases of sporadic acute hepatitis caused by these non-imported strains (Pina et al., 2000; Clemente-Casares et al., 2003). Strains of HEV may also infect pigs, wild boar and deer, and they are frequently detected in both slaughterhouse sewage, where pigs are treated, and urban sewage in areas in Europe that were considered non-endemic (Rodriguez-Manzano et al., 2010). Many of the viruses that are considered to be water contaminants produce primarily sub-clinical infections, causing symptoms only in a small proportion of the infected population. Enteroviruses are a good example, although they may cause a wide diversity of clinical syndromes, including diseases affecting the central nervous system. The syndromes produced by viral infections range from respiratory diseases, frequently associated with HAdV, to life-threatening conditions, such as acute hepatitis caused by HEV and HAV in adults. Other infections include severe gastroenteritis in small children and the elderly, commonly related to rotavirus, or adenovirus and norovirus respectively (Hart et al., 2009). Disease progression depends on the route of infection, the infective dose of the viral agent, the age, health, immunological, and nutritional status of the infected individual (pregnancy, presence of other infections or diseases), and the availability of health care. Viruses that are transmitted via contaminated food or water are typically stable because they lack the lipid envelopes that render other viruses more susceptible to environmental agents and because survive in the digestive track. Some viruses, such as human polyomaviruses JC (JCPyV) and BK (BKPyV) and some HAdV, infect humans during childhood, thereby establishing persistent infections. In recreational water-borne diseases, noroviruses are believed to be the single largest cause of documented outbreaks, followed by adenovirus (Sinclair et al., 2009).
2.2.
Bacteria
Salmonella and Campylobacter are the most frequent agents of bacterial gastroenteritis (Westrell et al., 2009). Salmonella is isolated from water in lower numbers than indicator bacteria such as faecal coliforms, faecal streptococci and enterococci, which are several orders of magnitude higher (Sidhu and Toze, 2009). However, low numbers (15e100 colony-forming units [CFU]) of Salmonella in water may pose a public health risk (Jyoti et al., 2009). Thermophilic Campylobacter species are widespread in the environment and are commonly found in surface water and sewage sludge (Sahlstro¨m et al., 2004). Other frequent waterborne pathogens are Shigella, Yersinia or Vibrio cholerae, with outbreaks linked to contaminated water and seafood (Sharma et al., 2003). Legionella pneumophila has a complex aquatic life
4327
cycle that strongly affects its state of activity. L. pneumophila is a ubiquitous bacterium in natural aquatic environments that can also persist in human-controlled systems containing water, such as air conditioning and plumbing infrastructures. Intracellular growth in protozoa can permit this pathogen to survive chlorination, and the generation of aerosols in these systems contributes to the transmission to humans where infection of alveolar macrophages results in respiratory illness (McDade et al., 1977; Steinert et al., 2002). In addition to these well-known water-borne pathogens, there are groups of bacteria, some of them considered as “emergent pathogens” that are now also regarded as being transmitted by the water route. E. coli O157:H7 is a faecal pathogen frequently isolated from waters (Bavaro, 2009). It has been found in 2% of raw sludges, but the numbers of E. coli O157:H7 in sewage and its survival during wastewater treatment are unknown (Sidhu and Toze, 2009). Listeria monocytogenes is a ubiquitous bacterium, isolated from a wide range of environmental sources, including soil, water, effluents, a variety of foods, and the faeces of humans and animals (Barbuddhe and Chakraborty, 2009). Recent outbreaks demonstrated that L. monocytogenes can also cause gastroenteritis in healthy individuals, and more severe invasive disease in immunocompromised patients (Barbuddhe and Chakraborty, 2009; Wilkes et al., 2009). Vibrio vulnificus is an opportunistic human pathogen that may cause gastroenteritis, severe necrotizing soft-tissue infections and primary septicaemia, with a high lethality rate. Illness is associated with ingestion of seafood or exposure to contaminated water. V. vulnificus has been recovered from fish, shellfish, water and sediments (Harwood et al., 2004). Recently, it has been isolated from wastewater samples (Igbinosa et al., 2009). Helicobacter pylori is an etiological agent of gastritis, and peptic and duodenal ulcers. In addition, infection is a recognized risk factor in the development of gastric mucosa-associated lymphoid tissue lymphoma and adenocarcinoma. H. pylori is present in surface water and wastewater (Queralt et al., 2005). Biofilms in drinking-water systems have been reported as possible reservoirs of H. Pylori (Park et al., 2001). Attempts to culture H. pylori cells from environmental water samples have been largely unsuccessful, and its ability to survive in an infectious state in the environment remains controversial. Due to the fastidious nature of the bacterium and the lack of standard culture methods for environmental samples, very few quantitative studies have been reported (Percival and Thomas, 2009). Arcobacter spp., which are considered to be emerging pathogens that cause diseases in domestic animals and diarrhoea in humans, are frequently isolated from animal food products, in particular from poultry, as well as various types of water such as groundwater, surface water, raw sewage and seawater (Ho et al., 2006). Various studies have concluded that water may also play an important role in the transmission of Arcobacter spp. and strongly suggest a faecaleoral route of transmission to humans and animals (Gonza´lez et al., 2007).
2.3.
Protozoa
A number of different types of pathogens, such as Cryptosporidium, Entamoeba, Cyclospora, Toxoplasma, Microsporidia and Giardia, among others, can be present in contaminated water
4328
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
(Smith et al., 2004). Among protozoan genera Cryptosporidium and Giardia are known to be highly resistant to environmental stress (Caccio`, 2003). Cryptosporidium has been detected in many drinking water sources and is considered an important water-borne contaminant (Xiao and Ryan, 2008). Infection occurs following ingestion of oocyst-contaminated food, drinking water, or recreational water. There are 20 valid Cryptosporidium species and over 40 genotypes of this parasite. Of these, Cryptosporidium parvum and Cryptosporidium hominis cause over 95% of the reported causes of human cryptosporidiosis (Xiao and Ryan, 2008). Giardia is a frequent water contaminant. Of the morphologically defined Giardia species, only Giardia duodenalis (syn. Giardia intestinalis or Giardia lamblia), has been recovered from humans and a wide variety of other mammals (van der Giessen et al., 2006). The major sources of surface water contamination with G. duodenalis are discharges of treated or untreated sewage, run-off or discharges of manure from agricultural lands and, in more pristine waters, wildlife. Concentrations of cysts as high as 88,000 per litre in raw sewage and 240 per litre in surface water have been reported (Wallis et al., 1996). Cysts are robust and can survive for weeks to months in freshwater. Cryptosporidium spp. and G. duodenalis are common food- and water-borne protozoa that affect humans and a wide range of domestic and wild animals (Fayer, 2004). These parasites are among the major causal agents of diarrhoeal disease in humans and animals worldwide, and they can even potentially shorten the life span of immunocompromised hosts (Smith et al., 2007; Reynolds et al., 2008). Giardia spp. cysts are more resistant to chlorine than enteric bacteria but they are not as resistant as Cryptosporidium. The number of parasites required to induce infection has been estimated to be as low as 10 Cryptosporidium spp. oocysts (Fayer et al., 2000) or 10 G. duodenalis cysts (Adam, 2001).
3. Available molecular techniques and applications Culturing pathogens is a laborious procedure, involving enrichment and selective media in an attempt to isolate pathogens from background bacteria. It is often difficult to achieve appropriate enrichment, which makes the work even more tedious. Moreover, concentrations may be too low for cultural detection but still be high enough to cause infection. Therefore, a molecular detection method is needed, since such methods are highly specific and sensitive. The methods used are typically based on the detection and quantification of specific segments of the pathogen’s genome (DNA or RNA). In order to reach the detection level, the specific segments are subjected to in vitro amplification. These methods allow researchers to rapidly and specifically detect microorganisms of public health concern. Additionally, recent improvements have allowed simultaneous detection of several microorganisms in a single assay (Maynard et al., 2005; Straub et al., 2005; Marcelino et al., 2006). The range of protocols available for the application of molecular techniques has increased over the last few years, and the costs involved, although still significant, have fallen.
These developments have allowed the potential standardization and automation of some of these techniques. In some cases they facilitate the identification, genotyping, enumeration, viability assessment, and source-tracking of human and animal contamination if host-specific highly prevalent pathogens are analyzed. Direct monitoring of pathogens has enjoyed wider application since the development of molecular technologies. However, the molecular techniques available today are being continuously refined in order to be standardized and make them applicable to a diversity of matrices, to increase their sensitivity, and to reduce the time and steps required in the analytical process. The standardization and validation of protocols is considered a very important requirement for the implementation of molecular techniques either in the clinical or in the environmental field and has a major impact on the evaluation of the data produced in the diverse studies (Raymaekers et al., 2009; Bustin, 2010; Doring et al., 2008; Harwood et al., 2009; Sen et al., 2007). External evaluation programs for the detection and quantification of many pathogens using molecular methods are already being routinely established in many laboratories in the clinical field. Although several validation studies have been developed for water-borne pathogens using molecular methods (Conraths and Schares, 2006; Cheng et al., 2009) there is still the need for validation assays for most of the pathogens and indicators in water samples. These validation assays should involve diverse laboratories and selected methodologies applicable in standard protocols evaluating inter and intra-laboratory variability and providing robust information on the efficiency of the methods for a diversity of pathogens and matrices. The question of suitable controls has also been discussed in reference to PCR assays. In addition to the specific positive and negative controls in PCR reactions, other controls are required. Known quantities of DNA are used as internal or external controls in the reactions, and neat and diluted samples are tested to evaluate the presence of potential inhibitors that may affect the accuracy of the quantification. Also the use of an affordable process control will be required in order to demonstrate that the concentration and extraction protocols worked correctly for every assay. Most methods utilize the following steps: (i) Concentration of the organism of interest from the environmental water sample into a suitable volume (if necessary); (ii) Extraction of the RNA or DNA from the target organism; (iii) Amplification of the genomic segment(s) chosen; (iv) Detection (or quantification) of the amplified genomic segment(s). Most applied molecular techniques are based on protocols of nucleic acid amplification, of which the polymerase chain reaction (PCR) is the most commonly used. Quantitative PCR (qPCR) is rapidly becoming established in the environmental sector. qPCR is, in many cases, more sensitive than either the bacterial culture method or the viral plaque assay (He and Jiang, 2005). Quantitative reverse transcriptase PCR (qRTPCR) uses RNA as a template molecule. qPCR commonly uses fluorescent dyes such as SYBR green for the detection of the amplified segment. However, molecular beacons or other fluorescent probes such as TaqMan assays (Applied Biosystems, Foster City, CA, USA), Scorpion primers (PREMIER Biosoft International, Palo Alto, CA, USA) or that used in the LightCycler (Roche, Indianapolis, Indiana, USA) lead to higher
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
specificity based on the use of complementary primers and probes for the quantification of the selected genome segment. The use of qPCR is extending, and is under consideration for monitoring the environment, water and food. Besides PCR, other methods are available to amplify nucleic acids, for example Nucleic Acid Sequence-Based Amplification (NASBA), an isothermal method designed to amplify RNA from either RNA or DNA templates, although it is most commonly used to amplify RNA (Cook, 2003; Goodwin and Litaker, 2008).
3.1.
Molecular techniques for the analysis of viruses
Practically all viral pathogens found in environmental waters e rivers, lakes, seawater and groundwater e originate from contamination with wastewater or directly with human or animal excreta. Most of the more than one hundred species of viruses of faecal origin cannot be detected with the conventional cell-culture methods, or their detection efficiency is very poor. For this reason, molecular methods for viruses in water have rapidly found their way in the analysis of the environment, and data are accumulating on the use of qPCR techniques to assess the presence and concentration of viruses in water. Even though PCR is a very sensitive detection technique, the amount of viruses found in many environmental waters is low and the viral particles must be subjected to a concentration step before the PCR can be attempted. A wide range of methodologies for the concentration and detection of viruses from water using molecular techniques have been tested. The techniques most commonly used are based on adsorptioneelution-based virus concentration protocols with various filters or glass wool columns and also on ultrafiltration (Donaldson et al., 2002; Haramoto et al., 2005; Rajal et al., 2007; Lambertini et al., 2008; Albinana-Gimenez et al., 2009a). The concentration of viruses by adsorptionelution-based protocols is less efficient in seawater than in freshwater, perhaps as a result of the high ionic strength in seawater. Alternative techniques are based on direct flocculation methods (Calgua et al., 2008) or ultrafiltration procedures (Jiang et al., 2001). Examples of the concentrations of viruses that have been found in sewage, freshwater and seawater by diverse concentration procedures and molecular methods are presented in Table 1.
3.2.
Molecular techniques for the analysis of bacteria
Classical microbiological quantification methodology relies on the cultivation of specific bacteria in appropriate culture media and on their further biochemical or immunological characterisation. For many well known and emerging pathogens, appropriate culture methods for environmental samples and biochemical schemes for valid identification at species level are lacking (Dong et al., 2008). On the other hand, target bacteria might be embedded in biofilms and not be accessible to the standard techniques. After prolonged exposure to water, bacterial pathogens might enter a viable but non-culturable stage, in which they cannot be detected by culture, although they retain their infective potential (CenciariniBorde et al., 2009). For these reasons, several studies have produced data on the quantification of bacteria in water by
4329
using qPCR techniques. However, molecular protocols, unlike traditional culture-based methods, do not distinguish between viable and non-viable organisms and although some approaches have been developed for discriminating damaged pathogens or naked nucleic acid from intact microorganisms, there is still the need for more information before replacing the current conventional methods by molecular ones. Molecular techniques for the specific detection and quantification of bacterial pathogens also offer several advantages over conventional methods: high sensitivity and specificity, speed, ease of standardization and automation. As with the viruses, direct PCR amplification of some bacterial pathogens from water samples is difficult due to the presence of only low numbers of these bacteria in environmental sources. Therefore, an enrichment step is usually required prior to performing a PCR (Noble and Weisberg, 2005). Quantitative PCR analysis of aquatic DNA provides the number of genomes per volume of water of a specific pathogen. This value is not identical with the number of cells that can be determined by microscopic techniques, such as fluorescent in situ hybridization (FISH) or immunofluorescence, or the CFUs obtained by plate counting. Bacterial pathogens can produce aggregates, many bacteria contain more than one genome per cell, depending on their physiological state, and also several of the gene markers used for detection, such as the rRNA genes, may be present in multiple copies in the bacterial genome. These features entail the application of different conversion factors from genome copies to cells for different pathogens, and such factors need to be determined for each targeted bacterial species individually (Brettar and Ho¨fle, 2008). Examples of the concentrations of bacterial pathogens that have been found in water by qPCR are presented in Table 2. The FISH technique, based on hybridization with rRNA oligonucleotide probes, has been used for the detection and identification of different microorganisms in mixed populations. This technique is considered a powerful tool for phylogenetic, ecological, diagnostic and environmental studies in microbiology (Bottari et al., 2006). It helps to reveal mechanisms of survival and infection at the cellular level (Brettar and Ho¨fle, 2008) and in the study of biofilms (Juhna et al., 2007). FISH has been applied with this aim for the detection of emerging pathogens from water, sewage and sludge (Gilbride et al., 2006). This method also allows for the detection of viable but non-culturable forms. However the methodology is still limited by a lack of sensitivity and enrichment steps are often required, either cultural preenrichment or a magnetic bead type enrichment, or both. An example of multiplex PCR (mPCR) assays developed for the detection of bacterial pathogens are a multiplex SYBR green I-based PCR assay developed for simultaneous detection of Salmonella serovars and L. monocytogenes (Jothikumar et al., 2003). Another quadruplex qPCR assay for detection and differentiation of O1, O139, and non-O1, non-O139 strains of V. cholerae and for prediction of their toxigenic potential was developed by Huang et al. (2009). L. monocytogenes has also been investigated in biofilms using qPCR techniques (Guilbaud et al., 2005). L. pneumophila is not transmitted through the oral route, however, it is also considered a water-borne pathogen and is the most commonly reported etiologic agent of legionellosis. It has an impact on public health in developed countries, as
4330
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
Table 1 e Examples of the concentration of viruses found in sewage, freshwater and seawater by qPCR. Results are expressed in genome copy logs (GC logs). Virus Adenovirus
Type of sample
Collection site
Concentration
% Positive samples
Sewage (raw) Sewage (secondary effluent) Biosolids River water
Spain
4e7 GC logs/100 ml 3 GC logs/100 ml
100% 100%
4e7 GC logs/100 g 1e4 GC logs/l
100% 90%
Seawater USA
4e5 GC logs/100 ml 3e4 GC logs/100 ml
Adenovirus 40, 41
River water
Japan
3e5 GC logs/l
JC Polyomavirus
Sewage (raw) Biosolids River water
Spain
5 GC logs/100 ml 3e5 GC logs/100 g 0e3 GC logs/l
Sewage (raw) Sewage (secondary effluent)
Brazil
River water
Japan
Sewage Sewage Sewage Sewage Sewage
(raw) (raw) (raw) (raw) (raw)
GII GI
Sewage (effluent)
France France Spain Spain United Kingdom Brazil
Reference Bofill-Mas et al. (2006)
Albinana-Gimenez et al. (2009b) Calgua et al. (2008) qPCR
Fong et al. (2010)
61%
qPCR
Haramoto et al. (in press)
100% 100% 90%
qPCR
Bofill-Mas et al. (2006)
4e7 GC logs/100 ml 4e5 GC logs/100 ml
96% 39%
qPCR
Fumian et al. (in press)
2e3 GC logs/l
11%
qPCR
Haramoto et al. (in press)
qRT-PCR qPCR qPCR qPCR qPCR
Le Cann et al. (2004) Schvoerer et al. (2001) Rodriguez-Manzano et al. (2010) Rodriguez-Manzano et al. (2010) Laverick et al. (2004)
qPCR
Victoria et al. (in press)
5e7 GC logs/100 ml 7 GC logs/100 ml 4 GC logs/100 ml 3 GC logs/100 ml 6 GC logs/100 ml
2e3 GC logs/l 2 GC logs/l
demonstrated by the many major outbreaks reported over the past years. Molecular techniques based on qPCR have been described for the detection of pathogenic strains of L. pneumophila (Morio et al., 2008). Due to the ubiquitous nature of L. pneumophila bacterium in the environment, molecular typing methods are needed both to determine the relatedness of outbreak strains and to identify the source of the outbreak.
3.3.
qPCR
1e3 GC logs/l
Sewage (raw) Sewage (tertiary effluent)
Astrovirus Enterovirus Hepatitis A virus Hepatitis E virus Norovirus
Quantification method
Molecular techniques for the analysis of protozoa
The identification of Cryptosporidium and Giardia (oo)cysts in environmental samples is largely achieved by the use of immunofluorescent assays (IFAs) after concentration processes using U.S. EPA method 1623 and its equivalents in other countries (USEPA, 2001). These microscopic-based methods produce total counts of live and dead Cryptosporidium oocysts in water samples, without distinguishing species or genotypes that can infect humans from those that cannot (Brescia et al., 2009). There are currently four methods by which oocyst viability can be assessed including (Millar et al., 2002): (i) animal infectivity, (ii) in vitro excystation, (iii) the exclusion/inclusion of vital fluorogenic dyes, and (iv) Reverse transcriptase PCR (RT-PCR). Given the limitations in the specific detection of Cryptosporidium and Giardia using microscopy, immunological
100%
Albinana-Gimenez et al. (2009b)
and/or flow cytometric methods, a range of nucleic acid-based methods have been developed and evaluated for the identification of species, the detection of genetic variation within and among species from faecal, environmental or water samples, and the diagnosis of cryptosporidiosis and giardiasis. Some methods rely on the in situ hybridization of probes to particular genetic loci within Cryptosporidium oocysts or Giardia cysts, whereas most rely on the specific amplification of one or more loci from small amounts of DNA by the PCR. The introduction of molecular techniques, in particular those based on the amplification of nucleic acids, has provided researchers with highly sensitive and specific assays for the detection and quantification of protozoa. Because Giardia and Cryptosporidium usually occur in low concentrations, detection methods should also include a procedure for concentrating the organisms from large-volume water samples. Filtration, flocculation, flow cytometry, immunomagnetic separation (IMS) and immunofluorescence with monoclonal antibodies together constitute the currently accepted methodology for detecting protozoa in drinking water (Slifko et al., 2000). These approaches are also used for testing raw and treated waters, although PCR-based procedures are increasingly being used in quality control (Thompson et al., 2004). In addition, molecular techniques can provide genotypic characterisation of the parasites isolated from water, thus helping to identify the
4331
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
Table 2 e Examples of the concentration of bacterial pathogens found in water by qPCR. Results are expressed as mean values of genome copy logs (GC logs) or cell logs. It must be considered that most data is approximate because some of the studies had only one or two positive results or were tested after rainfall events or by using standard curves of different bacterial strains. Bacteria
Quantitative assay
E. coli
Taq-Man
Enterococos
Taq-Man Taq-Man
Salmonella Campylobacter jejuni Arcobacter sp. Enterohemorragic E. coli
SYBR SYBR SYBR SYBR
V. vulnificus L. monocytogenes
Taq-Man Taq-Man
H. pylori Clostridium perfringens
SYBR green Taq-Man
V. cholera, Shigella dysenteriae, Salmonella thyphimurium and E. coli
SYBR green
green green green green
Type of samples
Concentration
Sewage (raw) Sewage (effluent) Lake water Sewage (raw) Sewage (effluent) Surface water Surface water Sewage (raw) Surface water
Seawater (estuarine) Sewage (raw) Sewage (effluent) Sewage Sewage (raw) Sewafe (effluent) River water (low level of pollution) River water (high level of pollution) 2 Urban lakes
source of contamination (Thompson et al., 2004). Molecular tools have been developed to detect and differentiate Cryptosporidium at the species/genotype and subtype levels (Xiao and Ryan, 2004; Caccio`, 2005), enabling researchers to perform more accurate risk assessment of environmental and drinking water contamination (Xiao and Ryan, 2008). Recently, a twocolor FISH assay, based on species-specific probes for C. parvum (Cpar 677 probe) and C. hominis (Chom253 probe), has been shown to distinguish between the two major species involved in human infections (Alagappan et al., 2009). The FISH assay is subject to the same limitations of sample recovery efficiency and purification losses that affect other methods (Jakubowski et al., 1996) and also the physiological condition of the (oo)cysts affects the fluorescence intensity of FISH-positive (oo)cysts. Quantitative RT-PCR is used to detect messenger RNA (mRNA), which is present only in viable organisms since its stability is short, often only a few seconds to minutes (Baeumner et al., 2001). Therefore, mRNA is an optimal target molecule if viable organisms need to be distinguished from non-viable organisms. However, not all mRNAs are produced throughout the life of an organism (Baeumner et al., 2001). qPCR protocols have been developed for the detection and identification of Cryptosporidium and Giardia species/genotypes from water samples. Examples of the molecular targets and sensitivities of the assays that have been used in water samples are presented in Table 3. Finally, other molecular assays, such as NASBA assays, have also been described for the detection of some bacterial pathogens and parasites in the environment and food (Baeumner et al., 2001; Caccio`, 2003; Cook, 2003; Thompson et al., 2003). The quantification of protozoa abundance in water by molecular methods has been described by several authors.
7 GC logs/100 ml 3 GC logs/100 ml 3 GC logs/100 ml 4 GC logs/100 ml 3 GC logs/100 ml 2 GC logs/100 ml 1 GC logs/100 ml 2e6 cell logs/100 ml 14/32 positives samples, levels under 10 GC/100 ml (assay quantification limit) Between 2 and 91 cells/100 ml 3 GC logs/100 ml Absence 2e3 GC logs/100 ml 5 GC logs/100 ml 3 GC logs/100 ml 1.8 GC logs/100 ml
Reference Shannon et al. (2007) Lavender and Kinzelman (2009) Shannon et al. (2007) Ahmed et al. (2009) Ahmed et al. (2009) Gonza´lez et al. (unpublished) Ahmed et al. (2009)
Wetz et al. (2008) Shannon et al. (2007) Nayak and Rose (2007) Shannon et al. (2007) Liu et al. (2009)
3.7 GC logs/100 ml 2.7e3.3 GC logs/100 ml
Guy et al. (2003) detected cyst concentrations ranging from 2653 to 13,408/l in sewage samples from wastewater treatment plants in Laval (Quebec) with TaqMan probe qPCR. Anceno et al. (2007) detected cyst concentrations ranging from 243 to 4103/l with SYBR green qPCR in an irrigation canal receiving discharges from a wastewater treatment plant in the periurban area of Thailand. Bertrand et al. (2004) detected cyst concentrations ranging from 250 to 2300/l in sewage influent samples from a wastewater treatment plant in Nancy (France) with qPCR.
4. Human and animal viruses as faecal indicators and MST tools The high stability of viruses in the environment, host-specificity and the high prevalence of some viral infections throughout the year in the world population strongly support the use of qPCR techniques for the identification and quantification of specific viruses that can be used as indicators of faecal contamination and as MST. HAdV and the JCPyV have been suggested as indicators of human faecal contamination, given their high prevalence in all the geographical areas studied as indicators of human faecal contamination (Puig et al., 1994; Pina et al., 1998; Bofill-Mas et al., 2000). The identification of faecal microbial contamination and their sources in the environment, water and food, plays a very important role in enabling effective management and remediation strategies. MST includes a group of methodologies that aim to identify, and in some cases quantify, the dominant sources of faecal contamination in the environment and, more specifically, in water resources (Field, 2004; Fong and Lipp, 2005). Molecular methods based on molecular detection of host-specific strains of bacteria from the
4332
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
Table 3 e qPCR assays for the detection and quantification of Cryptosporidium and Giardia in environmental samples. The genomic targets, the pathogen(s) detected, the sensitivity, and the nature of the investigated samples are indicated (adapted from Caccio`, 2003). Type of assay
Genomic target
Pathogen(s) detected
TaqMan
Oocyst wall protein gene (COWP)
TaqMan
Anonymous fragment (AF118110) b-giardin Elongation factor 1A 18S rRNA
C. hominis, C. parvum, Cryptosporidium meleagridis, Cryptosporidium wrairi C. hominis, C. parvum, C. meleagridis G. duodenalis (Assemblages A and B) G. duodenalis, Giardia ardae Cryptosporidium spp.
1 Cyst 1 Oocyst
Triose-phosphate isomerase
G. duodenalis (Assemblages A and B)
3 Cysts Assemblage A 4.5 cysts Assemblage B
TaqMan TaqMan Quenching probe TaqMan
Sensitivity 1 Oocyst
Environmental samples (water and sewage) River water
8 Oocysts 1 Cyst
Bacteroidales order and the genus Bifidobacterium (Bernhard and Field, 2000a,b; Harwood et al., 2009; Dorai-Raj et al., 2009) have been widely used. The quantification by PCR of DNA viruses has some advantages in relation to the quantification of RNA viruses, such as lower cost and lower sensitivity to inhibitors potentially present in the sample. In a study by Harwood et al. (2009), three laboratories evaluated library-independent MST methods for human sewage detection via conventional PCR: human-associated Bacteroidales, human polyomaviruses, and Methanobrevibacter smithii, and the results showed that human polyomaviruses were the most specific human marker for MST. The evaluation of the correlation between indicators and specific pathogens would require further studies. It has been shown that HAdV were the viruses most frequently detected throughout the year, and most of the samples that were positive for enteroviruses or HAVs were also positive for HAdV (Pina et al., 1998). The presence of human adenovirus or JCPyVs in water is an indication of faecal/urine contamination and potential risk for the presence of faecal/urine pathogens. However, the excretion patterns of some specific pathogens such as noroviruses or rotaviruses are different, and their dissemination as environmental contaminants in water change depending of the period of the year (higher in cold months in temperate areas) and the potential presence of outbreaks in the population. It is then clear that human adenoviruses and polyomaviruses may be considered a useful marker of human contamination but it is also clear that, in some cases and specific locations, the numbers of viruses related to specific outbreaks, such as rotaviruses, may exceed
Nature of the sample
References Guy et al. (2003)
Environmental samples (water and sewage) Wastewater River water Wastewater
Fontaine and Guillot (2003) Guy et al. (2003) Bertrand et al. (2004) Masago et al. (2006) Bertrand and Schwartzbrod (2007)
the numbers of human adenovirus in this specific environment (Miagostovich et al., 2008). Specific animal viruses have also been proposed as MST tools such as porcine adenoviruses (PAdV) and bovine polyomaviruses (BPyV) (Bofill-Mas et al., 2000, 2006; Formiga-Cruz et al., 2003; Hundesa et al., 2006; Hundesa et al., 2009a,b), bovine enterovirus and teschoviruses (Jimenez-Clavero et al., 2003, 2005). PAdVs and BPyV are disseminated widely in the swine and bovine population respectively but they do not produce clinically severe diseases. A summary of the concentration values reported for BPyV and PAdV in environmental samples is shown in Table 4.
5. Molecular techniques in the microbiological control of water quality: The pros and cons PCR techniques such as qPCR and qRT-PCR using specific probes, with a high level of sensitivity and specificity, represent rapid, cost-effective tools that generate significant information on the presence, quantity and distribution of classical and new emergent pathogens in water and food. For this reason molecular technologies for water quality analysis are widely used for the identification of pathogens in water. The final goal of molecular-based technologies is not simply to speed up slow culture methods, but to achieve an increase in the accuracy and sensitivity of the detection methods. To this end, great attention is being paid to the robustness of the
Table 4 e Results obtained in quantitative PCR (qPCR) assays for Bovine polyomaviruses (BPyV) and Porcine adenoviruses (PAdV) in different types of environmental samples. Results expressed as genome copies (GC) per l. qPCR assays were performed according to Hundesa et al. (2009a,b). Type of samples
Slaughterhouse wastewater (GC/l) Ter river water (GC/l)
No. of samples
10 6
Levels of BPyV and PAdV in the samples analyzed BPyV
PAdV
% Of positives samples
Mean values
Standard deviation
% Of positive samples
Mean values
Standard deviation
90.9
2.84 103
4.87 103
100
1.56 106
1.18 106
50
3.06 102
2.03 102
100
8.38
6.11
4333
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
5.1. The pros of molecular techniques applied to the detection of pathogens in water - For many pathogens and proposed new indicators, molecular techniques are the only method of detection and quantification because there is no efficient culture system. - Molecular techniques are used as identification tools for specific strains, genotyping and confirmation of the results. PCR and sequence analysis provide further information on the phylogenetic characteristics of the strains identified. - The qPCR methodology facilitates the evaluation of the efficiency of removal of pathogens or selected indicators in water treatment plants, including viruses. It is also a very useful tool for the identification of the sources of faecal contamination in water. - PCR is a powerful tool in risk assessment. The qPCR and qRT-PCR techniques provide quantitative estimations of the concentration of pathogens in water. qPCR and nucleic acid extraction protocols are easy to standardize and automate, and they can be used in the detection of pathogens that may be present below the limit of detection of other assays. The high numbers of samples that can be processed in this manner allow, in combination with epidemiological surveys, to carry out risk assessment studies that would otherwise be difficult to accomplish.
5.2. The cons of molecular techniques applied to the detection of pathogens in water - The detection of genomes by PCR-based techniques does not provide information about the infectivity of the pathogen or the indicator detected or the level of risk for the population. Disinfection of water by UV and chlorine treatments reduces the numbers of viral particles quantified by qPCR and qRT-PCR if severe treatments are applied. However, commonly applied disinfection treatments produce a significant reduction in the number of infectious viral particles without showing equivalent variations in the level of viral genomes quantified by qPCR and qRT-PCR. - The protocols used need to be improved and standardized. In the case of protocols for the concentration of viruses from water samples, differences of more than 2 logarithms are observed when using distinct concentration methods and qPCR (Albinana-Gimenez et al., 2009a). Fig. 1 shows the intra-laboratory variability for testing different methods of concentration of viruses in artificial seawater and freshwater. The methods tested are based on membrane concentration methods with different eluents, electropositive filters, glass wool columns and direct flocculation (Sobsey and Glass, 1980; Katzenelson et al., 1996; Pallin et al., 1997; Vilagine`s et al., 1997; Calgua et al., 2008). Moreover, the specific PCR conditions, primers and probes used may produce significant differences in the results (Bofill-Mas et al., 2006). - The presence of inhibitors for the molecular assays in the samples still represents a limitation in the analysis of
environmental samples. Nucleic acid extraction efficiencies vary considerably between different methods and the final nucleic acid yield depends on the methods used and the type of environmental sample. This makes direct comparison of absolute gene numbers between studies extremely problematic. Furthermore, the concentration at which inhibitors no longer affect the qPCR for any sample is not known a priori and must be determined empirically to ensure that the environmental template and the target gene for the standard curve have equivalent amplification efficiencies (Smith and Osborn, 2009). The efficiency of the reverse transcription may also be variable and in general qRT-PCR is considered to be more sensitive to inhibitors than qPCR.
6. Viability assays and molecular detection methods Water treatment procedures aim to kill pathogenic bacteria and protozoa, and it is clear that their physiological state in water is a major issue for water safety (Brettar and Ho¨fle, 2008). To assess whether viable cells, and not only DNA, are detected in the samples, molecular methods must be adapted. Various approaches have been evaluated, such as DNAintercalating dyes like ethidium monoazide (EMA) and propidium monoazide (PMA) to selectively remove cells with compromised membranes from the analysis. These dyes enter a cell and bind covalently to DNA when photo-activated. PCR amplification of such modified DNA is strongly inhibited. PMA has been used to discern whether a cell is alive or dead, in combination with qPCR (Nocker and Camper, 2009). A limitation of this technique, however, is that the principle is based on membrane integrity as the viability criterion. Another
10.00
Log10 virus recovery
assays (Signoretto and Canepari, 2008). The main pros and cons of molecular techniques are as follows.
1.00
0.10
Seawater
0.01 Method 1
Method 2
Fresh water Method 3
Method 4
Fig. 1 e Intra-laboratory variability of viral concentration methods in artificial seawater and freshwater. Virus recovery values obtained after spiking sets of ten 10-l samples with HAdV 2, concentrating by: Method 1: electronegative filters of nitrocellulose and glycine 0.05 M pH 9.5 e skimmed milk buffer. Method 2: electronegative filters of nitrocellulose and glycine 0.25 M pH 9.5 e beef extract buffer. Method 3: a column of glass wool and glycine 0.25 M pH 9.5-beef extract buffer. Method 4: Direct organic flocculation with skimmed milk and quantifying the recovery by qPCR according to Bofill-Mas et al. (2006).
4334
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
Table 5 e Quantification of Human adenovirus (HAdV 2 and 41) and JC polyomavirus (JCPyV) viral suspensions evaluated by immunofluorescence assay (IFA) and quantitative PCR (qPCR) with previous DNAse treatment. Results expressed as genome copies (GC) or focus forming units (FFU) per ml of suspension. qPCR assays were performed according to Bofill-Mas et al. (2006). Virus
Viral suspension
DNAse treatment
1
þ þ þ þ þ þ
HAdV 2
2 HAdV 41
3 4
JCPyV
5 6
qPCR
IFA
GC/ml
FFU/ml
2.70 105 NTa 1.25 105 4.1E 105 5.00 105 NT 5.10 105 7.17 106 5.00 105 NT 2.35 105 9.64 106
1.60 104 1.15 104 1.40 104 1.70 104 1.40 104 1.31 104
a NT: not tested.
possible solution is to target mRNA, as it tends to degrade relatively rapidly after cell death. However, there is increasing evidence that in many cases, mRNA persistence depends on the genes targeted and the conditions in which the cells were inactivated, and some RNA molecules can also persist in cells after loss of viability. Although weaker than that of DNA, this persistence of mRNA can, for example, lead to false positive results in the first hours after cell death when monitoring disinfection efficacy. Moreover, some mRNA molecules are not transcribed in the viable non-culturable (VBNC) state. Additionally, precautions must be taken to avoid the contamination with genomic DNA when RT-PCR is performed. Thus, it is currently impossible to develop a general protocol, and mRNA detection methods can be used only for specific study cases (Cenciarini-Borde et al., 2009). A new CryptoPMA-PCR assay, which allows genotyping and viability determination, may improve the data on water-
borne exposure to Cryptosporidium and enhance the validity of human risk assessment (Brescia et al., 2009). The use of qPCR with PMA treatment may also enable quantization of viable pathogenic Cryptosporidium oocysts in environmental samples (Brescia et al., 2009). Diverse studies have been also developed in order to approach the molecular detection assays to the infectivity of the viral particles present in the analyzed samples. Nuanualsuwan and Cliver (2002) treated with RNAse and Proteinase K viral suspensions of hepatitis A, poliovirus vaccine 1 and feline calicivirus after inactivation by heat, ultraviolet light or hypochlorite as a procedure for detecting only structured particles. Data on the stability of HAdV and JCPyV using qPCR with and without a DNase treatment before nucleic acid extraction indicate that viral DNA is not stable in urban sewage for long periods (Bofill-Mas et al., 2006). The t99 (time required to observe a reduction of 99% in the initial viral concentration) has been calculated from regression curves as 132.3 days for HAdV and 127.3 days for JCPyV at 20 C. When the viral concentrate was treated with DNase previously to the nucleic acid extraction, the t99 observed was 126.1 days for HAdV and 121.4 days for JCPyV. As expected, these viruses were more stable in phosphate buffered saline (PBS) than in sewage. Tables 5 and 6 report the quantification of viruses in viral suspensions, purified from cell-culture supernatants, and also of viruses in seawater, using cell-culture techniques, qPCR and qRT-PCR with and without DNAse or RNAse treatments.
7.
Future developments
The application of new technologies such as high-throughput mass sequencing to analyze stool samples collected from patients with acute diarrhoea, and the use of nested-PCR (nPCR) and nucleotide sequence analyses in the study of faecal and environmental samples has greatly increased the number of viruses that can be identified in the environment. For example, a new picornavirus has recently been identified in gastroenteritis patients, and sewage, Klassevirus I (Holtz et al., 2008, 2009; Blinkova et al., 2009), and it is expected to produce interesting new information in the future.
Table 6 e Quantification of Human Adenovirus 2 (HAdV 2) and Murine Norovirus 1 (MNV1) in natural and artificial seawater spiked with both viruses. Virus concentration was evaluated at 0 and 60 min by plaque Assay, qPCR and qRT-PCR with previous enzymatic treatment (RNase for MNV1 and DNase for HAdV2 respectively). Results expressed as genome copies (GC) or plaque forming units (PFU) per ml. qPCR assays were performed according to Baert et al. (2008) and Bofill-Mas et al. (2006). Mean values of two replicate experiments are showed. Virus HAdV 2
Seawater Natural Artificial
MNV1
Natural Artificial
DNAse
GC/ml (t0/t60) 5
PFU/ml (t0/t60) 5
þ þ
2.3 10 /1.00 10 1.71 107/1.65 107 4.02 105/3.47 104 1.02 107/1.90 106
2.50 104/1.80 104
RNAse
GC/ml (t0/t60)
PFU/ml (t0/t60)
þ þ
1.40 108/2.00 107 4.99 108/7.65 108 7.00 107/1,03 108 9.21 108/7.54 108
1.95 105/1.40 105
1.65 104/1.40 104
3.95 105/2.90 105
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
Several authors have suggested developing integrated systems for detecting multiple pathogens and indicators in source, drinking and recreational water. DNA microarray technologies could be the basis for such a test, although initial results have shown that direct hybridization of genomic DNA or RNA may not have the desired sensitivity. If microarray technologies could be coupled with PCR amplification of the target genes the signal sensitivity could be increased by 106fold (Gilbride et al., 2006). Other technical improvements are to be expected as a result of the advances in microfluidics and nanobiotechnology. Miniaturized systems could be based on microchips, and several approaches have been described (Ivniski et al., 2003; Gilbride et al., 2006). More information, on the stability of genetic markers and distribution of pathogens and indicators in diverse geographical areas and the diverse matrices is needed for the identification of the most suitable molecular targets for detection and quantification of pathogens and the evaluation of the applicability of new indicators and MST tools. More research is required on the identification of indicators that better correlate to pathogen presence as even the newly emerging indicators in MST development often have poor success in predicting pathogen presence. However, the epidemiological pattern of many pathogens is different which makes it necessary to distinguish between the significance of analyzing widely spread, highly prevalent indicators of faecal/urine contamination in water as an indication of potential contamination by many of the pathogens and the detection of specific pathogens that may be sporadically highly abundant in water and often do not correlate to other indicators. Several assays based on molecular techniques for detection and quantification of pathogens and potential indicators have been developed that may be validated and standardized, and the technology could be ready for routine implementation and automation in the near future.
8.
Conclusions
Molecular techniques, specifically nucleic acid amplification procedures, provide sensitive, rapid and quantitative analytical tools for studying specific pathogens, including new emergent strains and indicators. They can be used to evaluate the microbiological quality of water, the efficiency of pathogen removal in drinking and wastewater treatment plants, and in MST studies. Water disinfection treatments may have an effect on the viability of pathogens and the application of molecular techniques produces numbers of genome copies that may overestimate the concentration of infectious microorganisms. The molecular techniques available today and those under development would require further refinement in order to be standardized and applicable to a diversity of matrices.
Acknowledgments This work was supported partially by the “Ministerio de Educacio´n y Ciencia” of the Spanish Government (projects AGL2005-07776-C03-02, AGL2008-05275-C03-01/ALI and
4335
AGL2008-05275-C03-02/ALI). Jesus Rodriguez-Manzano (BES2006-13638) and Anna Carratala (BES-2009-014666) are fellows of the Spanish Ministry of Science. We thank Juan Lopez-Pila for critical review of this article.
references
Adam, R.D., 2001. Biology of Giardia lamblia. Clin. Microbiol. 14, 447e475. Ahmed, W., Sawant, S., Huygens, F., Goonetilleke, A., Gardner, T., 2009. Prevalence and occurrence of zoonotic bacterial pathogens in surface waters determined by quantitative PCR. Water Res. 43 (19), 4918e4928. Alagappan, A., Bergquist, P.L., Ferrari, B.C., 2009. Development of a two-color fluorescence in situ hybridization technique for species-level identification of human-infections Cryptosporidium spp. Appl. Environ. Microbiol. 75, 5996e5998. Albinana-Gimenez, N., Clemente-Casares, P., Calgua, B., Courtois, S., Huguet, J.M., Girones, R., 2009a. Comparison of methods for the quantification of human adenoviruses, Polyomavirus JC and Norovirus in source and drinking water. J. Virol. Methods 158, 104e109. Albinana-Gimenez, N., Miagostovich, M., Calgua, B., Huguet, J.M., Matia, L., Girones, R., 2009b. Analysis of adenoviruses and polyomaviruses quantified by qPCR as indicators of water quality in source and drinking water-treatment plants. Water Res. 43, 2011e2019. Anceno, A.J., Ozaki, M., Dang, Y.N., Chuluun, B., Shipin, O.V., 2007. Canal networks as extended waste stabilization ponds: fate of pathogens in constructed waterways in Pathumthani Province. Thailand Water Sci. Technol. 55 (11), 143e156. Baeumner, A.J., Humiston, M.C., Montagna, R.A., Durst, R.A., 2001. Detection of viable oocysts of Cryptosporidium parvum following nucleic acid sequence based amplification. Anal. Chem. 73, 1176e1180. Baert, L., Wobus, C.E., Van Coillie, E., Thackray, L.B., Debevere, J., Uyttendaele, M., 2008. Detection of murine norovirus 1 by using plaque assay, transfection assay, and real-time reverse transcription-PCR before and after heat exposure. Appl. Environ. Microbiol. 74 (2), 543e546. Barbuddhe, S.B., Chakraborty, T., 2009. Listeria as an enteroinvasive gastrointestinal pathogen. Curr. Top. Microbiol. Immunol. 337, 173e195. Bavaro, M.F., 2009. Escherichia coli O157: what every internist and gastroenterologist should know. Curr. Gastroenterol. Rep. 11 (4), 301e306. Bernhard, A.E., Field, K.G., 2000a. A PCR assay to discriminate human and ruminant feces on the basis of host differences in BacteroidesePrevotella genes encoding 16S rRNA. Appl. Environ. Microbiol. 66, 4571e4574. Bernhard, A.E., Field, K.G., 2000b. Identification of nonpoint sources of fecal pollution in coastal waters by using hostspecific 16S ribosomal DNA genetic markers from fecal anaerobes. Appl. Environ. Microbiol. 66, 1587e1594. Bertrand, I., Gantzer, C., Chesnot, T., Schwartzbrod, J., 2004. Improved specificity for Giardia lamblia cyst quantification in wastewater by development of a real-time PCR method. J. Microbiol. Methods 57, 41e53. Bertrand, I., Schwartzbrod, J., 2007. Detection and genotyping of Giardia duodenalis in wastewater: relation between assemblages and faecal contamination origin. Water Res. 41 (16), 3675e3682. Blinkova, O., Rosario, K., Li, L., Kapoor, A., Slikas, B., Bernardin, F., Breitbart, M., Delwart, E., 2009. Frequent detection of highly diverse variants of cardiovirus, cosavirus, bocavirus, and circovirus in sewage samples collected in the United States. J. Clin. Microbiol. 47 (11), 3507e3513.
4336
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
Bofill-Mas, S., Albinana-Gimenez, N., Clemente-Casares, P., Hundesa, A., Rodiguez-Manzano, J., Allard, A., Calvo, M., Girones, R., 2006. Quantification and stability of human adenoviruses and polyomavirus JCPyV in wastewater matrices. Appl. Environ. Microbiol. 72 (12), 7894e7896. Bofill-Mas, S., Pina, S., Girones, R., 2000. Documenting the epidemiologic patterns of polyomaviruses in human populations studying their presence in urban sewage. Appl. Environ. Microbiol. 66, 238e245. Bottari, B., Ercolini, D., Gatti, M., Neviani, E., 2006. Application of FISH technology for microbiological analysis: current state and prospects. Appl. Microbiol. Biotechnol. 73, 485e494. Brescia, C.C., Griffin, S.M., Ware, M.W., Varughese, E.A., Egorov, A.I. , Villegas, E.N., 2009. Cryptosporidium propidium monoazidePCR, a molecular biology-based technique for genotyping of viable Cryptosporidium oocysts. Appl. Environ. Microbiol. 75, 6856e6863. Brettar, I., Ho¨fle, M.G., 2008. Molecular assessment of bacterial pathogens da contribution to drinking water safety. Curr. Opin. Biotechnol. 19, 274e280. Bustin, S.A., 2010. Why the need for qPCR publication guidelines?eThe case for MIQE. Methods 50 (4), 217e226. Caccio`, S.M., 2005. Molecular epidemiology of human cryptosporidiosis. Parassitologia 47, 185e192. Caccio`, S.M., 2003. Molecular techniques to detect and identify protozoan parasites in the environment. Acta Microbiol. Pol. 52, 23e34. Calgua, B., Mengewein, A., Gru¨nert, A., Bofill-Mas, S., ClementeCasares, P., Hundesa, A., Wyn-Jones, A.P., Lo´pez-Pila, J.M., Girones, R., 2008. Development and application of a one-step low cost procedure to concentrate viruses from seawater samples. J. Virol. Methods 153, 79e83. Carey, C.M., Lee, H., Trevors, J.T., 2004. Biology, persistence, and detection of Cryptosporidium parvum and Cryptosporidium hominis oocyst. Water Res. 38, 818e862. Cenciarini-Borde, C., Courtois, S., La Scola, B., 2009. Nucleic acids as viability markers for bacteria detection using molecular tools. Future Microbiol. 4, 45e64. Chauret, C., Armstrong, N., Fisher, J., Sharma, R., Springthorpe, S., Sattar, S., 1995. Correlating Cryptosporidium and Giardia with microbial indicators. J. Am. Water Works Assoc. 87, 76e84. Cheng, C.M., Van Khanh, T., Lin, W., Ruby, R.M., 2009. Interlaboratory validation of a real-time PCR 24-hour rapid method for detection of Salmonella in foods. J. Food Prot. 72 (5), 945e951. Cizek, A.R., Characklis, G.W., Krometis, L.H., Hayes, J.A., Simmons, O.D., DiLonardo, S., Alderisiio, K.A., Sobsey, M.D., 2008. Comparing the partitioning behaviour of Giardia and Cryptosporidium with that of indicator organisms in stormwater runoff. Water Res. 42, 4421e4438. Clemente-Casares, P., Pina, S., Buti, M., Jardi, R., Martı´n, M., BofillMas, S., Girones, R., 2003. Hepatitis E virus epidemiology in industrialized countries. Emerg. Infect. Dis. 9, 448e454. Conraths, F.J., Schares, G., 2006. Validation of moleculardiagnostic techniques in the parasitological laboratory. Vet. Parasitol. 136 (2), 91e98. Cook, N., 2003. The use of NASBA for the detection of microbial pathogens in food and environmental samples. J. Microbiol. Methods 53 (2), 165e174. Donaldson, K.A., Griffin, D.W., Paul, J.H., 2002. Detection, quantitation and identification of enteroviruses from surface waters and sponge tissue from the Florida Keys using realtime RT-PCR. Water Res. 36, 2505e2514. Dong, J., Olano, J.P., McBride, J.W., Walker, D.H., 2008. Emerging pathogens: challenges and successes of molecular diagnostics. J. Mol. Diagn. 10 (3), 185e197. Dorai-Raj, S., O’Grady, J., Colleran, E., 2009. Specificity and sensitivity evaluation of novel and existing Bacteroidales and Bifidobacteria-specific PCR assays on feces and sewage
samples and their application for microbial source tracking in Ireland. Water Res. 43, 4980e4988. Doring, G., Unertl, K., Heininger, A., 2008. Validation criteria for nucleic acid amplification techniques for bacterial infections. Clin. Chem. Lab. Med. 46, 909e918. Fayer, R., 2004. Cryptosporidium: a water-borne zoonotic parasite. Vet. Parasitol. 126, 37e56. Fayer, R., Morgan, U., Upton, S.J., 2000. Epidemiology of Cryptosporidium: transmission, detection and identification. Int. J. Parasitol. 30, 1305e1322. Field, K.G., 2004. Faecal source identification. In: Cotruvo, J.A., Dufour, A., Reese, G., Bartram, J., Carr, R., Cliver, D.O., Craun, G.F., Fayer, R., Gannon, V.P.J. (Eds.), Waterborne Zoonosis: Identification, Causes and Control. IWA Publishing, London, England, pp. 349e366. Field, K.G., Bernhard, A.E., Brodeur, T.J., 2003. Molecular approaches to microbiological monitoring: fecal source detection. Environ. Monit. Assess. 81, 313e326. Fong, T.T., Lipp, E.K., 2005. Enteric viruses of humans and animals in aquatic environments: health risks, detection, and potential water quality assessment tools. Microbiol. Mol. Biol. Rev. 69, 357e371. Fong, T.T., Phanikumar, M.S., Xagoraraki, I., Rose, J.B., 2010. Quantitative detection of human adenoviruses in wastewater and combined sewer overflows influencing a michigan river. Appl. Environ. Microbiol. 76 (3), 715e723. Fontaine, M., Guillot, E., 2003. An immunomagnetic separationreal-time PCR method for quantification of Cryptosporidium parvum in water samples. J. Microbiol. Methods 54 (1), 29e36. Formiga-Cruz, M., Allard, A.K., Conden-Hansson, A.C., Hensilwood, H., Hernroth, B.E., Jofre, J., Lees, D.N., Lucena, F., Papapetropoulou, R.E., Rangdale, A., Tsibouxi, A., Vantarakis, A., Girones, R., 2003. Evaluation of potential indicators of viral contamination in shellfish with applicability to diverse geographical areas. Appl. Environ. Microbiol. 69, 1556e1563. Fumian, T.M., Guimares, F.R., Vaz, B.J.P., da Silva, M.T.T., Muylaert, F.F., Bofill-Mas, S., Girones, R., Leite, J.P.G., Miagostovich MP. Molecular detection, quantification and characterization of human polyomavirus JC from wastewater in rio de Janiero, Brazil. J. Water Health, in press. van der Giessen, J.W., de Vries, A., Roos, M., Wielinga, P., Kortbeek, L.M., Mank, T.G., 2006. Genotyping of Giardia in Dutch patients and animals: a phylogenetic analysis oh human and animal isolates. Int. J. Parasitol. 36, 849e858. Gilbride, K.A., Lee, D.Y., Beaudette, L.A., 2006. Molecular techniques in wastewater: Understanding microbial communities, detecting pathogens, and real-time process control. J. Microbiol. Methods 66 (1), 1e20. Gonza´lez, A., Botella, S., Montes, R.M., Moreno, Y., Ferru´s, M.A., 2007. Direct detection and identification of Arcobacter species by multiplex PCR in chicken and wastewater samples from Spain. J. Food Prot. 70, 341e347. Goodwin, K.D., Litaker, R.W., 2008. Emerging technologies for monitoring recreational waters for bacteria and viruses. In: Walsh, P.J., Smith, S.L., Fleming, L.E., Solo-Gabriele, H.M., Gerwick, W.H. (Eds.), Oceans and Human Health. Elsevier, Academic Press, Amsterdam, Netherlands, pp. 381e404. Guilbaud, M., de Coppet, P., Bourion, F., Rachman, C., Pre´vost, H., Dousset, X., 2005. Quantitative detection of Listeria monocytogenes in biofilms by real-time PCR. Appl. Environ. Microbiol. 71 (4), 2190e2194. Guy, R.A., Payment, P., Krull, U.J., Horgen, P.A., 2003. Real-time PCR for quantification of Giardia and Cryptosporidium in environmental water samples and sewage. Appl. Environ. Microbiol. 69 (9), 5178e5185. Haramoto, E., Katayama, H., Oguma, K., Ohgaki, S., 2005. Application of cation-coated filter method to detection of
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
noroviruses, enteroviruses adenoviruses and torque teno viruses in the Tamagawa River in Japan. Appl. Environ. Microbiol. 71 (5), 2403e2411. Haramoto, E., Kitajima, M., Katayama, H., Ohgaki, S. Real-time PCR detection of adenoviruses, polyomaviruses, and torque teno viruses in river water in Japan. Water Res. in press. Hart, C.A., Cunliffe, N.A., Nakagomi, O., 2009. Diarrhoea caused by viruses. In: Cook, G.C., Zumla, A.I. (Eds.), Manson’s Tropical Diseases. Saunders Elsevier, Philadelphia, USA, pp. 815e824. Harwood, V.J., Brownell, M., Wang, S., Lepo, J., Ellender, R.D., Ajidahun, A., Hellein, K.N., Kennedy, E., Ye, X., Flood, C., 2009. Validation and field testing of library-independent microbial source tracking methods in the Gulf of Mexico. Water Res. 43 (19), 4812e4819. Harwood, V.J., Gandhi, J.P., Wright, A.C., 2004. Methods for isolation and confirmation of Vibrio vulnificus from oysters and environmental sources: a review. J. Microbiol. Methods 59, 301e316. He, J., Jiang, S., 2005. Quantification of enterococci and human adenoviruses in environmental samples by real-time PCR. Appl. Environ. Microbiol. 71 (5), 2250e2255. Ho, H.T., Lipman, L.J., Gaastra, W., 2006. Arcobacter, what is known and unknown about a potential foodborne zoonotic agent! Vet. Microbiol. 115 (1e3), 1e13. Holtz, L.R., Finkbeiner, S.R., Kirkwood, C.D., Wang, D., 2008. Identification of a novel picornavirus related to cosaviruses in a child with acute diarrhea. Virol. J. 5, 159. Holtz, L.R., Finkbeiner, S.R., Zhao, G., Kirkwood, C.D., Girones, R., Pipas, J.M., Wang, D., 2009. Klassevirus 1, a previously undescribed member of the family Picornaviridae, is globally widespread. Virol. J. 6, 86. Horman, A., Rimhanen-Finne, R., Maunula, L., von Bonsdorff, C.H. , Torvela, N., Heikinheimo, A., Hanninen, M.L., 2004. Campylobacter spp., Giardia spp., Cryptosporidium spp., Noroviruses, and indicator organisms in surface water in southwestern Finland, 2000e2001. Appl. Environ. Microbiol. 70, 87e95. Huang, J., Zhu, Y., Wen, H., Zhang, J., Huang, S., Niu, J., Li, Q., 2009. Quadruplex real-time PCR assay for detection and identification of Vibrio cholerae O1 and O139 strains and determination of their toxigenic potential. Appl. Environ. Microbiol. 75 (22), 6981e6985. Hundesa, A., Bofill-Mas, S., Maluquer de Motes, C., RodriguezManzano, J., Bach, A., Casas, M., Girones, R., 2009a. Development of a quantitative PCR assay for the quantitation of bovine polyomavirus as a microbial source-tracking tool. J. Virol. Methods 163 (2), 385e389. Hundesa, A., Maluquer de Motes, C., Albinana-Gimenez, N., Rodriguez-Manzano, J., Bofill-Mas, S., Sun˜en, E., Girones, R., 2009b. Development of a qPCR assay for the quantification of porcine adenoviruses as an MST tool for swine fecal contamination in the environment. J. Virol. Methods 158, 130e135. Hundesa, A., Maluquer de Motes, C., Bofill-Mas, S., AlbinanaGimenez, N., Girones, R., 2006. Identification of human and animal adenoviruses and polyomaviruses for determination of sources of fecal contamination in the environment. Appl. Environ. Microbiol. 72, 7886e7893. Hurst, C.J., Crawford, R.L., Knudsen, G.R., McInerney, M.J., Stetzenbach, L.D., 2002. Manual of Environmental Microbiology. ASM Press, Washington, DC. Igbinosa, E.O., Obi, L.C., Okoh, A.I., 2009. Occurrence of potentially pathogenic vibrios infinal effluents of a wastewater treatment facility in a rural community of the Eastern Cape Province of South Africa. Res. Microbiol. 160 (8), 531e537. Ivniski, D., O’Neil, D.J., Gattuso, A., Schlicht, R., Calidonna, M., Fisher, R., 2003. Nucleic acid approaches for detection and identification of biological warfare and infectious disease agents. Biotechniques 35, 862e869.
4337
Jakubowski, W., Boutros, S., Faber, W., Fayer, R., Ghiorse, W., LeChevallier, M., Rose, J., Schaub, S., Singh, A., Stewart, M., 1996. Environmental methods for Cryptosporidium. J. Am. Water Works Assoc. 88, 107e121. Jiang, S., Noble, R., Chu, W., 2001. Human adenoviruses and coliphages in urban runoff-impacted coastal waters of Southern California. Appl. Environ. Microbiol. 67 (1), 179e184. Jimenez-Clavero, M.A., Escribano-Romero, E., Mansilla, C., Go´mez, N., Co´rdoba, L., Roblas, N., Ponz, F., Ley, V., Sa´iz, J.C., 2005. Survey of bovine enterovirus in biological and environmental samples by a highly sensitive real-time reverse transcription-PCR. Appl Environ. Microbiol. 71 (7), 3536e3543. Jimenez-Clavero, M.A., Fernadez, C., Ortiz, J.A., Pro, J., Carbonell, G., Tarazona, J.V., Roblas, N., Ley, V., 2003. Teschoviruses as indicators of porcine fecal contamination of surface water. Appl. Environ. Microbiol. 69, 6311e6315. Jothikumar, N., Wang, X., Griffiths, M.W., 2003. Real-time multiplex SYBR green I-based PCR assay for simultaneous detection of Salmonella serovars and Listeria monocytogenes. J. Food Prot. 66 (11), 2141e2145. Juhna, T., Birzniece, D., Larsson, S., Zulenkovs, D., Shapiro, A., Azevedo, N.F., Me´nard-Szczebara, F., Castagnet, S., Fe´liers, C., Keevil, C.W., 2007. Detection of Escherichia coli in biofilms from pipe samples and coupons in drinking water distribution networks. Appl. Environ. Microbiol. 73, 7456e7464. Jyoti, A., Ram, S., Vajpayee, P., Singh, G., Dwivedi, P.D., Jain, S.K., Shanker, R., 2009. Contamination of surface and potable water in South Asia by Salmonellae: Culture-independent quantification with molecular beacon real-time PCR. Sci. Total Environ. 408 (6), 1256e1263. Katzenelson, E., Fattal, B., Hostovesky, T., 1996. Organic flocculation: an efficient second-step concentration method for the detection of viruses in tap water. Appl. Environ. Microbiol. 32 (4), 638e639. Keeley, A., Faulkner, B.R., 2008. Influence of land use and watershed characteristics on protozoa contamination in a potential drinking water resources reservoir. Water Res. 42, 2803e2813. Lambertini, E., Spencer, S.K., Bertz, P.D., Loge, F.J., Kieke, B.A., Borchardt, M.A., 2008. Concentration of enteroviruses, adenoviruses and noroviruses from drinking water with glass wool filters. Appl. Environ. Microbiol. 74 (10), 2990e2996. Lavender, J.S., Kinzelman, J.L., 2009. A cross comparison of QPCR to agar-based or defined substrate test methods for the determination of Escherichia coli and enterococci in municipal water quality monitoring programs. Water Res. 43 (19), 4967e4979. Laverick, M.A., Wyn-Jones, A.P., Carter, M.J., 2004. Quantitative RTPCR for the enumeration of noroviruses (Norwalk-like viruses) in water and sewage. Lett. Appl. Microbiol. 39 (2), 127e136. Le Cann, P., Ranarijaona, S., Monpoeho, S., Le Guyader, F., Ferre´, V., 2004. Quantification of human astroviruses in sewage using real-time RT-PCR. Res. Microbiol. 155 (1), 11e15. Lipp, E.K., Farrah, S.A., Rose, J.B., 2001. Assesment of microbiological fecal pollution and human enteric pathogens in a coastal community. Mar. Pollut. Bull. 42 (4), 286e293. Liu, Y.J., Zhang, C.M., Wang, X.C., 2009. Simultaneous detection of enteric bacteria from surface waters by QPCR in comparison with conventional bacterial indicators. Environ. Monit. Assess. 158, 535e544. Marcelino, L.A., Backman, V., Donaldson, A., Steadman, C., Thompson, J.R., Preheim, S.P., Lien, C., Lim, E., Veneziano, D., 2006. Accurately quantifying low-abundant targets amid similar sequences by revealing hidden correlations in oligonucleotide microarray data. Polz MF. Proc. Natl. Acad. Sci. U. S. A. 103 (37), 13629e13634. Masago, Y., Oguma, K., Katayama, H., Ohgaki, S., 2006. Quantification and genotyping of Cryptosporidium spp. in river
4338
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
water by quenching probe PCR and denaturing gradient gel electrophoresis. Water Sci. Technol. 54 (3), 119e126. Maynard, C., Berthiaume, F., Lemarchand, K., Harel, J., Payment, P., Bayardelle, P., Masson, L., Brousseau, R., 2005. Waterborne pathogen detection by use of oligonucleotide-based microarrays. Appl. Environ. Microbiol. 71 (12), 8548e8557. McDade, J.E., Shepard, C.C., Fraser, D.W., Tsai, T.R., Redus, M.A., Dowdle, W.R., 1977. Legionnaires’ disease: isolation of a bacterium and demonstration of its role in other respiratory disease. N. Engl. J. Med. 297 (22), 1197e1203. McFeters, G.A., Bissonnette, G.K., Jezeski, J.J., Thomson, C.A., Stuart, D.G., 1974. Comparative survival of indicator bacteria and enteric pathogens in well water. Appl. Environ. Microbiol. 27, 823e829. Miagostovich, M.P., Ferreira, F.F., Guimara˜es, F.R., Fumian, T.M., Diniz-Mendes, L., Luz, S.L., Silva, L.A., Leite, J.P., 2008. Molecular detection and characterization of gastroenteritis viruses occurring naturally in the stream waters of Manaus, central Amazonia, Brazil. Appl. Environ. Microbiol. 74 (2), 375e382. Millar, B.C., Finn, M., Xiao, L., Lowery, J., Dooley, S.G., Moore, J.E., 2002. Cryptosporidium in foodstuffs-an emerging aetiological route of human foodborne illness. Trends Food Sci. Technol. 13, 168e187. Morio, F., Corvec, S., Caroff, N., Le Gallou, F., Drugeon, H., Reynaud, A., 2008. Real-time PCR assay for the detection and quantification of Legionella pneumophila in environmental water samples: utility for daily practice. Int. J. Hyg. Environ. Health 211 (3e4), 403e411. Nayak, A.K., Rose, J.B., 2007. Detection of Helicobacter pylori in sewage and water using a new quantitative PCR method with SYBR green. J. Appl. Microbiol. 103 (5), 1931e1941. Noble, R.T., Weisberg, S.B., 2005. A review of technologies for rapid detection of bacteria in recreational waters. J. Water Health 3 (4), 381e392. Nocker, A., Camper, A.K., 2009. Novel approaches toward preferential detection of viable cells using nucleic acid amplifcation techniques. FEMS Microbiol. Lett. 291, 137e142. Nuanualsuwan, S., Cliver, D.O., 2002. Pretreatment to avoid positive RT-PCR results with inactivated viruses. J. Virol. Methods 104, 217e225. Pallin, R., Wyn-Jones, A.P., Place, B.M., Lightfoot, N.F., 1997. The detection of enteroviruses in large volume concentrates of recreational waters by the polymerase chain reaction. J. Virol. Methods 67 (1), 57e67. Park, S.R., MacKay, W.G., Reid, D.C., 2001. Helicobacter sp. recovered from drinking water biofilm sampled from a water distribution system. Water Res. 35, 1624e1626. Percival, S.L., Thomas, J.G., 2009. Transmission of Helicobacter pylori and the role of water and biofilms. J. Water Health 7 (3), 469e477. Pina, S., Buti, M., Cotrina, M., Piella, J., Girones, R., 2000. HEV identified in serum from humans with acute hepatitis and in sewage of animal origin in Spain. J. Hepatol. 33 (5), 826e833. Pina, S., Puig, M., Lucena, F., Jofre, J., Girones, R., 1998. Viral pollution in the environment and shellfish: human adenovirus detection by PCR as an index of human viruses. Appl. Environ. Microbiol. 64, 3376e3382. Pote, J., Haller, L., Kottelat, R., Sastre, V., Arpagaus, P., Wildi, W., 2009. Persistence and growth of faecal culturable bacterial indicators in water column and sediments of Vidy Bay, Lake Geneva, Switzerland. J. Environ. Sci. 21, 62e69. Puig, M., Jofre, J., Lucena, F., Allard, A., Wadell, G., Girones, R., 1994. Detection of adenoviruses and enteroviruses in polluted waters by nested PCR amplification. Appl. Environ. Microbiol. 60, 2963e2970. Queralt, N., Bartolome´, R., Araujo, R., 2005. Detection of Helicobacter pylori DNA in human faeces and water with
different levels of faecal pollution in the north-east of Spain. J. Appl. Microbiol. 98, 889e895. Rajal, V.B., McSwain, B.S., Thompson, D.E., Leutenegger, C.M., Kildare, B.J., Wuertz, S., 2007. Validation of hollow fiber ultrafiltration and real-time PCR using bacteriophage PP7 as surrogate for the quantification of viruses from water samples. Water Res. 41 (7), 1411e1422. Raymaekers, M., Smets, R., Maes, B., Cartuyvels, R.J., 2009. Checklist for optimization and validation of real-time PCR assays. Clin. Lab. Anal. 23 (3), 145e151. Reynolds, K.A., Mena, K.D., Gerba, C.P., 2008. Risk of waterborne illness via drinking water in the United States. Rev. Environ. Contam. Toxicol. 192, 117e158. Rodriguez-Manzano, J., Miagostovich, M., Hundesa, A., ClementeCasares, P., Carratala, A., Buti, A., Jardı´, R., Girones, R., 2010. Analysis of the evolution in the circulation of HAV and HEV in Eastern Spain by testing urban sewage samples. J. Water Health 8 (2), 346e354. Sahlstro¨m, L., Aspana, A., Baggea, E., Danielsson- Thamb, M.L., Albihna, A., 2004. Bacterial pathogen incidences in sludge from Swedish sewage treatment plants. Water Res. 38, 1989e1994. Savichtcheva, O., Okabe, S., 2006. Alternative indicators of fecal pollution: relations with pathogens and conventional indicators, current methodologies for direct pathogen monitoring and future application perspectives. Water Res. 40, 2463e2476. Schvoerer, E., Ventura, M., Dubos, O., Cazaux, G., Serceau, R., Gournier, N., Dubois, V., Caminade, P., Fleury, H.J., Lafon, M.E., 2001. Qualitative and quantitative molecular detection of enteroviruses in water from bathing areas and from a sewage treatment plant. Res. Microbiol. 152 (2), 179e186. Scott, T.M., Rose, J.B., Jenkins, T.M., Farrah, S.R., Lukasik, J., 2002. Microbial source tracking: current methodology and future directions. Appl. Environ. Microbiol. 68 (12), 5796e5803. Sen, K., Schable, N.A., Lye, D.J., 2007. Development of an internal control for evaluation and standardization of a quantitative PCR assay for detection of Helicobacter pylori in drinking water. Appl. Environ. Microbiol. 73, 7380e7387. Shannon, K.E., Lee, D.Y., Trevors, J.T., Beaudette, L.A., 2007. Application of real-time quantitative PCR for the detection of selected bacterial pathogens during municipal wastewater treatment. Sci. Total Environ. 382 (1), 121e129. Sharma, S., Sachdeva, P., Virdi, J.S., 2003. Emerging water-borne pathogens. Appl. Microbiol. Biotechnol. 61 (5e6), 424e428. Sidhu, J.P.S., Toze, S.G., 2009. Human pathogens and their indicators in biosolids: a literature review. Environ. Int. 35, 187e201. Signoretto, C., Canepari, P., 2008. Towards more accurate detection of pathogenic Gram-positive bacteria in waters. Curr. Opin. Biotechnol. 19, 248e253. Simpson, J.M., Santo Domingo, J.W., Reasoner, D.J., 2002. Microbial source tracking: state of the science. Environ. Sci. Technol. 36 (24), 5279e5288. Sinclair, R.G., Jones, E.L., Gerba, C.P., 2009. Viruses in recreational water-borne disease outbreaks: a review. J. Appl. Microbiol. 107, 1769e1780. Sobsey, M., Glass, J.S., 1980. Poliovirus concentration from tap water with electropositive adsorbent filters. Appl. Environ. Microbiol. 40 (2), 201e210. Solo-Gabriele, H.M., Wolfert, M.A., Desmarais, T.R., Palmer, C.J., 2000. Sources of E. coli in a coastal subtropical environment. Appl. Environ. Microbiol. 66, 230e237. Slifko, T.R., Smith, H.V., Rose, J.B., 2000. Emerging parasite zoonoses associated with water and food. Int. J. Parasitol. 30, 1379e1393. Smith, C.J., Osborn, A.M., 2009. Advantages and limitations of quantitative PCR (Q-PCR)-based approaches in microbial ecology. FEMS Microbiol. Ecol. 67, 6e20. Smith, H.V., Caccio`, S.M., Cook, N., Nichols, R.A.B., Tait, A., 2007. Cryptosporidium and Giardia as foodborne zoonoses. Vet. Parasitol. 149, 29e40.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 3 2 5 e4 3 3 9
Smith, J.J., Gunasekera, T.S., Barardi, C.R., Veal, D., Vesey, G., 2004. Determination of Cryptosporidium parvum oocyst viability by fluorescence in situ hybridization using a ribosomal RNAdirected probe. J. Appl. Microbiol. 96, 409e417. Steiner, T.S., Thielman, N.M., Guerrant, R.L., 1997. Protozoal agents: what are the dangers for the public water supply? Annu. Rev. Med. 48, 329e340. Steinert, M., Hentschel, U., Hacker, J., 2002. Legionella pneumophila: an aquatic microbe goes astray. FEMS Microbiol. Rev. 26, 149e162. Straub, T.M., Dockendorff, B.P., Quin˜onez-Dı´az, M.D., Valdez, C.O. , Shutthanandan, J.I., Tarasevich, B.J., Grate, J.W., BrucknerLea, C.J., 2005. Automated methods for multiplexed pathogen detection. J. Microbiol. Methods 62 (3), 303e316. Stewart, J.R., Gast, R.J., Fujioka, R.S., Solo-Gabriele, H.M., Meschke, J.S., Amaral-Zettler, L.A., Del Castillo, E., Polz, M.F., Collie, T.K., Strom, M.S., Sinigalliano, C.D., Moeller, P.D., Holland, A.F., 2008. The coastal environment and human health: microbial indicators, pathogens, sentinels and reservoirs. Environ. Health 7 (2), S3. Thompson, H.P., Lowery, C.J., Moore, J.E., Millar, B.C., Dooley, J.S.G., 2003. Detection of Viable Oocysts of Cryptosporidium parvum by Nucleic Acid Sequence Based Amplification (NASBA) of the DNA Replication Gene Cp-RPA1. Teagasc Irish Agriculture and Food Development Authority. http://www.teagasc.ie/ publications/2003/conferences/cryptosporidiumparvum/ poster20.htm. Thompson, R.C., Hopkins, R.M., Homan, W.L., 2004. Nomenclature and genetic groupings of Giardia infecting mammals. Parasitol. Today 16 (5), 210e213. Tree, J.A., Adams, M.R., Lees, D.N., 2003. Chlorination of indicator bacteria and viruses in primary sewage effluent. Appl. Environ. Microbiol. 69, 2038e2043. USEPA, April 2001. Method 1623: Giardia and Cryptosporidium in Water by Filtration/IMS/FA. United States Environmental Protection Agency publication, Office of Water (EPA/821/R-801/025). Victoria, M., Rigotto, C., Moresco, V., de Abreu Correˆa, A., Kolesnikovas, C., Leite, J.P., Miagostovich, M.P., Barardi, C.R.
4339
Assessment of norovirus contamination in environmental samples from Floriano´polis City, Southern Brazil. J. Appl. Microbiol, in press. Vilagine`s, P., Sarrette, B., Champsaur, H., Hugues, B., Dubrou, S. , Joret, J.C., Laveran, H., Lesne, J., Paquin, J.L., Delattre, J.M., Oger, C., Alame, J., Grateloup, I., Perrollet, H., Serceau, R., Sine`gre, F., Vilagine`s, R., 1997. Round robin investigation of glass wool method for poliovirus recovery from drinking water and sea water. Water. Sci. Technol. 35, 445e449. Wallis, P.M., Erlandsen, S.L., Isaac-Renton, J.L., Olson, M.E., Robertson, W.J., van Keulen, H., 1996. Prevalence of Giardia cysts and Cryptosporidum oocysts and characterization of Giardia spp. isolated from drinking water in Canada. Appl. Environ. Microbiol. 62, 2789e2797. We´ry, N., Lhoutellier, C., Ducray, F., Delgene`s, J.P., Godon, J.J., 2008. Behaviour of pathogenic and indicator bacteria during urban wastewater treatment and sludge composting, as revealed by quantitative PCR. Water Res. 42, 53e62. Westrell, T., Ciampa, N., Boelaert, F., Helwigh, B., Korsgaard, H., Chrı´el, M., Ammon, A., Ma¨kela¨, P., 2009. Zoonotic infections in Europe in 2007: a summary of the EFSA-ECDC annual report. Euro Surveill. 14 (3) pii: 19100. Wetz, J., Blackwood, A.D., Fries, J.S., Williams, Z.F., Noble, R.T., 2008. Trends in total Vibrio spp. And Vibrio vulnificus concentrations in the eutrophic Neuse River estuary, North Carolina, during storm events. Aquat. Microb. Ecol. 53, 141e149. Wilkes, G., Edge, T., Gannon, V., Jokinen, C., Lyautey, E., Medeiros, D., Neumann, N., Ruecker, N., Topp, E., Lapen, D.R., 2009. Seasonal relationships among indicator bacteria, pathogenic bacteria, Cryptosporidium oocysts, Giardia cysts, and hydrological indices for surface waters within an agricultural landscape. Water Res. 43 (8), 2209e2223. Xiao, L., Ryan, U.M., 2004. Cryptosporidiosis: an update in molecular epidemiology. Curr. Opin. Infect. Dis. 17, 483e490. Xiao, L., Ryan, U.M., 2008. Molecular epidemiology. In: Fayer, R., Xiao, L. (Eds.), Cryptosporidium and Cryptosporidiosis. CRC Press, Boca Raton, Florida, USA, pp. 120e171.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 8 7 e4 4 9 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Predicting the vulnerability of reservoirs to poor water quality and cyanobacterial blooms Catherine Leigh a,*, Michele A. Burford a, David T. Roberts b, James W. Udy b a b
Australian Rivers Institute, Griffith University, Nathan QLD 4111, Australia Queensland Bulk Water Supply Authority (Seqwater), Brisbane, QLD 4000, Australia
article info
abstract
Article history:
Cyanobacterial blooms in drinking water reservoirs present a major ecosystem functioning
Received 6 April 2010
and human health issue. The ability to predict reservoir vulnerability to these blooms would
Received in revised form
provide information critical for decision making, hazard prevention and management. We
2 June 2010
developed a new, comparative index of vulnerability based on simple measures of reservoir
Accepted 7 June 2010
and catchment characteristics, rather than water quality data, which were instead used to
Available online 12 June 2010
test the index’s effectiveness. Testing was based on water quality data collected over a number of seasons and years from 15 drinking water reservoirs in subtropical, southeast
Keywords:
Queensland. The index correlated significantly and strongly with algal cell densities,
Agricultural land use
including potentially toxic cyanobacteria, as well as with the proportions of cyanobacteria in
Algae
summer months. The index also performed better than each of the measures of reservoir and
Climate change
catchment characteristics alone, and as such, was able to encapsulate the physical charac-
Decision making
teristics of subtropical reservoirs, and their catchments, into an effective indicator of the
Eutrophication
vulnerability to summer blooms. This was further demonstrated by calculating the index for
Watershed
a new reservoir to be built within the study region. Under planned dimensions and land use, a comparatively high level of vulnerability was reached within a few years. However, the index score and the number of years taken to reach a similar level of vulnerability could be reduced simply by decreasing the percentage of grazing land cover via revegetation within the catchment. With climate change, continued river impoundment and the growing demand for potable water, our index has potential decision making benefits when planning future reservoirs to reduce their vulnerability to cyanobacterial blooms. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
In the tropics and subtropics, climate and the physical characteristics of reservoirs create conditions that promote algal blooms (Jones and Poplawski, 1998). Higher air temperatures and longer daylight hours in summer months lead to stronger thermal stratification and, potentially, the release of bioavailable nutrients from anoxic sediments (Burford and O’Donohue, 2006). Summer-dominated rainfall and subsequent inflows
also tend to increase nutrient supply during this time. Together with the long water residence times typically associated with reservoirs, as opposed to river systems, these factors can combine with other, complex causative factors to make summer blooms a common phenomenon (McGregor and Fabbro, 2000). Indeed, there is growing concern over the increased reporting of toxic cyanobacterial blooms in reservoirs the world over, particularly during summer months (e.g. Padisa´k, 1997; Bouvy et al., 2000; Wiedner et al., 2007). In
* Corresponding author. Tel.: þ61 07 37357457; fax: þ61 07 37357615. E-mail addresses:
[email protected] (C. Leigh),
[email protected] (M.A. Burford),
[email protected] (D.T. Roberts),
[email protected] (J.W. Udy). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.016
4488
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 8 7 e4 4 9 6
addition, global warming is predicted to increase nutrient loads and algal growth in temperate systems (Matzinger et al., 2007; Paerl and Huisman, 2008). Catchment (¼watershed) characteristics are also associated with changes in water quality and algal blooms within impounded systems. In particular, agricultural land use can lead to reservoir eutrophication: (1) by increasing soil erosion and nutrient loads in runoff, which can cause changes in nitrogen and phosphorus ratios; or (2) through a combined effect with other catchment and reservoir characteristics, such as water storage capacity and catchment size (Arbuckle and Downing, 2001; Knoll et al., 2003; Davis and Koop, 2006; Yang et al., 2008). Because summer conditions favour algal blooms and, particularly in the tropics and subtropics, tend to occur in tandem with increased precipitation (which increases the inflow of nutrients and sediments), the effects of land use on reservoir water quality may be most apparent during warmer months. In response to the increased demand for irrigation and drinking water supply by growing human populations, new reservoirs are being built and/or the storage capacities of existing reservoirs increased (L’Vovich, 1990; Pringle, 2001; Zengi et al., 2007). In developing countries, particularly China and India, new reservoirs in both urban and remote areas continue to be constructed (Dudgeon, 2000). Much of this reservoir expansion is occurring in the tropics and subtropics, and often in conjunction with conversion of forest or savannah into agricultural lands (Blanch, 2008; Gu¨cker et al., 2009). This combination of factors suggests that the incidence of algal blooms will become more likely. In turn, there is a need to reliably forecast the occurrence and frequency of toxic algal blooms in existing or future reservoirs. Regular monitoring within drinking water reservoirs is conducted to ensure that the relevant water quality guidelines for ecosystem and human health are met. This process, and the supply of safe drinking water, however, consumes considerable human and fiscal resources (De Ceballos et al., 1998). In addition, monitoring often commences after reservoirs are built and water quality problems have begun to occur. The ability to predict whether reservoirs may be more or less vulnerable to poor water quality and toxic cyanobacterial blooms, and why, is critical for reliable hazard prevention, planning and management. Therefore, our objectives were to develop an index of vulnerability to poor water quality based on simple measures of reservoir and catchment characteristics, and to test the index’s ability to predict this vulnerability using water quality and cyanobacteria data collected from 15 drinking water reservoirs in subtropical southeast Queensland. We expected that an effective index would show positive correlation with nutrients, chlorophyll a concentrations and algal and cyanobacterial cell densities in summer months.
2.
Materials and methods
2.1.
Study region
The 15 reservoirs examined in this study supply drinking water to the urban and semi-rural populations of southeast Queensland (c. 2.8 million) and vary in catchment size and full
supply capacity (Fig. 1, Table 1). In the past 30 years, average annual rainfall in the region has ranged from 800 to 1600 mm (www.bom.gov.au). Land use in the catchments is dominated by natural bushland and pastoral activities (mainly cattle grazing), with smaller proportions of cropping and residential lands (Fig. 2).
2.2.
Index calculation
Reservoir and catchment characteristics can be summarised by several parameters, many of which have been examined for their ability to explain variation in reservoir water quality (e.g. Forbes et al., 2008). The parameters we used to calculate the vulnerability index (VI) satisfied the following four conditions: (1) correlation with water quality was well established in the literature, either theoretically or empirically; (2) parameters were easily calculated from readily available data on reservoir or catchments characteristics; (3) parameters were not strongly correlated with each other (Spearman correlation, R < 0.70 and/or P > 0.05), and (4) parameters were relatively static or predictable through time so that the index was unaffected by substantial spatial and temporal variation (e.g. this would exclude parameters like nutrient concentrations and water transparency). This last condition also ensured that the index was not self-forecasting (e.g. high nutrient concentrations predicting that a reservoir was vulnerable to high nutrient concentrations). Five parameters satisfied the conditions listed above and the VI, which ranges from 0 (lowest vulnerability) to 1 (highest vulnerability), was calculated as follows:
VI ¼ (percentage grazing land covera þ reservoir shoreline to surface area ratiobc þ reservoir volume at full supply capacityab þ reservoir volume to catchment area ratiobc þ age since dam constructionab)/5 where a is the range standardised so that the highest value ¼ 1 and the lowest ¼ 0; b,log transformed to reduce skew; c, range standardised so that the highest ratio ¼ 0 and the lowest ¼ 1. Log transformation and range standardisation gave each parameter an equivalent weighting in the formula and created a comparative index of vulnerability among the 15 reservoirs.
2.3.
Index testing
The ability of the index to predict the vulnerability of reservoirs to poor water quality and algal blooms was assessed by testing the correlation between index scores and water quality parameters in the 15 reservoirs. Each reservoir was sampled once between 9 February and 3 March 2009 in the late summer period. Over the past 30 years in the study area, mean rainfall and temperature during February has been 100e300 mm and 18e27 C (minemax ranges, www.bom.gov.au). Heavy rainfall was experienced while sampling Kurwongbah reservoir and this rain event caused overflow at the dam walls of both Kurwongbah and Somerset at the time of sampling. There was also a bloom of the toxic cyanobacterium Cylindrospermopsis raciborskii in Borumba reservoir during sampling. Three
4489
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 8 7 e4 4 9 6
151oE
152oE
153oE
25oS
25oS
Pacific ocean
26oS
26oS
Macdonald Borumba Cooloolabin Baroon
Ewen Maddock
27oS
27oS Somerset Kurwongbah North Pine Wivenhoe Manchester
28oS
Moogerah
N
Wyaralong
Leslie Harrison
Hinze
28oS
Little Nerang Maroon 50
25
151oE
0
50 kilometers
152oE
153oE
Fig. 1 e Dam wall locations of 15 reservoirs (closed triangles) in subtropical southeast Queensland, Australia, sampled during late summer 2009, and a new reservoir currently in construction (open triangle), shown with catchment boundaries of the major river systems in which the reservoirs are located.
reservoirs had destratification units in use near the dam wall to vertically mix the water: Leslie Harrison, Macdonald and North Pine (see Burford and O’Donohue, 2006). At least three sites were sampled in each reservoir. Sites were near the dam wall, mid-reservoir, and in the upstream section of each reservoir. Sampling was conducted at the deepest point of each site, which was at least 6 m (except for the upstream sites at Kurwongbah and Little Nerang which were 2e4 m in depth). Four sites were sampled in the larger reservoirs (e.g. Somerset) and in those with two major arms (Borumba, Hinze, Leslie Harrison and North Pine). In Wivenhoe, the largest reservoir out of the 15, only three sites were sampled due to an aquatic weed infestation restricting access to the most upstream reaches; however, the three sites were still representative of dam wall, mid-reservoir and upstream locations. At each site, a 3 m depth-integrated sample of surface water was collected with a modified PVC pipe. Bottom water
was collected with a van Dorn (3.2 L Vertical Beta Plus) sampler from 1 m above the bottom of each site. Surface and bottom water were each transferred to a clean bucket from which samples for water quality analyses were taken. Each process was repeated to obtain duplicate samples. Surface water was subsampled for analyses of total nitrogen and phosphorus (TN, TP; for both whole and dissolved fractions), dissolved inorganic nitrogen and phosphorus (DIN and DIP), and chlorophyll a (Chl) concentrations as well as algal identification and cell densities (cells mL1), including potentially toxic cyanobacterial species (Anabaena circinalis, Anabaena bergii, Aphanizomenon ovalisporum, C. raciborskii and Microcystis aeruginosa). Bottom water subsamples were analysed for these same parameters, excluding chlorophyll a concentration and algal identification. All subsamples that were analysed for dissolved fractions were pre-filtered, in situ, through 0.45-mm membrane filters (Millipore) and
4490
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 8 7 e4 4 9 6
Table 1 e Physical characteristics of 16 subtropical reservoirs and their catchments. Code
Reservoir
Shore (km)
SA (km2)
Vol (ML)
Shore: SA (km1)
Shore: vol (km-2)
Depth (m)
Age (y)
CA (ha)
Vol:CA (ML ha1)
Bar Bor Cool Ewen Hin Kur LHD LNer Mac Man Mar Moog NPD Som Wiv Wya
Baroon Borumba Cooloolabin Ewen Maddock Hinze Kurwongbah Leslie Harrison Little Nerang Macdonald Manchester Maroon Moogerah North Pine Somerset Wivenhoe Wyaralong
25.5 44.5 12.8 14.3 67.4 33.4 34.5 10.9 38.3 27.3 17.9 29.3 167 237 462 110
3.9 3.8 1.4 2.3 9.3 3.4 4.2 0.6 2.6 2.6 3.3 7.7 21.2 39.7 110 1.1
61,000 46,000 14,200 16,600 163,000 14,400 24,800 9280 8000 26,000 44,300 83,800 215,000 369,000 1,150,000 103,000
6.6 11.9 8.9 6.4 7.2 9.9 8.2 19.6 14.7 10.5 5.4 3.8 7.9 6.0 4.2 9.0
4.2 9.7 9.0 8.6 4.1 23.3 13.9 11.8 47.9 10.5 4.0 3.5 7.8 6.4 4.0 10.8
15.7 12.2 9.9 7.4 17.5 4.2 5.9 16.6 3.1 10.0 13.3 10.9 10.1 9.3 10.5 8.4
21 46 30 27 20 40 25 48 29 93 35 48 33 56 25
6530 46900 736 2130 17600 5250 8890 3600 4960 7260 10,500 22,700 34,900 133,000 568,000 54,590
9.3 1.0 19.3 7.8 9.3 2.7 2.8 2.6 1.6 3.6 4.2 3.7 6.2 2.8 2.0 1.9
e
Shore, shoreline perimeter; SA, reservoir surface area at full supply level (FSL); vol, water storage capacity at FSL; Depth (mean) ¼ Vol/SA; age, age to 2009 since the completion of dam construction; CA, catchment area; e, Wyaralong dam due for completion in 2011.
subsamples for algal identification were fixed with Lugol’s iodine solution (0.6% final concentration). Concentrations of dissolved organic nutrients (DON and DOP) were determined by subtraction (e.g. DON ¼ dissolved TN fraction e DIN). All samples were stored in the dark and on ice until transported to the laboratory. Nutrient analyses were conducted following standard colorimetric methods; chlorophyll a was extracted in 100% acetone and measured spectrophotometrically (American Public Health Association, 1985). DIN and DIP concentrations were often near or at detection limit (0.002 mg L1) and were not used in further analyses. Algal taxa were identified to species level, where possible, under 400 phase-contrast microscopy and cells were counted using a Sedgewick Rafter counting chamber (Lund et al., 1958; Burford et al., 2007). The top layer (c. 2 cm) of sediments at the bottom of each site was collected in duplicate using a weighted sediment corer. These samples were analysed for TN and TP concentrations as well as stable carbon and nitrogen isotope ratios (d13C and d15N &), which were determined using a mass
spectrometer (GV Isoprime, Manchester, UK), following standard methods (American Public Health Association, 1985; Jardine et al., 2003). Data was subjected to Spearman’s rank correlation as this test handles parameters with skewed distributions and/or heteroskedacity, which included the majority of the water quality parameters used in the analysis as well as the VI itself. Statistical significance was recorded at P < 0.05 and tests were conducted within the R stats package (www.r-project.org/). The strength (size of the correlation coefficient, R) and significance of correlations between VI and water quality parameters were also compared with those between the individual VI (grazing, shore: SA, Vol, Vol:CA, age) and water quality parameters. Data from a previous study on 7 of the 15 reservoirs were also used as an independent dataset to assess the VI’s effectiveness (Burford et al., 2007). In this study, water quality was assessed at two sites (dam wall and upstream) in each reservoir during spring (October 2004), early summer (December 2004), late summer (February 2005) and autumn (April 2005).
1.0 0.9 0.8 0.7 Other Cropping Urban Natural bush Grazing
0.6 0.5 0.4 0.3 0.2 0.1 0.0 Cool Ewen LNer Mac Kur Bar Man LHD Mar Hin MoogNPD Bor Som Wiv Increasing catchment area
Fig. 2 e Proportions of different land use cover in the catchments of 15 reservoirs. See Table 1 for the key to reservoir coding.
4491
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 8 7 e4 4 9 6
Correlations between these data and the VI were tested using the same methods outlined for the February 2009 study, with the VI calculated based on reservoir age in 2004 and 2005 for the corresponding years of sampling.
exceptions were Hinze (mid-reservoir only) and Maroon, in which Anabaena circinalis dominated. Five reservoirs had no potentially toxic cyanobacterial species identified (Cooloolabin, Macdonald, Ewen Maddock, Leslie Harrison and Little Nerang).
3.2.
3.
Results
3.1.
Water quality and algal cell densities (Feb 2009)
Index of vulnerability (VI)
Based on the index calculation outlined in the Methods, Wivenhoe reservoir had the highest level of vulnerability to poor water quality and algal blooms (VI ¼ 0.77), followed closely by Somerset (0.74) and Moogerah (0.67) (Fig. 4). Cooloolabin had the lowest VI (0.18). Ewen Maddock (0.29), Little Nerang (0.30), Hinze (0.33) and Leslie Harrison (0.36) also had low indices. The remaining reservoirs had intermediate VI scores (Macdonald ¼ 0.43, Manchester ¼ 0.44, Baroon ¼ 0.46, Maroon ¼ 0.48, Kurwongbah and North Pine ¼ 0.50, Borumbah ¼ 0.52).
Mean surface water temperatures, integrated over the top 3 m, ranged from 26.7 0.5 C at Baroon reservoir to 28.7 0.7 C at Moogerah. The exception was Little Nerang reservoir, which had a mean surface water temperature of 22.5 1.8 C. Within each reservoir, TN and TP concentrations were generally higher in bottom waters than surface waters (see Accessory Publication). Borumba, Moogerah and Manchester had the highest mean concentrations of TN, in both and bottom waters surface (0.66e0.88 mg L1) (1.18e1.31 mg L1). TP concentrations were similar among most reservoirs; mean concentrations were all <0.05 mg L1 except for the bottom waters of Borumba, Somerset, Wivenhoe, Maroon and Moogerah (0.135e0.372 mg L1). Many reservoirs had high proportions of DON in surface waters and high mean concentrations of DOP were found in bottom waters of Borumba, Wivenhoe, Somerset, Maroon and Moogerah (0.094e0.271 mg L1). The algal composition of surface water samples from all reservoirs, in terms of mean cell densities, was dominated by cyanobacteria (Fig. 3; Accessory Publication). Borumba had the highest mean algal cell density of all reservoirs (455,000 91,000 cells mL1), followed by Moogerah, Somerset and Wivenhoe (means > 200,000 cells mL1). Hinze had the lowest mean algal cell density (20,000 5000 cells mL1). Potentially toxic cyanobacteria, when present within reservoirs, were most often dominated by C. raciborskii. The
3.3.
Index performance
Statistically significant correlations between water quality parameters and the VI scores were all positive (Table 2). These correlations were also stronger and more often statistically significant than the correlations between water quality parameters and each of the five parameters used to calculate the VI (Table 3). Significant correlations were found between the VI and both the densities and proportions of algal cells in all study periods, except October 2004 (early spring) for which correlations were all non-significant. Among the significant correlations, the strongest were with total algal and cyanobacterial cell densities, and the strongest of these were found in February 2009 (R ¼ 0.82 and 0.83) and December 2004 (R ¼ 0.86 and 0.86). Correlations with the proportion of cyanobacterial cells were strongest during the 2004e2005 study (R ¼ 0.74e0.82). February was the only month for which significant correlations with potentially toxic species were detected (2005: R ¼ 0.69 for density only; 2009: R ¼ 0.71 and 0.64
600000 Total algae Cyanobacteria Potentially toxic taxa
Cells mL-1
500000
400000
300000
200000
100000
0 Bar
Bor
Cool
Ewen
Hin
Kur
LHD
LNer
Mac
Man
Mar
Moog
NPD
Som
Wiv
Fig. 3 e Algal densities (cells mLL1) within reservoirs (means with standard errors as bars) sampled in February 2009. Potentially toxic cyanobacteria include Anabaena circinalis, Aphanizomenon ovalisporum, Cylindrospermopsis raciborskii and Microcystis aeruginosa.
4492
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 8 7 e4 4 9 6
2009 Vulnerability Index (VI)
1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 Cool Ewen LNer Hin LHD Mac Man Bar Mar Kur NPD Bor Moog Som Wiv
Fig. 4 e Vulnerability Index for 15 reservoirs in subtropical Queensland (based on ages of reservoirs in 2009). See Table 1 for the key to reservoir coding.
for density and proportion respectively). This was also the case for chlorophyll a concentrations (February 2005, R ¼ 0.56; February 2009, R ¼ 0.41). In addition, correlations with nutrient concentrations measured in all months (water column TN and TP) were only significant in February 2009 (except for surface TN in December 2004; R ¼ 0.55) and the strongest was with TP in bottom waters (R ¼ 0.82). The VI also correlated well with sediment nutrient and carbon data, in particular with d13C (R ¼ 0.72), which were measured in February 2009 only.
4.
densities, including potentially toxic cyanobacteria, and the proportions of cyanobacteria within the subtropical reservoirs during summer months. Cyanobacteria are capable of regulating their buoyancy, surviving low light conditions, storing nutrients and utilising forms of nutrient that are inaccessible to other taxa, all of which allows them to dominate the algal community under various physicochemical conditions (Padisa´k, 1997; Burford et al., 2007; Posselt et al. 2009). Given this flexibility, the ability of the VI to reflect increased summer densities and proportions of cyanobacteria, based on physical characteristics of the reservoirs and catchments alone, suggests that it may be more capable of assessing bloom susceptibility than traditional measures based on nutrient concentrations, trophic status or light availability (e.g. Downing et al., 2001). Our index also performed better than
Discussion
The index of vulnerability to poor water quality and algal blooms correlated strongly and significantly with algal cell
Table 2 e Correlation between the VI and water quality parameters (mg LL1 unless otherwise indicated) within 15 subtropical reservoirs. Italic text indicates significant correlations (P < 0.05). Oct 2004a
1
Algae (cells mL ) Cyano (cells mL1) Cyano (%) Toxic (cells mL1) Toxic (%) Chl (mg L1) TN TN TN (mg kg1) TP TP TP (mg kg1) DON DON DOP DOP d13C (&) d15N (&)
S S S S S S S B Sed S B Sed S B S B Sed Sed
Dec 2004a
Feb 2005a
Apr 2005a
Feb 2009
R
P
n
R
P
n
R
P
n
R
P
n
R
P
n
0.14 0.10 0.10 0.20 0.35 0.17 0.51 0.33
0.6727 0.7633 0.7175 0.5014 0.2293 0.5621 0.0652 0.2520
0.86 0.86 0.74 0.20 0.00 0.28 0.55 0.40
0.0001 0.0001 0.0023 0.4797 0.9758 0.3260 0.0414 0.1574
0.0027 0.0019 0.0015 0.0061 0.1205 0.0394 0.0676 0.7743
0.0010 0.0013 0.0003 0.9400 0.5543 0.5634 0.0933 0.5861
0.10 0.33
0.7515 0.2386
0.10 0.26
0.7978 0.3747
14 14 14 14 14 14 14 14 0 14 14 0 0 0 0 0 0 0
0.61 0.59 0.68 0.00 0.03 0.03 0.22 0.03
0.3017 0.5856
14 14 14 14 14 14 14 14 0 14 14 0 0 0 0 0 0 0
0.73 0.75 0.76 0.69 0.44 0.56 0.50 0.10
0.30 0.17
14 14 14 14 14 14 14 14 0 14 14 0 0 0 0 0 0 0
0.01 0.13
0.7504 0.2018
14 14 14 14 14 14 14 14 0 14 14 0 0 0 0 0 0 0
0.82 0.83 0.37 0.71 0.64 0.41 0.66 0.50 0.10 0.50 0.82 0.51 0.55 0.28 0.10 0.66 0.72 0.57
<0.0001 <0.0001 0.0064 <0.0001 <0.0001 0.0028 <0.0001 0.0002 0.4905 0.0002 <0.0001 0.0002 <0.0001 0.0454 0.4886 <0.0001 <0.0001 <0.0001
50 50 50 50 50 50 50 49 48 50 49 48 50 49 50 49 42 42
a Raw data sourced from a previous study (Burford et al., 2007). Cyano, cyanobacteria; toxic, potentially toxic cyanobacteria; S, surface; B, bottom; sed, sediment.
4493
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 8 7 e4 4 9 6
Table 3 e A comparison of correlation between the VI versus individual physical parameters and water quality within 15 subtropical reservoirs sampled in February 2009. Italics text indicates significant correlations (P < 0.05). n
1
Algae (cells mL ) Cyano (cells mL1) Cyano (%) Toxic (cells mL1) Toxic (%) Chl (mg L1) TN (mg L1) TN (mg L1) TN (mg kg1) TP (mg L1) TP (mg L1) TP (mg kg1) DON (mg L1) DON (mg L1) DOP (mg L1) DOP (mg L1) d13C (&) d15N (&)
S S S S S S S B Sed S B Sed S B S B Sed Sed
50 50 50 50 50 50 50 49 48 50 49 48 50 49 50 49 42 42
Grazing
Shore:SA
Vol
Vol:CA
Age
VI
R
P
R
P
R
P
R
P
R
P
R
P
0.41 0.47 0.47 0.34 0.32 0.14 0.17 0.04 0.07 0.46 0.59 0.42 0.05 0.16 0.28 0.47 0.74 0.62
0.0029 0.0006 0.0007 0.0163 0.0255 0.3237 0.2276 0.7829 0.6491 0.0008 0.0000 0.0026 0.7519 0.2744 0.0456 0.0005 0.0000 0.0000
0.26 0.35 0.65 0.06 0.00 0.36 0.05 0.18 0.11 0.04 0.42 0.47 0.09 0.02 0.26 0.39 0.28 0.38
0.0698 0.0115 0.0000 0.6921 0.9992 0.0095 0.7136 0.2004 0.2628 0.7591 0.0025 0.0000 0.5319 0.8914 0.0677 0.0056 0.0043 0.0061
0.53 0.56 0.36 0.50 0.44 0.05 0.22 0.23 0.48 0.17 0.53 0.48 0.14 0.04 0.03 0.50 0.44 0.71
0.0001 0.0000 0.0108 0.0002 0.0015 0.7532 0.1023 0.1265 0.0004 0.2404 0.0001 0.0004 0.3189 0.7948 0.8251 0.0002 0.0015 0.0000
0.47 0.41 0.10 0.26 0.22 0.62 0.48 0.19 0.03 0.50 0.39 0.03 0.35 0.24 0.15 0.29 0.12 0.10
0.0006 0.0028 0.4764 0.0733 0.1319 0.0000 0.0005 0.1779 0.8373 0.0002 0.0050 0.8373 0.0117 0.0914 0.2946 0.0379 0.3963 0.4861
0.51 0.45 0.04 0.35 0.29 0.57 0.53 0.67 0.25 0.33 0.38 0.08 0.50 0.19 0.07 0.31 0.10 0.22
0.0001 0.0009 0.8083 0.0126 0.0389 0.0000 0.0001 0.0000 0.0854 0.0185 0.0064 0.5665 0.0002 0.6229 0.1820 0.0260 0.5050 0.1282
0.83 0.83 0.38 0.70 0.64 0.41 0.66 0.50 0.10 0.50 0.82 0.51 0.55 0.28 0.10 0.66 0.72 0.58
0.0000 0.0000 0.0064 0.0000 0.0000 0.0028 0.0000 0.0002 0.4905 0.0002 0.0000 0.0002 0.0000 0.0454 0.4886 0.0000 0.0000 0.0000
Cyano, cyanobacteria; toxic, potentially toxic cyanobacteria; S, surface; B, bottom; sed, sediment.
each of the VI parameters alone, which further supported its ability to comparatively assess the reservoirs’ vulnerability. Overall, our analyses suggest that strong links exist among the physical environment of dammed river systems, their physicochemical characteristics and algal ecology, although further work is required to understand and show causality. Land use, particularly animal agriculture, has been implicated in the eutrophication of streams, rivers, lakes and reservoirs the world over (Søndergaard and Jeppesen, 2007). In our study of subtropical reservoirs, 12 out of the 18 water quality parameters analysed showed significant correlation with the percentage of grazing land cover in the reservoirs’ catchments. Reservoirs with lower reservoir volume to catchment area ratios are more likely to have stronger links with catchment characteristics, including land use and the consequent reduction in water quality, than those with higher ratios, regardless of climatic zone (cf. Burford et al., 2007). Indeed, lower ratios have been linked with higher concentrations of chlorophyll a and total phosphorus in temperatezone reservoirs, Ohio USA (Knoll et al., 2003). Physical characteristics of reservoirs also affect internal water quality. For example, increased water residence time, through the increased net loading of nutrients in reservoirs, has long been implicated in the promotion of algal blooms and reduced water quality (Søballe and Kimmel, 1987; Harris, 2001). We did not include this parameter in our index, however, as the data needed to calculate residence time was not available for all reservoirs. In addition, water levels in the reservoirs fluctuate through time due to variation in inflow and outflow volumes and timing, such that residence time is not constant. Rather, we used age since completion of the dam wall as an alternative indicator of the nutrient loading capacity of reservoirs. Older reservoirs were assumed to have increased stores of nutrients, and strong correlations between nutrient concentrations in the water column and reservoir age
were found. The specific processes linking reservoir age to present-time water quality are not clear; however, it may be that as reservoirs age, sediment loading into reservoirs results in siltation and reduced water depth, particularly in the upper reaches. Nutrients released from sediments would therefore be more readily available for algal growth in surface waters, consistent with increased benthic-pelagic coupling (see also No˜ges et al., 1999). In addition, our index was based on the ratio of reservoir shoreline length to surface area. Reservoirs with lower ratios (shoreline length to surface area or reservoir volume) are likely to have a stronger pelagic and hypolimnic influence on reservoir water quality and ecosystem processes than those with higher ratios (Wetzel, 2001). For the tropics and subtropics in particular, reservoirs that have a greater proportion of pelagic than littoral habitat may become more susceptible to poor water quality when internal processes, such as stratification and sediment remineralisation, start to affect water quality in the summer months (Jones and Poplawski, 1998). The combination of these parameters (percentage grazing, shoreline to surface area ratio, reservoir volume, volume to catchment area ratio and reservoir age) produced a good index of vulnerability to poor water quality and algal blooms in the subtropical reservoirs, and in particular, to increased cyanobacterial densities and proportions in summer months. Correlation between the VI and pre-summer water quality (Oct 2004) was not detected. Algal composition during this month was significantly different to summer and postsummer months (Burford and O’Donohue, 2006; Burford et al., 2007) and recent studies of Wivenhoe reservoir suggest that this pre-summer/summer difference may be expected for other reservoirs in the study region (P. Muhid, unpublished results). As expected, correlations were highly significant in summer months (Dec 2004, Feb 2005, Feb 2009), which included both small (n ¼ 14) and larger (n ¼ 42e50) datasets.
4494
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 8 7 e4 4 9 6
The index also showed positive correlation with nutrient concentrations and/or stable isotope ratios in the water column and sediments measured during the 2009 summer sampling period. Warm temperatures and stratification can lead to sediment remineralisation and the renewed availability of nutrients to cyanobacteria. In addition, enriched carbon and nitrogen isotope ratios in reservoirs have been linked to increased autotrophic production (for carbon) and inputs of nitrogen associated with agricultural or urban land use (Leavitt et al., 2006; Wu et al., 2006; Tomaszek et al., 2009). In summary, our analysis indicated that the VI may be useful for assessing summer bloom vulnerability in subtropical reservoirs. However, the link between summer rainfall and reservoir water quality, like that with summer temperatures, was inherent in the reasoning behind the index. Southeast Queensland experienced drought conditions (minimum summer rainfall) between c. 2002 and December 2009. In conjunction with summer temperatures, the recent rainfall and inflow events prior to the February 2009 sampling period, although not substantial, may have improved the VI’s performance in comparison with the other sampling periods, including February 2005. As such, the risk of summer bloom events in reservoirs with high index scores may decrease in drought conditions. However, if water depth declines to a critical threshold where sediment remineralisation processes promote algal growth throughout the water column, summer blooms may be inevitable (P. Muhid, unpublished results), particularly given the known effect of increased temperatures on algal growth. The ultimate aim of the VI is to provide water authorities and managers with a rapid tool to confidently assess how vulnerable a reservoir is (or may be) to poor water quality, and in particular, cyanobacterial blooms. For example, we applied the index to a new dam (Wyaralong) being constructed about 20 km northeast of Maroon reservoir, to compare its potential vulnerability with the 15 reservoirs examined above (Fig. 5). Construction is scheduled for
One year after construction
Five years after construction
completion by end 2011. Based on planned dimensions and current grazing cover (46.9%), Wyaralong’s VI (0.55) predicts mid-range vulnerability, comparative with reservoirs like Maroon and Kurwongbah, for at least 5 years after competition (Table 1, Fig. 5). However, 10 years after completion, the VI is more comparable with that of Baroon, and 20 years after completion with Borumbah and North Pine. After 100 years, the VI is at the higher end of vulnerability to eutrophication and cyanobacterial blooms, such that Wyaralong is the fourth most vulnerable reservoir with respect to Moogerah, Somerset and Wivenhoe (Fig. 5). A simple exercise in decreasing or increasing the percentage of grazing land cover in Wyaralong’s catchment by 10% via reforestation, predicts that the VI will either reach the same endpoint (the fourth most vulnerable reservoir) after only 5 years (þ10% grazing cover) or remain below this point for at least 100 years (10% grazing cover) (all other parameters except age were unchanged for all reservoirs; Fig. 5). This demonstrates that the VI could provide input to the planning of new reservoirs and assist in decision making about investment to mitigate for adverse water quality outcomes. This may include such comparisons as costs of land use change versus increased treatment and may lead to the expansion of impact assessments to include the possibility of new reservoirs meeting water quality targets and to consider the potential impacts of algal blooms. Our paper encapsulates the physical characteristics of a group of reservoirs and their catchments into an effective indicator of the potential for summer blooms and water quality issues. However, the ability of the VI to predict the comparative susceptibility to summer blooms of cyanobacteria and eutrophic conditions was assessed for a limited number of reservoirs and in the subtropics alone. Adaptations may be required to achieve an acceptable level of correlation between the VI and water quality parameters in any one set of reservoirs (e.g. using residence time instead of reservoir age to calculate the index). To confirm the generic usefulness of the
20 years after construction
1.0 0.8
100 years after construction
Wyaralong under current grazing cover (47%)
Wyaralong under current grazing cover (47%)
Wyaralong under current grazing cover (47%)
Wyaralong under current grazing cover (47%)
Decreased g r a z i n g c ov e r (-10%)
Decreased grazing cover (-10%)
Decreased grazing cover (-10%)
Decreased grazing cover (-10%)
Increased grazing cover (+10%)
Increased grazing cover (+10%)
Increased grazing cover (+10%)
Increased grazing cover (+10%)
0.6 0.4
Vulnerability Index
0.2
1.0 0.8 0.6 0.4 0.2
0.8 0.6 0.4 0.2
Fig. 5 e Vulnerability Index for 16 reservoirs in subtropical Queensland, at one, five, twenty and one hundred years since the planned completion of Wyaralong dam wall in 2011, given: the current percentage of grazing land cover in Wyaralong catchment (top row); minus 10% (middle row); plus 10% (bottom row). Unbroken arrows show the change in the level of among-reservoir vulnerability between grazing cover scenarios. Broken arrows show the change in among-reservoir vulnerability through time. See Table 1 for the key to reservoir coding.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 8 7 e4 4 9 6
VI, similar tests are recommended in other reservoirs within subtropical, tropical and even temperate climates.
5.
Conclusions
This is the first index to encapsulate the physical characteristics of subtropical reservoirs and their catchments into an effective indicator of summer bloom vulnerability. The index of vulnerability to poor water quality and cyanobacterial blooms in the subtropical reservoirs examined in this study was based on the percentage of agricultural land use in catchments and physical characteristics of reservoirs. The index correlated strongly with increased cyanobacterial cell densities in summer months, as well as their proportional contribution to the total algal density. The index has the capability to predict vulnerability to poor water quality and summer blooms of cyanobacteria in subtropical, and potentially, tropical and temperate-zone reservoirs. With climate change, continued river impoundment and the growing demand for potable water, our index may provide decision making support when planning reservoirs, in the subtropics and elsewhere, to reduce their vulnerability to cyanobacterial bloom events.
Acknowledgements We thank rangers, boat drivers and water quality monitoring staff at Seqwater; Stephen Faggotter and Deb Gale for their assistance with sampling; Queensland Forensic and Scientific Services for nutrient and chlorophyll analyses and algal identification, in particular Gary Prove, Eugene Lee, Priya Muhid and Karen Reardon; Jason Kerr and Rene Diocares for stable isotope sample preparation and analysis at Griffith University; Seqcatchments (Brisbane), Monica Poel and Leanne Bowen for catchment boundary and land use spatial data, which was based on Statewide Land cover and Trees Study data from the State of Queensland (Department of Natural Resources and Water) 2006 (http://www.nrw.qld.gov. au/slats/index.html). This work was supported and funded by Queensland Bulk Water Supply Authority (Seqwater). We are grateful to two anonymous reviewers for their comments on our manuscript.
Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.watres.2010.06.016.
references
American Public Health Association, 1985. Standard Methods for the Examination of Water and Wastewater. APHA-AWWAWEF, Washington, DC.
4495
Arbuckle, K.E., Downing, J.A., 2001. The influence of watershed land use on lake N:P in a predominantly agricultural landscape. Limnology and Oceanography 46 (4), 970e975. Blanch, S., 2008. Steps to a sustainable Northern Australia. Ecological Management & Restoration 9 (2), 110e115. Bouvy, M., Dalcao, D., Marinho, M., Pagano, M., Moura, A., 2000. Occurrence of Cylindrospermopsis raciborskii (Cyanobacteria) in 39 tropical reservoirs during the 1998 drought. Aquatic Microbial Ecology 23 (1), 13e27. Burford, M.A., Johnson, S.A., Cook, A.J., Packer, T.V., Taylor, B.M., Townsley, E.R., 2007. Correlations between watershed and reservoir characteristics, and algal blooms in subtropical reservoirs. Water Research 41 (18), 4105e4114. Burford, M.A., O’Donohue, M.J., 2006. A comparison of phytoplankton community assemblages in artificially and naturally mixed subtropical water reservoirs. Freshwater Biology 51 (5), 973e982. Davis, J.R., Koop, K., 2006. Eutrophication in Australian rivers, reservoirs and estuaries e a southern hemisphere perspective on the science and its implications. Hydrobiologia 559 (1), 23e76. De Ceballos, B.S.O., Konig, A., De Oliveira, J.F., 1998. Dam reservoir eutrophication: a simplified technique for a fast diagnosis of environmental degradation. Water Research 32 (11), 3477e3483. Downing, J.A., Watson, S.B., McCauley, E., 2001. Predicting Cyanobacteria dominance in lakes. Canadian Journal of Fisheries and Aquatic Sciences 58 (10), 1905e1908. Dudgeon, D., 2000. Large-scale hydrological alterations in tropical Asia: prospects for riverine biodiversity. Bioscience 50 (9), 793e806. Forbes, M., Doyle, R., Scott, J., Stanley, J., Huang, H., Brooks, B., 2008. Physical factors control phytoplankton production and nitrogen fixation in eight Texas reservoirs. Ecosystems 11 (7), 1181e1197. Gu¨cker, B., Boe¨chat, I.G., Giani, A., 2009. Impacts of agricultural land use on ecosystem structure and whole-stream metabolism of tropical Cerrado streams. Freshwater Biology 54 (10), 2069e2085. Harris, G.P., 2001. Biogeochemistry of nitrogen and phosphorus in Australian catchments, rivers and estuaries: effects of land use and flow regulation and comparisons with global patterns. Marine and Freshwater Research 52 (1), 139e149. Jardine, T.D., McGeachy, S.A., Paton, C.M., Savoie, M., Cunjak, R.A. , 2003. Stable isotopes in aquatic systems: sample preparation, analysis, and interpretation. Canadian Manuscript Report of Fisheries and Aquatic Sciences No. 2656. Jones, G.J., Poplawski, W., 1998. Understanding and management of cyanobacterial blooms in sub-tropical reservoirs of Queensland, Australia. Water Science and Technology 37 (2), 161e168. Knoll, L.B., Vanni, M.J., Renwick, W.H., 2003. Phytoplankton primary production and photosynthetic parameters in reservoirs along gradient of watershed land use. Limnology and Oceanography 48 (2), 608e617. Leavitt, P.R., Brock, C.S., Ebel, C., Patoine, A., 2006. Landscapescale effects of urban nitrogen on a chain of freshwater lakes in central North America. Limnology and Oceanography 51 (5), 2262e2277. Lund, J.W.G., Kipling, C., Cren, E.D., 1958. The inverted microscope method of estimating algal numbers and the statistical basis of estimations by counting. Hydrobiologia 11 (2), 143e170. L’Vovich, M.I., 1990. In: Turner II, B., Clark, W., Kates, R., Richards, J., Mathews, J., Meyer, W. (Eds.), Earth as Transformed by Human Action: Global and Regional Changes in the Biosphere Over the Past 300 Years. Cambridge University Press, New York, pp. 235e252. Matzinger, A., Schmid, M., Veljanoska-Sarafiloska, E., Patceva, S., Guseska, D., Wagner, B., Mu¨ller, B., Sturm, M., Wu¨est, A., 2007. Eutrophication of ancient Lake Ohrid: global warming
4496
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 8 7 e4 4 9 6
amplifies detrimental effects of increased nutrient inputs. Limnology and Oceanography 52 (1), 338e353. McGregor, G.B., Fabbro, L.D., 2000. Dominance of Cylindrospermopsis raciborskii (Nostocales, Cyanoprokaryota) in Queensland tropical and subtropical reservoirs: implications for monitoring and management. Lakes & Reservoirs: Research and Management 5 (3), 195e205. No˜ges, P., Tuvikene, L., No˜ges, T., Kisand, A., 1999. Primary production, sedimentation and resuspension in large shallow Lake Vo˜rtsja¨rv. Aquatic Sciences 61 (2), 168e182. Paerl, H.W., Huisman, J., 2008. Blooms like it hot. Science 320 (5872), 57e58. Padisa´k, J., 1997. Cylindrospermopsis raciborskii (Wo1oszynska) Seenayya and Subba Raju, an expanding highly adaptive cyanobacterium: worldwide distribution and review of its ecology. Archiv fur Hydrobiologie Supplementbund. Monographic Studies 107 (4), 563e593. Posselt, A.J., Burford, M.A., Shaw, G., 2009. Pulses of phosphate promote dominance of the toxic cyanophyte Cylindrospermopsis raciborskii in a subtropical water reservoir. Journal of Phycology 45 (3), 540e546. Pringle, C.M., 2001. Hydrologic connectivity and the management of biological reserves: a global perspective. Ecological Applications 11 (4), 981e998. Søndergaard, M., Jeppesen, E., 2007. Anthropogenic impacts on lake and stream ecosystems, and approaches to restoration. Journal of Applied Ecology 44 (6), 1089e1094.
Søballe, D.M., Kimmel, B.L., 1987. A large-scale comparison of factors influencing phytoplankton abundance in rivers, lakes, and impoundments. Ecology 68 (6), 1943e1954. Tomaszek, J.A., Koszelnik, P., Gruca-Rokosz, R., 2009. The distribution and isotopic composition of carbon and nitrogen as indicators of organic-matter fluxes in the Solina Reservoir (south-east Poland). Marine and Freshwater Research 60 (7), 647e652. Wetzel, R.G., 2001. Limnology: Lake and River Ecosystems. Academic Press, San Diego. Wiedner, C., Ru¨cker, J., Bru¨ggemann, R., Nixdorf, B., 2007. Climate change affects timing and size of populations of an invasive cyanobacterium in temperate regions. Oecologia 152 (3), 473e484. Wu, J.L., Lin, L., Gagan, M.K., Schleser, G.H., Wang, S.M., 2006. Organic matter stable isotope (d13C, d15N) response to historical eutrophication of Lake Taihu, China. Hydrobiologia 563 (1), 19e29. Yang, X., Anderson, N.J., Dong, X., Shen, J., 2008. Surface sediment diatom assemblages and epilimnetic total phosphorus in large, shallow lakes of the Yangtze floodplain: their relationships and implications for assessing long-term eutrophication. Freshwater Biology 53 (7), 1273e1290. Zengi, H., Lirong, S., Yu, Z., Chen, H., 2007. Post-impoundment biomass and composition of phytoplankton in the Yangtze River. International Review of Hydrobiology 92 (3), 267e280.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 3 0 e4 6 4 2
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Reduction of fecal indicator bacteria (FIB) in the Ballona Wetlands saltwater marsh (Los Angeles County, California, USA) with implications for restoration actions John H. Dorsey a,*, Patrick M. Carter a, Sean Bergquist b, Rafe Sagarin c a
Loyola Marymount University, Department of Natural Science, One LMU Drive, Los Angeles, CA 90045, USA Santa Monica Bay Restoration Commission, One LMU Drive, Los Angeles, CA 90045, USA c Institute of the Environment, University of Arizona, P.O. Box 210156, Tucson, AZ 85721, USA b
article info
abstract
Article history:
A benefit of wetland preservation and restoration is the ecosystem service of improving
Received 10 October 2009
water quality, typically assessed based on bacterial loading. The Ballona Wetlands,
Received in revised form
a degraded salt marsh of approximately 100 ac located on the southern border of Marina
30 April 2010
Del Rey (Los Angeles County, California, USA) are currently the focus of publicly funded
Accepted 6 June 2010
restoration planning. The wetlands receive tidal water, usually contaminated with fecal
Available online 12 June 2010
indicator bacteria (FIB: total and fecal coliforms, Escherichia coli, enterococci) from the adjacent Ballona Creek and Estuary. During the summer of 2007, two 24-h studies were
Keywords:
conducted to determine FIB tidal dynamics within the wetland. Measurements of water
Fecal indicator bacteria
flow and mean FIB concentrations (n ¼ 3) were measured every 1.5 h to determine total FIB
Wetlands
load estimates. FIB loading rates (MPN/s) were greatest during flood tides as water entered
FIB loading
the wetlands, and then again during spring tide conditions when sediments were resus-
Tidal flux
pended during swifter spring ebb flows. During daylight hours, the wetland acted as a sink for these bacteria as loads diminished, presumably by sunlight and other processes. Conversely, during late afternoon and night, the wetlands shifted to being a source as excess FIB departed on ebb flows. Therefore, the wetlands act as both a source and sink for FIB depending on tidal conditions and exposure to sunlight. Future restoration actions would result in a tradeoff e increased tidal channels offer a greater surface area for FIB inactivation, but also would result in a greater volume of FIB-contaminated resuspended sediments carried out of the wetlands on stronger ebb flows. As levels of FIB in Ballona Creek and Estuary diminish through recently established regulatory actions, the wetlands could shift into a greater sink for FIB. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Maintaining good water quality along beaches is a prime goal of resource managers. Contamination of recreational waters by sewage or runoff can lead to increased swimmer illness from exposure to water-borne pathogens and consequential
loss of millions of dollars for a region (Given et al., 2006). A variety of measures to reduce microbial loads impacting coastal waters have been implemented with varying levels of success (e.g. Dorsey, 2009). Among these, the use of natural or constructed wetlands appears to be a good strategy to reduce levels of pathogenic bacteria or fecal indicator bacteria (FIB:
* Corresponding author. E-mail address:
[email protected] (J.H. Dorsey). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.012
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 3 0 e4 6 4 2
total and fecal coliforms, Escherichia coli, enterococci). For example, in studying a series of constructed wetlands, Rifai (2006) documented an average FIB removal efficiency of 88%. Coastal salt marsh systems have been shown to be both sinks and sources of FIB. Steets and Holden (2003) found that a small coastal lagoon-wetland system in Santa Barbara, California, acted as a sink for FIB during the dry summer months. However, during wet-weather winter months, the system switched to a source of FIB to the adjacent surf zone. In this situation, bacteria were associated with sediments resuspended by increased runoff flowing through the lagoon system. During dry weather, Grant et al. (2001) determined that increased densities of enterococci in tidal water flowing from the Talbert Marsh system significantly impacted surf zone water quality along Huntington Beach, California. Later studies at this site showed that as increased volumes of contaminated runoff were diverted from the marsh into the sanitary sewers, indicator bacteria were more efficiently removed within the marsh by the natural processes (Jeong et al., 2008). Further, Jeong et al. (2008) calculated that the wetland could receive runoff equaling <1% of the average tidal prism volume (about 2 105 m3/d) before becoming a source of FIB to coastal waters. These studies imply that an urban saltwater marsh system could act as either a sink or source for FIB depending on tidal flows and the amount of urban runoff it receives. Here, we define a sink as an area whereby loads of FIB are reduced through various wetland processes. Conversely, the wetland would act as a source if FIB loads increase and are released to adjacent receiving waters. A management goal would be to understand factors influencing whether a wetland acts as a sink or source of FIB, and if these systems can be utilized to clean contaminated runoff, thus improving water quality along adjacent ocean beaches. In turn, this information can affect restoration design considerations, particularly for urban wetlands.
1.1.
The Ballona Wetlands
The Ballona Wetlands in Los Angeles County offer a good opportunity to address these questions. These wetlands are the last remaining major saltwater marsh in Los Angeles County (West, 2001). Surrounded by urban areas, the marsh has been degraded from numerous past activities, chief of which was construction of the Ballona Creek flood control channel, a largely cemented box or trapezoidal channel that diverted water from the Creek straight to Santa Monica Bay, thus severely limiting tidal flow to the marsh system (Phillip Williams & Associates, Ltd., 2006). Presently, the Ballona Wetlands are undergoing restoration planning under the direction of the California Coastal Conservancy (http://www. scc.ca.gov). These wetlands comprise a saltwater marsh system of approximately 100 ac located just south of Marina Del Rey (Fig. 1). Based on a series of recent monitoring studies by the City of Los Angeles and Keane Biological Consulting (2005), the marsh flat was characterized by a low diversity plant community dominated by the pickle weed Salicornia virginica, while the upper banks of the tidal channels are covered by stands of Fleshy jaumea, Jaumea carnosa. Invasive plant
4631
species were prevalent throughout the marsh habitat and upland areas. Plants dominating the intertidal sediments mainly were the algae Enteromorpha and Ulva. Within the tidal channels, corophid amphipods, spionid and capitellid polychaetes, and the California horn snail Cerithidea californica dominated a low diversity animal assemblage. Common fish included the California killifish Fundulus parvipinnis, the Topsmelt Atherinops affinis, and several species of goby (Arrow goby Clevelandia ios, Longjaw mudsucker Gillichthys mirabilis). Feeding on the infaunal and fish assemblage were a variety of shorebirds, particularly the Willet Catoptrophorus semipalmatus, and several species of larger wading birds, mainly the Great blue heron Ardea herodias, Snowy egret Egretta thula and Great egret Ardea alba. The marsh is subjected to mixed semidiurnal tidal flows through approximately 3.2 ac of tidal channels. Water enters wetlands from the Ballona Creek Estuary through a single selfregulating east tide gate during flood flows. This gate was constructed in 2003 replacing an older, non-functioning gate. The tidal flow is muted as the gate allows only a tidal height of up to 1.1 m. At medium to high tides, the water column within the tidal channels is stratified with the surface being more brackish (Dorsey, unpublished data), reflecting the input of freshwater from Ballona Creek upstream from the estuary. During ebb flows, water departs wetlands via the east tide gate and a second smaller west tide gate. This latter gate is a flapvalve system allowing water to flow out of the wetlands. At the lowest spring tides, the tidal channels are nearly completely drained with water depths of 10 cm or less. Dorsey (2006) found that FIB concentrations within the wetland tidal channels vary greatly with tidal flows and that the wetlands may act as a sink for FIB under flood-tide conditions. Four 12-h surveys within the wetlands to generate preliminary data on FIB tidal dynamics similarly demonstrated that densities changed by several orders of magnitude over the period, and that densities tended to diminish during daylight hours (Dorsey, unpublished data).
1.2.
This study
Previous work by Dorsey (2006) to describe FIB tidal dynamics within the Ballona Wetlands was based on samples collected only at peak flood and ebb tides, and four subsequent 12-h surveys (Dorsey, unpublished data) captured only a portion of the tidal cycle. In this study we ask under what tidal conditions does the Wetland act as a FIB sink or source. To address this question, we determined FIB loading and total FIB loads throughout 24-h tidal cycles by coupling measurements of FIB densities with water flow within the tidal channels. This approach enabled us to determine how the loading of bacteria changed over tidal periods.
2.
Methods
2.1.
Station locations
Water samples for FIB analyses, flow rate, and water quality measurements were collected at BW2 in the Ballona Wetlands positioned 35 m south of the East Tide gate (Fig. 1). This
4632
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 3 0 e4 6 4 2
Fig. 1 e Location of sampling sites in the Ballona Wetlands (BW2) and Estuary (BE). Images from Google Earth.
location was selected to measure water entering and leaving the wetlands. A second station (BE) was positioned in the Ballona Estuary 60 m east of the tide gate as a reference point to measure concentrations of FIB upstream of the gate. Sampling events were conducted only during dry weather conditions in an attempt to clarify FIB tidal dynamics. Previous studies showed that during rain events, concentrations of all FIB groups could increase by several orders of magnitude throughout the wetland (Dorsey, 2006).
2.2.
Sampling frequency
Two 24-h sampling events were conducted during the summer of 2007 to elucidate trends over several tidal cycles (Table 1). Replicate water samples (n ¼ 3) were collected every 1.5 h during each sampling event at the two wetland stations, and every 4.5 h at the estuary station. Water for bacterial and turbidity tests was collected in 125 ml sterile polypropylene
jars, placed on ice, and transported to the nearby laboratory at Loyola Marymount University for analyses.
2.3.
Bacterial and turbidity tests
Concentrations of FIB (total coliforms, E. coli, enterococci) were determined using defined substrate technology (APHA et al., 1998: Standard Methods Section 9223 B). Idexx media Colilert-18 was used for total coliforms and E. coli, and Enterolert media for enterococci (http://www.idexx.com). For each sample, 10 ml of sample was added to 90 ml of dilution water, sealed into Quanti-Tray 2000 97-well trays, then incubated 18e22 h at 35 C for total coliforms/E. coli, and 24 h at 41 C for enterococci. After incubation, reactive wells in the trays were counted and the most probable number (MPN) of bacteria/ 100 ml was determined for each sample. Turbidity (NTU) was determined using a HACH 2100N turbidimeter.
4633
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 3 0 e4 6 4 2
Table 1 e Sampling dates and tidal information for the 24h FIB surveys in the Ballona Wetlands (BW2) and Estuary (BE). Samples were collected every 1.5 h at BW2 and 4.5 h at BE. Date
Start time Time period (h) (h)
Tidal flow
Tidal range (m)
12e13 Jul’07
0445
0310e0952 0952e1404 1404e2025 2025e0353
Flood Ebb Flood Ebb
1.44 0.39 1.29 2.40
2e3 Aug’07
0600
0624e1253 1253e1839 1839e0030 0030e0657
Flood Ebb Flood Ebb
1.49 0.99 0.98 1.30
2.4.
Chemical and physical electronic measurements
During each sampling event, water quality measurements of depth (m), temperature ( C), salinity (ppt), pH, and dissolved oxygen (mg/L) were measured using a YSI 6600 EDS sonde positioned in the deepest part of the channel, and with the bottom of the sonde resting on the sediment surface. Measurements were made every 5 min during sampling events. Light intensity (lums/ft2) was measured during the 24h sampling events in 2007 at Station BW2 using a HOBO data logger positioned on the west channel bank where each set of water samples were collected.
2.5.
FIB loading estimates
2.5.1.
Tidal channel flow rate measurements
Tidal channel flow rate measurements were recorded every 1.5 h. Velocity in ft/s was determined at 1-ft intervals across the channel using a McBirney Flo-Mate (Model 2000) electromagnetic flow meter positioned approximately 2/3 of the distance between the channel bottom and water surface. For each interval, the flow rate (Q) in ft3/s was calculated as follows: Q ¼ ðwÞðdÞðvÞ
(1)
where w was the interval width (ft), d was the depth (ft), and v was the flow velocity (ft/s). The flow rates for each interval were then summed as follows to yield a total flow rate for the channel, QT, and then converted to metric units (m3/s): QT ¼
n X
ðQÞi
(2)
i¼1
For each of the total flow rates, a volume (V) in m3 of water moving into or out of the wetlands at that sample time was calculated by simply multiplying the total flow rate (QT) by the duration of flow (in seconds).
2.5.2.
FIB loading estimates
A loading rate of bacteria (MPN/s) was determined by multiplying the concentration of bacteria by the flow rate measured at each sample time. This approach needed to be modified since incoming/outgoing flows do not occur simultaneously and the concentration of bacteria and the flow rates are
constantly changing. Therefore, no direct comparison of influent and effluent loading rates is applicable. To mitigate this situation, a total load of bacteria entering the wetlands over the course of a flood tide and a total load of bacteria leaving the wetlands over the course of an ebb tide was calculated. The beginning and end of a tidal cycle from low to high and back to low tide was sampled for this experiment, and based on direct measurements of water depth collected with the YSI sonde (Section 2.4 above). The resulting flood load was divided by the ebb load to determine whether the wetlands acted as a source or sink. Similarly, Shellenbarger et al. (2008) used total loads of FIB to determine a source or sink term for system of managed tidal ponds in San Francisco Bay. The total load of each FIB group per sample (L) was determined by multiplying the volume of water moving (calculated above) by the mean FIB concentration (FIBc) measured at the sampling time. For example, FIB loads for Sample 1 (L1) would be calculated as follows: ðL1 Þ ¼ V1 ðFIBc1 Þ
(3) 3
where V1 was the volume in m , and FIBc1 was the indicator group’s concentration (MPN/100 ml) collected at a specific time t1. To calculate total FIB load for each flood (LF) and ebb (LE) tidal flows, the total ebb volume of water departing the wetlands from both tidal gates needed to be estimated. Of the total flow entering the wetlands, most of the ebb flow exits the wetlands from the east gate with the remainder exiting from the west gate. For the total load calculations described below, the ebb volume from the West Gate over the entire ebb period was estimated by subtracting the East Gate ebb volume from the Total Flood Volume: VEBB
WG
¼ VFLD VEBB
(4)
EG
The Total Ebb Volume for each time interval was then estimated by adding the East Gate Ebb volume and the West Gate Ebb volume together. Total Ebb Volume ¼ VEBB
EG
þ VEBB
WG
(5)
Where VEBB EG is the ebb volume measured at the East Tide Gate, VEBB WG is the ebb volume calculated for the West Tide Gate and VFLD is the total flood volume at the East Tide Gate. Depending on the tide characteristics and the operating condition of the tide gates, the percentage of the water leaving via each tide gate can vary significantly from day to day. Because of this, these calculations were performed for each time interval during each 24 h study. It is important to note that equal volumes of flood and ebb water were used in calculating the total load of bacteria. This was achieved by only using samples taken between the same depths in the tidal channel. Also note that that the infiltration, evapotranspiration, and inputs from other sources were not included in the model because in this system, which is overwhelmingly driven by tidal inputs and outflows from Ballona creek and the Pacific Ocean, these factors are relatively insignificant (Phillip Williams & Associates, Ltd., 2006). Tidal load ratios for each FIB group were calculated between similar depths in the tidal channel from low tide through the high tide crest then back through the following
4634
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 3 0 e4 6 4 2
equivalent low tide. Load estimates for each sampling time (L) were summed for each flood-flow and ebb-flow period over this truncated tidal cycle. For example, the total loading during the flood tide (LFF) would be calculated by the expression L1 þ L2. þ Lx ¼ LF. The total load for the flood flow then was divided by the total load for the ebb flow to yield a FIB tidal ratio (FIBR) for each indicator group as follows: FIBR ¼ LF =LE
2.5.3.
(6)
Correlations
Pearson correlation analyses were performed between FIB concentrations, YSI measured water depth at the time of sampling, and averaged data from the YSI sonde measurements. For each YSI parameter, means were determined by averaging three readings prior to each FIB collection time, the reading at the time of collection, and the subsequent three readings to yield n ¼ 7.
3.
Results
3.1.
Tidal flow rates and volumes
Flow rates ranged from <0.01 m3/s during slack water periods to 1.24 m3/s during the July sampling event (Fig. 2). The most rapid flow rates occurred approximately midway between the high and low tide peaks, especially when tidal ranges (Table 1) exceeded 1 m during spring tides. Maximum flow rates during each flood and ebb flow period were positively correlated with corresponding tidal ranges (Spearman r ¼ 0.67), but this relationship was not significant ( p ¼ 0.08). Based on calculated flood and ebb volumes (equation (4) above), we estimated that during the two tidal cycles of the July survey, 93% of the ebb flow departed the wetlands through the east tide gate during the first cycle followed by 70% during the second cycle. The first cycle had a relatively smaller tidal range compared to the second cycle that day (Fig. 2). During August, the two tidal cycles were more similar in their tidal ranges (Fig. 2) with similarly proportioned volumes leaving the wetlands through the east tide gate (first tide cycle equaled 77%, the second equaled 67%).
3.2.
Water quality measurements
Nearly 500 sets of measurements for temperature ( C), salinity (ppt), dissolved oxygen (mg/L) and pH were collected with the YSI sonde during the two 24-h sampling events. Summary statistics for these variables are presented in Table 2. In addition to the YSI recordings, 102 turbidity measurements were made in samples collected for bacterial analyses (Table 2). The trends of water quality variables generally appeared to be associated with tidal conditions and time of day. The range of salinities (22.31e33.62 ppt) reflected the estuarine nature of the water entering the wetlands. Salinity tended to be positively correlated with tidal heights (Table 3) where higher salinities occurred during flood tides and lower salinities were associated with ebb flows. Several actions could account for this relationship. Most likely, the wetland salinities are
Fig. 2 e Flow rates and corresponding water depths at Ballona Wetlands Station BW2 during the 24-h surveys in a) July, and b) August.
governed by water flooding in from the estuary where the brackish/freshwater lens in the estuary first enters the wetlands during the initial flood tide followed by more saline water. During the flood tide, the water column would become water progressively more stratified. During spring ebb flows, as measured in this study, the more brackish surface lens would eventually reach the channel bottom as the water level drops to 10 cm or less in the tide channel at maximum low tide. This scenario was best demonstrated during the July sampling event as water at the bottom of the tide channel (where the YSI conductivity sensor was positioned) became progressively more saline with the flood tide, then more brackish as the tide fell later in the day (Fig. 3). We assumed
Table 2 e Summary statistics for water quality variables measured at BW2. Values for water levels <0.2 m not used since sensors were exposed. Date
Variables Temp Salinity DO pH Turbidity (ppt) (mg/L) (NTU)a ( C)
12e13 Jul’07 n ¼ 228
Mean
22.42
30.53
6.67
8.11
3.77
S.D. Min Max
1.15 21.11 25.11
2.91 22.31 33.62
2.38 2.76 13.94
0.13 7.90 8.46
2.17 1.83 9.51
2e3 Aug’07 n ¼ 252
Mean
22.52
29.36
5.63
7.96
2.92
S.D. Min Max
1.31 20.62 25.73
2.01 22.57 31.85
3.53 1.00 15.48
0.20 7.64 8.47
1.92 1.27 7.60
a n ¼ 51 for the July and 51 for the August 2007 surveys.
4635
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 3 0 e4 6 4 2
Table 3 e Matrix of Pearson correlation coefficients (r) for FIB groups and mean water quality variables measured during the two 24-h sampling events. Significant correlations ( p < 0.05) indicated by bold, italicized font. Total
E. coli
Entero
Tide Hgt
Flow
Light
Wat Temp
12e13 July, 2007 E. coli 0.79 Entero 0.77 Tide Hgt 0.48 Flow 0.41 Light 0.16 Wat Temp 0.47 Sal 0.45 DO 0.12 pH 0.10 Turbidity 0.42
1.00 0.99 0.47 0.15 0.37 0.34 L0.68 0.28 0.06 0.65
1.00 0.47 0.10 0.33 0.35 L0.66 0.25 0.04 0.71
1.00 0.30 0.14 0.34 0.76 0.19 0.39 0.64
1.00 0.02 0.00 0.08 0.02 0.18 0.09
1.00 0.57 0.42 0.82 0.81 0.01
1.00 0.00 0.65 0.63 0.11
2e3 August, 2007 E. coli 0.79 Entero 0.54 Tide Hgt 0.25 Flow 0.00 Light 0.15 Wat Temp 0.01 Sal 0.19 DO 0.10 pH 0.02 Turbidity 0.07
1.00 0.62 0.23 0.14 0.00 0.03 0.27 0.32 0.19 0.20
1.00 0.06 0.23 0.08 0.14 0.33 0.19 0.04 0.35
1.00 0.01 0.28 0.40 0.58 0.17 0.25 0.54
1.00 0.12 0.08 0.04 0.07 0.08 0.26
1.00 0.35 0.64 0.51 0.56 0.20
1.00 0.29 0.29 0.20 0.02
Sal
DO
pH
1.00 0.23 0.05 0.62
1.00 0.95 0.04
1.00 0.18
1.00 0.47 0.41 L0.58
1.00 0.97 0.41
1.00 0.39
that water column became fully stratified and horizontally mixed throughout the wetland as high slack water was reached. Other sources of brackish and/or freshwater entering the tidal channels could be dry-weather runoff from the surrounding community, and intrusion of shallow groundwater (Saez, 2007). These latter two sources are not significant relative to estuarine tidal water entering the wetlands, and were not included in this study. Mean water temperatures generally reflected warmer temperatures during the summer with the greatest temperatures occurring in the afternoons during slack tide conditions. This relationship was best demonstrated during the July sampling event where water temperatures and light intensity were significantly correlated (Table 3; r ¼ 0.57, p < 0.05). Dissolved oxygen displayed a wide range with lowest values (e.g. 1.0 mg/L during the August survey) measured during predawn hours. Dissolved oxygen and pH were positively correlated during both sampling events, and also displayed positive correlations with water temperatures and light intensity during the July sampling event (Table 3). Turbidity negatively correlated with salinity (Table 3) indicating that more turbid water was associated with ebb flows when more brackish water was present. When mean turbidities were regressed against corresponding water flow rates (Fig. 4), the swifter flow rates corresponded with increased turbidities (r2 ¼ 0.97, p < 0.0001). Increased turbidities at the tide gate probably resulted from tidal scouring that resuspended sediments, mainly during spring tide ebb flows approaching minus tide conditions.
adjacent Ballona Estuary, and during the mid-point of the ebband flood-flows when turbidity spikes occurred. Within sampling events, mean densities of these indicator bacteria could span up to three orders of magnitude with total coliforms ranging from 102 to 104, and E. coli and enterococci from 101 to 103 MPN/100 ml. Concentrations of FIB at the estuary station were elevated during early morning hours, fell throughout the day, but rose again at night. During the July sampling event, the estuary station displayed a spike during the mid-point of the ebb flow, similar to that seen within the wetland. A ratio of mean FIB concentrations between the upstream estuary (BE) and wetland (BW2) stations was calculated for each sampling time to indicate the magnitude of estuary concentrations relative to those in the wetlands (Table 4).
3.3.
Fig. 3 e Relation between mean salinity (ppt) and water depth (m) during the July sampling event at BW2. Salinity was measured at a point approximately 5 cm off the channel bottom where the salinity sensor of the YSI sonde was positioned.
FIB concentrations and loading
The mean concentrations of total coliforms, E. coli, and enterococci (Appendix A) generally were greater during floodtide conditions as water entered the wetlands from the
4636
Flow (m3/s)
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 3 0 e4 6 4 2
during a single tidal cycle (low tide through high then back to low) are presented in Table 5. During daylight hours, the FIB tidal ratio (FIBR) was >1 for all three indicator groups indicating a reduction of bacteria during this period. The reduction of the total number of load of FIB during daylight hours ranged from approximately 60% total coliforms to 100% for enterococci. Conversely, during late afternoon and nighttime, the FIBR ratios all were <1 indicating that the wetlands were acting as a source of FIB. Increases in the total load of FIB ranged from around 270% for total coliforms to 836% for enterococci. These load increases during the nighttime resulted from the spikes of suspended solids during the strong ebb flow rates (Fig. 5). During the July survey densities of E. coli and enterococci displayed significant ( p < 0.05) correlations with increased turbidity (r ¼ 0.65 and 0.71, respectively) and lower salinities (r ¼ 0.68 and 0.66, respectively) associated with ebb tide conditions (Table 3). In contrast, during the August sampling event, none of the FIB groups formed significant correlations with the water quality parameters.
BW2, 2007
0.45 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05
r2=0.9704 p<0.0001 n=42
0.00 0
15 10 5 Mean Turbidity (NTU)
20
Fig. 4 e Linear regression between flow rate and mean turbidity at BW2. All data from the two surveys were pooled for the analysis.
Calculations of RE:W were followed by Student’s t-tests to determine if the means used to calculate the ratios differed significantly (Table 4). In the estuary at Station BE, concentrations of all three FIB groups tended to be significantly greater than those within wetland during the July survey as indicated by values of RE:W > 1.00 (Table 4). The highest RE:W values for all three FIB groups occurred during low slack water when the tidal height was less than 0.5 m (Fig. 2). Conversely, ratios of 1.0 or less tended to be associated with flood conditions as slack water was reached. Similar trends were seen during the August survey, although not as evident (Table 4). However, as in July, some of the greatest values of RE:W occurred during low slack water. The FIB loading rates (MPN/s) were greatest in early mornings during flood flows as water flowed into the wetland channel from Ballona Estuary (Fig. 5). Loading rates generally diminished throughout the day for all FIB groups although midflow spikes occurred, especially during spring tide ebbs (e.g. July at 12:00 AM). Total load estimates for flood and ebb flows
4.
Discussion
Three main points have emerged from this research: 1) densities of FIB groups can vary by up to three orders of magnitude within a 24-h period during dry weather; 2) the wetlands can act as a sink for these bacteria during daylight hours; and 3) although re-suspension of sediments during ebb flow can cause the wetlands to be a source of FIB to the estuary, water flowing from the wetlands can dilute the more contaminated receiving waters of the estuary.
4.1.
Variation in FIB densities
Water in the Ballona Estuary is a mixture of marine water from Santa Monica Bay and freshwater flowing into the
Table 4 e Ratio of mean FIB concentrations between the Estuary (BE) and Wetland (BW2) stations. Mean concentrations for FIB groups per site and per time are given in Appendix A, and were tested for significant differences using a Student-t test on log10-transformed data. Sample time
Total coliforms a
E. coli
Enterococci
Tidal state
RE:W
p
RE:W
p
RE:W
p
July 12e13 survey 5:51AM 10:27AM 2:58PM 7:30PM 12:00AM 4:20AM
1.00 2.93 1.81 0.24 3.40 7.16
Not testedb 0.0003* 0.0322* 0.0239* 0.0699 ns 0.0003*
1.75 3.58 1.00 1.00 8.37 12.70
0.0356* 0.0220* Not tested Not tested 0.0108* 0.0002*
1.08 3.85 1.67 1.00 9.63 19.28
0.4839 ns 0.0236* 0.3753 ns Not tested 0.0035* 0.0033*
Flood Ebb Flood High Slack water Ebb Low slack water
August 2e3 survey 7:30AM 12:00PM 4:30PM 9:00PM 1:30AM 6:00AM
1.00 1.00 1.64 0.69 0.79 1.31
Not tested Not tested 0.0015* 0.0138* 0.3077 ns 0.0271*
5.85 0.49 1.24 0.41 2.35 18.51
0.0268* 0.0666 ns 0.6037 ns 0.0520 ns 0.0604 ns 0.0001*
0.08 0.18 2.40 0.82 0.75 6.09
0.0007* 0.0266* 0.1180 ns 0.4552 ns 0.3808 ns 0.0001*
Flood High slack water Low slack water Flood High Slack water Low slack water
a *p 0.05; ns p > 0. b Means were equal, so not tested.
4637
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 3 0 e4 6 4 2
Fig. 5 e FIB loading rate estimates (MPN/s) for the two 24-h surveys conducted during 2007 at BW2 in the Ballona Wetlands: a) July sampling event, and b) August sampling event.
estuary from Ballona Creek that is contaminated by FIB. Estuarine surface salinities have been measured to range from <1 to 32 ppt depending on tidal conditions and runoff from rainfall (Dorsey, 2006), and are stratified with fresher water forming a surface lens, both within the estuary and wetland tidal channels (Dorsey, unpublished data). The main source of
lower salinity water within the wetlands most likely is tidal water from the estuary. Other sources of freshwater entering the tidal channels could be dry-weather runoff from the surrounding community, or groundwater intrusion, but we assume that these sources are minimal compared to tidal flows.
Table 5 e Total tidal loads during flood (TLF) and ebb (TLE) flows and the corresponding FIB Tidal Ratio (FIBR) during the 24-h surveys in 2007. TLE
TLF
FIBR
% Reduction () or increase (þ)
Date
Time (h)
Total coliforms 12e13 Jul 2007 12e13 Jul 2007 3-Aug-07 2e3 Aug-2007
0721e1321 1351e2416 0751e1825 1825e0600
8.93E 3.29E 1.57E 1.05E
þ þ þ þ
13 12 14 14
3.71E 2.94E 6.43E 1.21E
þ þ þ þ
13 13 13 14
2.41 0.11 2.44 0.87
58.45 793.62 59.04 15.24
E. coli 12e13 Jul 2007 12e13 Jul 2007 3-Aug-07 2e3 Aug-2007
0721e1321 1351e2416 0751e1825 1825e0600
2.82E 6.17E 2.64E 8.76E
þ þ þ þ
11 10 12 11
7.08E 2.31E 2.57E 8.76E
þ þ þ þ
10 11 11 11
3.98 0.27 10.27 1
74.89 274.39 90.27 0
Enterococci 12e13 Jul 2007 12e13 Jul 2007 3-Aug-07 2e3 Aug-2007
0721e1321 1351e2416 0751e1825 1825e0600
1.50E þ 12 6.26E þ 10 1.35E þ 13 4.80E þ 11
1.16E 5.86E 6.23E 7.15E
þ 11 þ 11 þ 10 þ 11
12.93 0.11 216.69 0.67
92.27 836.1 99.54 48.96
4638
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 3 0 e4 6 4 2
During this study, some of the greatest FIB concentrations were measured in water flooding in from the Ballona Estuary through the tide gates at BW2 (Fig. 5, Table 5, Appendix A). Ballona Creek is a major source of FIB entering the estuary, hence the wetlands. Upstream of the tidal prism, geometric mean densities of total coliforms have been measured to range from 102 to 103 MPN/100 ml for E. coli and enterococci, and 104 for total coliforms during dry weather (Stein and Tiefenthaler, 2004; Tiefenthaler et al., 2009). Dorsey and Lindaman (2004) measured similar FIB densities in the lower reaches of the Creek for E. coli and enterococci during dry weather with total coliforms averaging 104e105 MPN/100 ml. During this study, FIB concentrations in Ballona Estuary generally was an order of magnitude greater than those found in the wetlands (Appendix A), following similar trends with mean concentrations ranging from 102 to 104 MPN/100 ml for total coliforms, and 101 to 103 MPN/100 ml for both E. coli and enterococci. Once FIB-contaminated water entered the wetland tidal channels, densities of FIB diminished by up to two orders of magnitude over a period of several daylight hours, presumably due to sunlight and other processes. This variability demonstrates rapidly changing FIB densities associated with tides and time of day. Grant et al. (2001) found similar enterococci variability in the Talbert Marsh saltwater system at Huntington Beach, California. Working in the same locality, Boehm et al. (2002) measured similar variability along the open ocean beach at Huntington Beach where the surf zone was impacted by runoff from various sources. Boehm’s group found FIB to vary over time scales ranging from minutes to hours depending on tidal and daylight conditions. This short-term variability emphasizes how a single grab sample taken at a monitoring locality would potentially provide misleading information on whether or not water meets standards for recreational contact (e.g. swimming, surfing). Water quality monitoring of FIB by public health and other agencies typically is done by acquiring a single grab sample, usually at the same time of the day, with results available 18e24 h later. Boehm et al. (2002) stress that this strategy is flawed given how FIB can vary over short time periods resulting from changing environmental conditions, and that a more effective strategy would be to obtain the geometric mean from a series of sample collected over a longer period, such as a week. This approach would enable beach managers to judge water quality based on 30-day standards while still being able to respond to spikes in FIB concentrations measured from instantaneous samples.
4.2. FIB
The Ballona Wetlands as both source and sink for
It appeared that densities of the FIB groups diminished throughout the daylight hours (Table 5), although the poor correlation between FIB and light intensity (Table 3) suggests that the relationship between light and FIB densities is complex. Sunlight has been demonstrated to be a prime factor in reducing FIB densities within wetland and other water systems. Boehm et al. (2002) measured a one- to two-order log reduction of enterococci and E. coli in mesocosm experiments exposing unseeded FIB-contaminated water from an open
coastal beach to sunlight. Other studies have demonstrated the ability of sunlight to destroy FIB within experimental mesocosms (Noble et al., 2004), treatment wetlands (Jillson et al., 2001; Karathanasis et al., 2003; Mayo, 2004; Vymazal, 2005) and ponded treatment systems (Stinton et al., 2002; Whitman et al., 2008). Photochemically produced oxidants from the reaction of sunlight with organic matter appear to be the active agents that destroy bacteria (Chamberlin and Mitchell, 1978). Data presented herein suggest that various FIB reduction processes are at play within the Wetlands, and that effects of sunlight may interact with tidal flows and turbidity. Other FIB reduction processes likely include settling onto submerged plant surfaces (e.g. Karathanasis et al., 2003) and predation by protozoans (e.g. Surbeck et al., 2010). Further studies are needed to develop a model of FIB reduction and input pathways, similar to the box-model approach used by Shellenbarger et al. (2008) in modeling the FIB dynamics in ponds and sloughs bird habitats. The wetlands also acted as a source of FIB to the adjacent estuary as evident by the total load ratios during late afternoon (2:00 PM on) through nighttime hours (Table 5). Increased loads leaving the wetlands probably were associated with spikes in turbidity during swifter spring tide ebb flows when sediments harboring FIB were resuspended. During spring tide ebb flows, water depth in the tidal channel at BW2 was only about 10e15 cm. Therefore, it is possible that the lens of contaminated brackish surface water would add more bacteria to the FIB associated with resuspended sediments. Although FIB concentrations during the spikes were one to two orders of magnitude lower compared with night and early morning flood flows (Appendix A), the greater flow rate of water moving out of the wetlands resulted in higher loading estimates. At nearby Del Rey Lagoon, a similar association was found where concentrations of FIB correlated with increased turbidity during strong ebb flows (Dorsey et al., 2008). Resuspended sediments have proven to be a source of sediment-dwelling FIB. For example, Ferguson et al. (2005) showed that densities of enterococci were much greater in sediments impacted by contaminated runoff, indicating retention and growth. Similarly, FIB were found to be two to four times more concentrated in intertidal sediments impacted by sewage effluents than densities measured in the overlying water column (Shiaris et al., 1987). Evanson and Ambrose (2006) found that within a wetland system adjacent the Santa Ana River mouth (California), the sediment-associated FIB populations may be distinct from those in the overlying water column based on the ratio of Total Coliforms: E. coli (TC:EC). Sediment populations generally had ratios >10 compared to overlying water where TC:EC ratios were <10, suggesting possible human fecal contamination (Haile et al., 1999). Based on this ratio, resuspended sediments from this wetland were determined not to impact the adjacent coastal beach using this ratio. Rather, water quality along the beach was impacted by discharge from the Santa Ana River (Evanson and Ambrose, 2006). Within the Ballona Wetlands, additional studies like that of Evanson and Ambrose (2006) would be required to conclusively determine if increased FIB during ebb flows is caused by resuspended sediments.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 3 0 e4 6 4 2
A key question here is if FIB outwelling from the Ballona Wetlands could impact the adjacent surf zones along Playa Del Rey to the south, and Venice Beach to the north, similar to the situation described at Huntington Beach by Grant et al. (2001). This scenario is unlikely because the concentrations of FIB within the estuary tend to be greater relative to those outwelling from the wetlands (Table 4), even during strong ebb flows when the wetlands contributed their greatest loads of FIB (Fig. 5, Table 5). In essence, water exiting the wetland may act to dilute the more FIB-contaminated estuary water, thus reducing potential contamination of adjacent coastal beaches.
4.3.
Implications for wetland restoration
Presently, tidal channels within the main Ballona Wetlands encompass approximately 3.2% of the area. Restoration planning, now underway by the California Coastal Conservancy, calls for increasing the number of tidal channels. To date, three options have been developed of increasing in complexity and area covered by tidal or open water (http:// www.ballonarestoration.org). Increasing the tidal channels along with surface area of flooded banks will greatly increase FIB reduction. Channels should be designed to maximize flooding during high tides by grading channels to have gentle slopes of 5:1 or more. Gentler slopes also will create more intertidal area, thus increasing extent of plants and biodiversity of infauna and bird and fish predators. Increasing the extent of tidal channels comes with a tradeoff e increased FIB entering the estuary during spring tide ebb flows associated with resuspended sediments. Therefore, channels should be designed to reduce flow velocity, thus minimizing re-suspension. Further, increasing the extent of intertidal areas will increase diversity and abundance of birds, especially waders and shorebirds that have proven to be a significant source of FIB (Ricca and Cooney, 1998; Alderisio and DeLuca, 1999). We predict that this balance of FIB sink vs. source may shift to that of the wetlands acting primarily as a sink particularly as concentrations of FIB diminish in Ballona Creek and Estuary due to pollution abatement measures. A Total Maximum Daily Load (TMDL) to control levels of FIB in Ballona Creek and Estuary was adopted by the California State Water Resources Control Board and the U.S. Environmental Protection Agency in 2007 (http://www.waterboards. ca.gov/losangeles/water_issues/programs/tmdl/tmdl_list. shtml). This TMDL establishes the maximum limit of FIB that can be discharged into these waters during dry weather without requiring corrective measures. Control measures will be implemented over about a 13e14 yr period via the Municipal Stormwater NPDES permit issued to the County of Los Angeles and co-permittee cities within the Ballona Watershed. Water quality within the Creek should improve as control measures are adopted, resulting in less contaminated water flowing into the adjacent wetlands. As the pressure from contaminated runoff diminishes, we would expect that the natural disinfection processes within the wetland would shift it into a predominantly “sink” mode. This situation would be similar to conditions Jeong et al. (2008) documented for the Talbert
4639
Marsh as urban runoff was increasingly rerouted from the wetland into the sanitary sewers via low-flow diversion structures. An important ecosystem service of wetlands is maintaining good water quality, usually by removal of organic matter by settling (Gopal, 1999) and nutrient by macrophytes (Zedler and Kercher, 2005). Increasing wetland biodiversity through restoration actions has been shown to increase ecosystem services (Benayas et al., 2009) like water quality. In southern California, more natural salt marsh wetlands, or at least those receiving limited urban runoff, have been shown to retain FIB more often than acting as a source. Examples include wetlands adjacent the Santa Ana River (Jeong et al., 2008), the Carpinteria salt marsh (Ambrose, unpublished data), and the Bolsa Chica Wetlands (Moore, 2007). Therefore, restoration of urban salt marsh systems will have many benefits as ecosystem services are regained, of which water quality ranks high.
5.
Conclusions
1. After two 24-h sampling events, densities of FIB varied up to three orders of magnitude over a tidal cycle. 2. These bacteria mainly are introduced into the wetland via contaminated water from the adjacent Ballona Creek. 3. During daytime, the total load of bacteria is significantly reduced, most likely by sunlight and other processes, resulting in a sink. 4. Sediment resuspension during stronger tidal flows may reintroduce FIB into the water column, or these bacteria could be enriched within the tidal channel water. 5. During stronger ebb flows, the wetlands tend to be a source of FIB entering the estuary, but because their concentrations of FIB are lower relative to those in the estuary, wetland flows can dilute the more contaminated receiving waters of the estuary. 6. By better understanding these processes, we may predict the potential outcome as more water is introduced into the wetlands through restoration projects. 7. We would anticipate that as pollution control measures are adopted within the watershed to reduce FIB contamination, the wetlands would shift primarily into a sink-mode rather than seesawing between sink- and source-mode, as they appear to do so now.
Acknowledgments We wish to sincerely thank the following students who enthusiastically participated in the field and laboratory studies: Loyola Marymount University e Sarika Doshi, Nick Fugate, Tim Tawakol, Anh Nguyen, Arthur Bilikian, Lindsay Vitort; UCLA e Liz Bernier and Natalie Diaz; Chadwick High School e Swati Yanamadala. This study would not have been possible without their help. Dr. Eric Stein and Ken Schiff, both from the Southern California Coastal Water Research Project, kindly reviewed the manuscript, providing many valuable
4640
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 3 0 e4 6 4 2
suggestions. We also wish to acknowledge the anonymous reviewers who also made excellent recommendations to improve the manuscript. Funding for this study was provided through several sources: a LMU Continuing Faculty Summer Grant, a grant from the Santa Monica Bay Restoration Commission, and in part from a University of California Marine Council grant to Stanley Grant (UCI) and Richard Ambrose (UCLA).
Appendix A. Summary of FIB densities (MPN/ 100 ml) collected during the 24-h surveys in the Ballona Wetlands (BW2) and Estuary (BE), 2007. For each collection time, n [ 3.
Total coliforms Date
E. coli
Enterococci
Station
Time
Mean
S.D.
Mean
S.D.
Mean
S.D.
12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 12-Jul-07 13-Jul-07 13-Jul-07 13-Jul-07 12-Jul-07
BW2 BW2 BC BW2 BW2 BW2 BC BW2 BW2 BW2 BC BW2 BW2 BW2 BC BW2 BW2 BW2 BC BW2 BW2 BW2 BC
4:45 AM 5:51 AM " 7:22 AM 8:58 AM 10:27 AM " 11:58 AM 1:28 PM 2:58 PM " 4:30 PM 6:05 PM 7:30 PM " 8:57 PM 10:31 PM 12:00 AM " 1:30 AM 3:58 AM 4:20 AM "
12230.7 >24192.0 24196.0 >24192.0 14119.7 8258.3 >24196.0 10953.3 1461.3 1028.3 1860.7 407.0 388.3 503.3 123.0 887.7 2354.7 7119.0 >24196.0 2326.7 5145.0 3378.0 >24196.0
4668.8 0.0 0.0 0.0 2820.8 1355.8 0.0 425.5 330.6 265.2 259.2 128.3 191.8 161.8 69.9 95.1 1039.6 8844.3 0.0 1491.8 3868.5 1050.4 0.0
1205.0 246.3 430.0 89.7 44.3 16.7 59.7 <10 <10 <10 10.0 13.3 16.7 13.3 10.0 16.7 10.0 55.7 466.0 20.0 <10 <10 127.0
1536.6 25.5 110.0 35.2 5.8 11.5 13.3 0.0 0.0 0.0 0.0 5.8 11.5 5.8 0.0 5.8 0.0 35.4 234.6 17.3 0.0 0.0 37.6
86.3 1074.3 1155.7 264.0 241.3 37.3 143.7 13.3 <10 <10 16.7 13.3 <10 <10 10.0 30.0 20.3 178.7 1721.3 34.0 16.7 20.3 392.0
69.2 186.2 51.6 73.4 92.3 15.8 54.4 5.8 0.0 0.0 11.5 5.8 0.0 0.0 0.0 10.0 17.9 48.5 1025.8 6.1 5.8 17.9 112.6
2-Aug-07 2-Aug-07 2-Aug-07 2-Aug-07 2-Aug-07 2-Aug-07 2-Aug-07 2-Aug-07 2-Aug-07 2-Aug-07 2-Aug-07 2-Aug-07 2-Aug-07 2-Aug-07 2-Aug-07 2-Aug-07 3-Aug-07 3-Aug-07 2-Aug-07 3-Aug-07 3-Aug-07 3-Aug-07 2-Aug-07
BW2 BW2 BC BW2 BW2 BW2 BC BW2 BW2 BW2 BC BW2 BW2 BW2 BC BW2 BW2 BW2 BC BW2 BW2 BW2 BC
6:00 AM 7:30 AM " 9:00 AM 10:30 AM 12:00 PM " 1:30 PM 3:00 PM 4:30 PM " 6:00 PM 7:30 PM 9:00 PM " 10:30 PM 12:00 AM 1:30 AM " 3:00 AM 4:30 AM 6:00 AM "
827.7 >24192.0 >24192.0 >24192.0 >24192.0 >24192.0 >24192.0 >24192.0 5252.7 10535.3 17329.0 12777.3 14820.7 21307.3 14686.3 12178.0 5944.3 7044.3 34903.3 12777.3 14686.0 18419.0 >24192.0
302.2 0.0 0.0 0.0 0.0 0.0 0.0 0.0 685.6 1149.5 0.0 4003.1 2245.7 2501.7 1463.0 1695.7 857.4 1881.4 44861.1 1480.8 5647.1 2501.1 0.0
<10 334.7 1958.3 523.7 358.0 243.0 119.0 125.3 41.3 42.3 52.7 27.3 110.0 205.0 83.3 132.0 27.0 52.0 122.0 92.0 3727.0 115.7 2141.7
0.0 385.7 75.8 140.9 24.6 79.5 43.6 28.3 18.5 26.8 29.3 6.4 70.0 103.9 11.5 52.6 6.1 21.5 48.0 44.2 6266.6 17.6 531.9
59.0 7094.0 1958.3 783.3 6045.3 2161.7 119.0 15.0 13.3 <10 52.7 16.7 106.0 94.0 83.3 34.0 26.7 13.3 122.0 <10 184.0 237.0 2141.7
6.1 2772.0 75.8 183.9 729.2 1293.0 43.6 14.4 5.8 0.0 29.3 11.5 33.5 28.0 11.5 12.1 11.5 5.8 48.0 0.0 38.5 11.5 531.9
> ¼ Greater than method detection limits. < ¼ Less than method detection limits.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 3 0 e4 6 4 2
references
Alderisio, K.A., DeLuca, N., 1999. Seasonal enumeration of fecal coliform bacteria from the feces of ring-billed gulls (Larus delawarensis) and Canada Geese (Branta canadensis). Applied and Environmental Microbiology 65 (12), 5628e5630. American Public Health Association, American Water Works Association, Water Environment Federation, 1998. Standard Methods for the Examination of Water and Wastewater, 20th ed. American Public Health Association, 1015 Fifteenth St., Washington, D.C. Benayas, J.M., Newton, A., Diaz, A., Bullock, J.M., 2009. Enhancement of biodiversity and ecosystem services by ecological restoration: a meta-analysis. Science 325 (28 August 2009), 1121e1124. Boehm, A.B., Grant, S.B., Kim, J.H., Mowbray, S.L., McGee, C.D., Clark, C.D., Foley, D.M., Wellman, D.E., 2002. Decadal and shorter period variability of surf zone water quality at Huntington Beach, California. Environmental Science and Technology 36 (18), 3885e3892. Chamberlin, C.E., Mitchell, R., 1978. A decay model for enteric bacteria in natural waters. In: Mitchell, R. (Ed.), Water Pollution Microbiology, vol. 2. John Wiley and Sons, New York, pp. 325e348 (Chapter 12). City of Los Angeles and Keane Biological Consulting, 2005. Ballona Wetlands 1135 restoration project biological study, Los Angeles County, California. Final Report. Prepared for the U.S. Army Corps of Engineers, Los Angeles District, Los Angeles, CA, 56 pp. Dorsey, J.H., 2006. Densities of fecal indicator bacteria in tidal waters of the Ballona Wetlands, Los Angeles, County, California. Bulletin Southern California Academy of Sciences 105 (2), 59e75. Dorsey, J.H., 2009. Improving water quality through California’s Clean Beach Initiative: an assessment of 17 projects. Environmental Monitoring and Assessment. doi:10.1007/ s10661-009-0987-5. Dorsey, J.H., Lindaman, E., 2004. Water quality. Report submitted to the California State Legislature. In: McNeill, D., Landry, J. (Eds.), Ballona Creek Trail and Bikeway: Environmental Monitoring & Assessment. Baldwin Hills Conservancy, Los Angeles, CA, pp. 62e81. Dorsey, J.D., Raymond, M., Leary, C., 2008. A study of the fecal indicator bacteria (FIB) in Del Rey Lagoon, Playa Del Rey, California. Report submitted to City of Los Angeles, Environmental Monitoring Division, 14 pp. Evanson, M., Ambrose, R., 2006. Sources and growth dynamics of fecal indicator bacteria in a coastal wetland system and potential impacts to adjacent waters. Water Research 40 (3), 475e486. Ferguson, D.M., Moore, D.F., Getrich, M.A., Zhowandai, M.H., 2005. Enumeration and speciation of enterococci found in marine and intertidal sediments and coastal water in southern California. Journal of Applied Microbiology 99 (3), 598e608. Given, S., Pendleton, L.H., Boehm, A.B., 2006. Regional public health cost estimates of contaminated coastal waters: a case study of gastroenteritis at southern California beaches. Environmental Science and Technology 40 (16), 4851e4858. Gopal, B., 1999. Natural and constructed wetlands for wastewater treatment: potentials and problems. Water Science and Technology 40 (3), 27e35. Grant, S.B., Sanders, B.F., Boehm, A.B., Redman, J.A., Kim, J.H. , Mrse, R.D., Chu, A.K., Gouldin, M., McGee, C.D., Gardiner, N.A., Jones, B.H., Svejkovsky, J., Leipzig, G.V., Brown, A., 2001. Generation of enterococci bacteria in a coastal saltwater marsh and its impact on surf zone water quality. Environmental Science and Technology 35 (12), 2407e2416.
4641
Haile, R.W., Witte, J.S., Gold, M., Cressey, R., McGee, C.D., Millikan, R.C., Glasser, A., Harawa, N., Ervin, C., Harmon, P., Harper, J., Dermand, J., Alamillo, J., Barrett, K., Nides, M., Wang, G.Y., 1999. The health effect of ocean water contaminated by storm drain runoff. Epidemiology 10 (4), 355e363. Jeong, Y., Sanders, B.F., McLaughlin, K., Grant, S.B., 2008. Treatment of dry weather urban runoff in tidal saltwater marshes: a longitudinal study of the Talbert Marsh in southern California. Environmental Science and Technology 42 (10), 3609e3614. Jillson, S.J., Dahab, M.F., Woldt, W.E., Surampalli, R.Y., 2001. Pathogen and pathogen indicator removal characteristics in treatment wetlands systems. Practice Periodical of Hazardous, Toxic, and Radioactive Waste Management 5 (3), 153e160. Karathanasis, A.D., Potter, C.L., Coyne, M.S., 2003. Vegetation effects on fecal bacteria, BOD, and suspended solid removal in constructed wetlands treating domestic wastewater. Ecological Engineering 20 (3), 157e169. Mayo, A.W., 2004. Kinetics of bacterial mortality in granular bed wetlands. Physics and Chemistry of the Earth, Parts A/B/C 29 (15e18), 1259e1264. Moore, D., 2007. The effects of the Bolsa Chica channel opening. California Coastal Coalition, Headwaters to Ocean Conference (H2O 2007). Presentation available at http://www.websurfer. us/coastal/h20_2008/2007_presentations.php. Noble, R.T., Lee, I.M., Schiff, K.C., 2004. Inactivation of indicator micro-organisms from various sources of faecal contamination in seawater and freshwater. Journal of Applied Microbiology 96 (3), 464e472. Phillip Williams & Associates, Ltd., 2006. Ballona Wetlands Existing Conditions Report. Report submitted to California Coastal Conservancy. Ricca, D.M., Cooney, J.J., 1998. Coliphages and indicator bacteria in birds around Boston Harbor. Journal of Industrial Microbiology and Biotechnology 21 (1e2), 28e30. Rifai, H., 2006. Study on the Effectiveness of BMPs to Control Bacteria Loads. Final Quarterly Report No. 2. Prepared for Texas Commission on Environmental Quality, P.O. Box 13087, MC-150, Austin, Texas 78711-3087. Saez, J., 2007. Characterization of soils and subsurface hydrology. Report submitted to the California Coastal Conservancy and the Santa Monica Bay Restoration Commission. In: Dorsey, J., Bergquest, S. (Eds.), A Baseline Survey of the Ballona Outdoor Learning & Discovery (BOLD) Area. Ballona Wetlands, Los Angeles County, California. Ballona Wetlands Foundation, pp. 6e18 (Chapter 2) Grant Agreement No. 04-118. Shiaris, M.P., Rex, A.C., Pettibone, G.W., Keay, K., McManus, P., Rex, M.A., Ebersole, J., Gallagher, E., 1987. Distribution of indicator bacteria and Vibrio parahaemolyticus in sewagepolluted intertidal sediments. Applied and Environmental Microbiology 53 (8), 1756e1761. Shellenbarger, G.G., Athearn, N.D., Takekawa, J.Y., Boehm, A.B., 2008. Fecal indicator bacteria and Salmonella in ponds managed as bird habitat, San Francisco Bay, California, USA. Water Research 42 (12), 2921e2930. Steets, B.M., Holden, P.A., 2003. A mechanistic model of runoffassociated fecal coliform fate and transport through a coastal lagoon. Water Research 37 (3), 589e608. Stein, E.D., Tiefenthaler, L.L., 2004. Characterization and source identification of dry-weather metals and bacteria in Ballona Creek. In: Weisberg, S., Elmore, D. (Eds.), Southern California Coastal Water Research Project, Biennial Report 2003e2004. Stinton, L.W., Hall, C.H., Lynch, P.A., Davies-Colley, R.J., 2002. Sunlight inactivation of fecal indicator bacteria and bacteriophages from waste stabilization pond effluent in fresh and saline waters. Applied and Environmental Microbiology 68 (3), 1122e1131.
4642
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 6 3 0 e4 6 4 2
Surbeck, C.Q., Jiang, S.C., Grant, S.B., 2010. Ecological control of fecal indicator bacteria in an urban stream. Environmental Science and Technology 44 (2), 631e637. Tiefenthaler, L.L., Stein, E.D., Lyon, G.S., 2009. Fecal indicator bacteria (FIB) levels during dry weather from southern California reference streams. Environmental Monitoring and Assessment 155 (1e4), 477e492. Vymazal, J., 2005. Removal of enteric bacteria in constructed treatment wetlands with emergent macrophytes: a review. Journal of Environmental Science and Health, Part A: Toxic/
Hazardous Substances and Environmental Engineering 40 (6e7), 1355e1367. West, J., 2001. Ballona wetland. In: Zedler, J.B. (Ed.), Handbook for Restoring Tidal Wetlands. CRC Press, Boca Raton, FL, pp. 10e20. Whitman, R.L., Przybyla-Kelly, K., Shively, D.A., Nevers, M.B., Byappanahalli, M.N., 2008. Sunlight, season, snowmelt, storm, and source affect E. coli populations in an artificially ponded stream. Science of the Total Environment 390 (2e3), 448e455. Zedler, J.B., Kercher, S., 2005. Annual Review of Environment and Resources 30, 39e74.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 2 5 e4 4 3 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
River phosphorus cycling: Separating biotic and abiotic uptake during short-term changes in sewage effluent loading M.I. Stutter*,1, B.O.L. Demars 1, S.J. Langan The Macaulay Institute, Craigiebuckler, Aberdeen AB15 8QH, UK
article info
abstract
Article history:
Medium to small scale point sources continue to threaten river ecosystems through P
Received 4 December 2009
loadings. The capacity and timescales of within-river processing and P retention are
Received in revised form
a major factor in how rivers respond to, and protect downstream ecosystems from,
31 May 2010
elevated concentrations of soluble reactive P (SRP). In this study, the bio-geochemical
Accepted 4 June 2010
response of a small river (w40 km2 catchment area) was determined before, during and
Available online 12 June 2010
after exposure to a fourteen day pulse of treated sewage effluent using an upstream reach as a control. A wide array of approaches (batch and column simulations to in-situ whole
Keywords:
stream metabolism) allowed independent comparison and quantification, of the relative
Nutrient spiralling
contribution of abiotic and biotic processes in-river P cycling. This enabled, for the first
Whole stream metabolism
time, separating the relative contributions of algae, bacteria and abiotic sorption without
Stoichiometry
the use of labelled P (radioisotope). An SRP mass balance showed that the ecosystem
Algal assays
switched from a P sink (during effluent inputs) to a P source (when effluent flow ceased).
Kinetic sorption studies
However, 65e70% of SRP was retained during the exposure time and remained sequestered
Open flow-through columns
two-weeks after-effluent flow ceased. Batch studies treated with biocide gave unrealistic results, but P uptake rates derived by other methods were highly comparable. Downstream of the effluent input, net P uptake by algae, bacteria and sediment (including the biofilm polysaccharide matrix) were 0.2 (0.1), 0.4 (0.3), and 1.0 (0.9) mmol m2 day1 during effluent exposure. While autotrophic production did not respond to the effluent exposure, heterotrophic production increased by 67% relative to the control and this translated into a 50% increase in biological P uptake rate. Therefore, both biological and abiotic components of stream ecosystems uptake P during exposure to treated sewage effluent P inputs, and maintain a long ‘memory’ of this input in terms of P storage for considerable timescales after loading. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Phosphorus is the key limiting nutrient controlling biological processes in many river systems (Slavik et al., 2004) and is the focus of research and legislation aiming to reduce loads to these systems (e.g. European Waste Water Treatment
Directive, CEC 1991). Despite past efforts in addressing point sources and a current shift towards tackling agricultural diffuse pollution, there remains a complex interplay of P sources to rivers (Withers and Jarvie, 2008). Point sources should not be considered a past problem since much of what is considered ‘diffuse’ sources concerns an abundance of
* Corresponding author. Tel.: þ44 1224498200; fax: þ44 1224395037. E-mail address:
[email protected] (M.I. Stutter). 1 Both these authors contributed equally to this work. 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.014
4426
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 2 5 e4 4 3 6
smaller point sources in rural landscapes (farm yard drains, septic tank outflows, etc.). Hence, understanding the behaviour of point sources and their impact on river ecosystems remains an important goal, particularly under summer low flow conditions when biota are most sensitive (Rodda, 2007; Van Vliet and Zwolsman, 2008). For example, while P concentration in the water column decreases rapidly following P removal at point sources, river bed sediment P dynamic may not respond as fast (Demars et al., 2005; Demars and Harper, 2005; Jarvie et al., 2006). An improved understanding of biotic and geochemical processes of river P retention would enable better prediction of current and future buffering capacity as pollutant, land management and climatic pressures evolve. A river’s capacity for internal nutrient processing provides an important function of buffering downstream ecosystems by removing (in the case of N; Mulholland et al., 2008), or delaying (in the case of P; Dorioz et al., 1989) increases in nutrient concentrations due to upstream pollution. Shortterm classic radioisotope studies with 32P have revealed crucial insights in our understanding of P cycling in streams, particularly the biological uptake and translocation of P into the food web (Ball and Hooper, 1963; Newbold et al., 1983). Traditional P cycling methods (not relying on 32P), using: a) laboratory closed systems such as batch bottles, chambers or re-circulating flumes (House et al., 1995; Jarvie et al., 2005; Gainswin et al., 2006); b) outdoor open channels and in-situ whole river studies (House et al., 1995; Gu¨cker and Pusch, 2006; Demars, 2008), have not yet resolved the relative proportions of biotic and abiotic uptake. Only a few studies have tried to partition the relative role of the biota, but these relied on the use of biocide in batch bottles or flow-through channels and did not distinguish the role of autotrophs from heterotrophs (e.g. D’Angelo et al., 1991; Haggard et al., 1999; McDowell, 2003; Lottig and Stanley, 2007). Therefore, more amenable methods still need to be developed (Webster et al., 2009; Small et al., 2009) to quantify the relative role of the biota (autotrophs and heterotrophs) in P cycling rates. There are also large discrepancies between laboratory and in-situ whole stream P cycling rates (Demars, 2008) that need to be further investigated. In this work we therefore sought to (i) compare several insitu whole stream and laboratory P exchange capacity methods between river bed sediment and water column; (ii) separate the relative role of algae, bacteria and abiotic P uptake-release dynamics using independent approaches. Our field manipulation experiment exposed a rural river reach (37 km2 mixed-agriculture catchment) to a 14 day pulse of treated sewage effluent. This novel approach enabled study of the river bed retention of P during exposure to elevated P, and fate of the transient storage of P post treatment.
2.
Material and methods
To structure the work and experimental work done we first describe, the water and sediment of the field site and its characteristics to provide essential and detailed experimental context. Then, we present two methods to estimate total whole river P cycling (Section 2.2.). This is followed by two
batch laboratory methods (i.e. within a closed system in a bottle) to study abiotic P exchange capacity of river bed sediment with water column, with a method allowing comparison with in-situ whole stream measurements (Section 2.3.). Finally, we undertook three independent methods to quantify P algal and bacterial uptake (Section 2.4.). The combination of the best results was used to partition the algal, bacterial and abiotic (total-biotic) P uptake rate (Section 3.5.). The design of the study is further illustrated in Fig. 1 and summarised in Table 1.
2.1.
Experimental site, water and sediment sampling
In 2004 the Tarland village (w600 people) Waste Water Treatment Works (WWTP) was upgraded as part of an ongoing catchment-wide programme to improve water quality. Treated effluent was then discharged to a self-contained wetland, but the operator included a switch valve to discharge treated effluent directly to the stream (under licensed consent). The WWTP is a combined sewer overflow and so intermittent discharges to the stream occur during high rainfall events when the storage tank facility is exceeded. Short (few minutes) intermittent overflows were also observed under dry weather when WWTP effluent discharge was around 4 L s1 due to an overflow at the switch valve. Our experiment diverted the treated effluent directly to the stream for 14 days in Aug 2006, after which it went back to the wetland (Fig. 1). The experiment had three periods: (A) prior to-, (B) during- and (C) after-effluent discharge. Stream water samples were collected at three to six hourly intervals upstream and two downstream reaches from the WWTP effluent pipe using auto-samplers (Fig. 1). River stage was continually monitored at the downstream end of reach 2 in a stream section rated by depth velocity profiling. Discharge of
Fig. 1 e Experimental design (see Table 1). Three sections (U, D1, D2) of Tarland stream were studied before (A), during (B) and after (C) WWTP effluent diversion from a wetland to the stream. Whole stream mass balance measurements are reported from these sections. Note however that the measurements based on whole stream metabolism and nutrient spiralling (U and D1) excluded the mixing zone necessary for tracer studies (NaCl and propane). Discharge (Q) was measured continuously at the downstream end. The position of auto-samplers is indicated by a cylinder symbol.
4427
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 2 5 e4 4 3 6
Table 1 e Experimental design. The present study sought to compare whole river in-situ P dynamics with laboratory essays (Sections 2.2. and 2.3.), and to disentangle algal, bacterial and abiotic P uptake (Section 2.4.). The measurements were taken upstream and downstream a WWTP (see Fig. 1), in three consecutive periods, before (A), during (B) and after (C) effluent diversion. Measurements
Upstream
Downstream 1
Downstream 2
Period
2.2. Whole river P cycling 2.2.1. Mass balancea (MB ¼ SCQ) 2.2.2. Whole stream SRP addition studies (nutrient spiralling)
n/a U
U U
U n/a
A, B, C A, B
2.3. Laboratory sediment P sorption capacity 2.3.1. Batchb bottle P exchange capacity with biocide (EPC0) 2.3.2. Kinetic bottle P sorption capacity with biocide
U n/a
U U
n/a n/a
A, B, C Simulated B
2.4. Biological P uptake rate 2.4.1. Laboratory column sorption experiment (no biocide) 2.4.2. Biofilm accrual on bricks 2.4.3. Whole stream metabolismc (WSM) Stoichiometric approach based on WSM and nutrient spiralling
n/a U U n/a
U U U U
n/a n/a n/a n/a
Simulated A, B, C A, B, C A, C B (C)
n/a, not applicable. a C ¼ concentration, Q ¼ discharge. b Batch of eight different water SRP concentrations. c Respiration and photosynthesis.
reach 1 was catchment-scaled from reach 2 measurements, which agreed within 5% with depth velocity profile and salt dilution gauging. The 37 km2 catchment is subject to additional agricultural pressures with land areas of arable and intensive grassland of 23% and 40%, respectively (see Stutter et al., 2008). The upper 2 cm of sediment from the river bed was sampled in the upstream reach and downstream reach 1 using a flat plastic scoop. At the upstream and downstream locations ten spot samples across 10 m2 area of river bed were combined at each site. Sediments were passed through a 2mm aperture sieve and air-dried (30 C). The decision to dry sediments was taken on the basis that different storage times for sediments sampled over time prior to determining isotherms in a single batch would introduce as much errors as the drying process itself (cf McDowell, 2003). The sediments are characterised as having low organic C contents <2 g kg1 and oxalate extractable Fe and Al contents of 9 and 3 g kg1, respectively (Stutter et al., 2008). Sediments typically had size class contributions clay, silt, fine sand, medium to coarse sand of 1, 10, 28, 61%, respectively. Further analytical details are given in Supporting Information.
2.2.
Whole river P cycling
Two approaches were used to determine in-situ SRP cycling rates: (i) a mass balance approach using auto-samplers at the top and bottom of downstream reaches and (ii) detailed SRP addition studies (Fig. 1, Table 1).
2.2.1.
SRP mass balance
The mass balance (MB; mmol s1) of the downstream river reaches (subscripts 1 and 2) was undertaken according to: MB1 ¼ CD1 QD1 ðCU QU þ CE QE Þ;
(1) 1
where C and Q equal the SRP concentration (mmol L ) and discharge (L s1), respectively, and subscripts D1, U and E
denote downstream reach 1, upstream and effluent, respectively. For the downstream reach 2 (D2) this was expanded to include the inputs from two tributaries (T1 and T2): MB2 ¼ CD2 QD2 ðCU QU þ CE QE þ CT1 QT1 þ CT2 QT2 Þ:
(2)
The combined tributary inputs averaged 17% of the P flux at D2. Mass balances were scaled to mmol P m2 day1 on the basis of the average measured river width of 2.9 m.
2.2.2.
Nutrient addition approach
Whole river phosphate cycling studies following Webster and Valett (2006) were performed before and after the start of the effluent diversion upstream and downstream of the WWTP (2, 4, 6, 24 and 72 h). This was undertaken on the same river reaches used for whole stream metabolism measurements. These studies allowed the measurement of background and subsequent changes in: nutrient uptake length (average distance travelled by a molecule of phosphate before uptake by the river bed); phosphate uptake rate; and phosphate uptake velocity (a measure of uptake rate, relative to availability in the water, which normalises for stream velocity and depth). The nutrient spiralling studies were undertaken during ‘period A’ and described the background uptakerelease dynamic equilibrium. The studies done at 24 and 72 h downstream of the WWTP described the net uptake rate (Haggard et al., 2001), directly comparable to the mass balance approach. The uptake velocity is comparable across streams and sensitive to anthropogenic pressures (e.g. Newbold et al., 2006).
2.3.
Laboratory sediment abiotic P sorption capacity
Two batch laboratory approaches were undertaken in order to determine the SRP exchange capacity of sediment collected in the different periods (A, B, C) of the field manipulation (Fig. 1, Table 1) and a method is presented to allow comparison with whole stream total P exchange dynamics (Section 2.2.).
4428 2.3.1.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 2 5 e4 4 3 6
2.4.
Batch study, SRP exchange capacity
These experiments were performed using 1 g sediment (oven dried equivalent) to 30 ml of 0, 10, 20 and 50, 100, 500, 1000, 5000 mmol P L1 soluble reactive phosphorus (SRP as KH2PO4). This was made up with an ‘artificial river water’ matrix of 0.5 mmol L1 CaCl2, with 0.02% azide as a biocide. After 18 h equilibration at 20 C, solutions were filtered prior to SRP analyses. Triplicate equilibrations had coefficients of variation of final SRP concentrations within 10%. Example isotherms are shown in Supporting Information (Fig. S1). A Langmuir model was fitted describing DNa, the measured P uptake during equilibration (mmol kg1) and C, the remaining SRP concentration after equilibration (mmol L1): DNa ¼ Nmax
KC Ni 1 ¼ KC
(3)
where K (L kg1) is a constant, Ni the initial ‘native’ adsorbed P (mmol kg1) and Nmax the sorption maximum (mmol kg1). The positive intercept on the x axis (the SRP concentration which gives no change in DNa over the period of equilibration) represents the equilibrium phosphorus concentration (EPC0; Taylor and Kunishi, 1971). Note that only the four lowest batch concentrations were used to determine the EPC0 (as shown in Fig S1). Readily desorbable ‘native’ P contents of the sediments were determined using the Fe oxide paper strip test according to Chardon et al. (1996).
2.3.2.
Batch study, kinetics of SRP sorption
These experiments were performed (same conditions as in Section 2.3.1) at 3.8 mmol SRP L1 (equivalent to river concentrations during effluent discharge) and time points 0, 0.08, 0.25, 0.5, 1, 3, 5, 24 h equilibration. Sorption was converted from mmol kg1 to mmol m2 of river reach by multiplying by the river bed area, specific density of fine sediment <2 mm (1600 kg m3) and assuming that the first 2 cm depth (as per sampling method) of the river bed sediment layer interacted with overlying waters and a homogeneity of sediment across the river bed. While this depth is rather pragmatic it serves to act as a linear scaling factor in the P uptake modelling and corresponded to previous studies (Fischer and Pusch, 2001). Phosphate kinetic data was interpreted with the Elovich equation (e.g. Demars, 2008): dUNET ¼ aexpðbUNET Þ dt
(4)
where UNET is the net uptake (mmol m2), t time (day), a is the initial uptake rate (mmol m2 day1), and b is the rate constant (m2 mmol1). With the condition that U ¼ 0 at t ¼ 0, this integrates (Allen and Scaife, 1966): 1 UNET ¼ lnð1 þ abtÞ b
(5)
The nutrient uptake length SW (m), for comparative purpose with whole stream uptake studies, was calculated as follows (Demars, 2008): SW ¼
u ab
with u, average water velocity (m day1).
(6)
Biological P uptake rate
Three independent methods were used to provide estimates of biological uptake from the autotrophs (mostly algae) and heterotrophs (mostly bacteria) at two scales of observation: laboratory or field bioassays and in-situ whole river reach (Fig. 1, Table 1).
2.4.1.
Laboratory column experiments
Column sorption experiments were undertaken without biocide to: (i) evaluate any hysteresis in desorption relative to adsorption and (ii) determine any microbial P uptake. Column parameters are described in Supplementary Information Table S1. Briefly, columns (47 g sediment, sieved <2 mm) were equilibrated at 20 C (flow rate 0.5 mL min1), until eluant outflow concentrations matched inflows, or became stable following saturation of abiotic P sorption. Columns were equilibrated sequentially to (i) 0.7 mmol L1 SRP, then (ii) elevated SRP (3.8 mmol L1), then (iii) 0.7 mmol L1 SRP again, all using a background electrolyte of 0.5 mmol L1 CaCl2. Columns were packed with previously air-dried sediments, but were equilibrated for a fortnight (approximately 1000 pore volumes during phase (i), cf McDowell, 2003). One day of flow equated to 60e75 column pore volumes.
2.4.2.
Algal growth rates
Accumulation of benthic algae was determined on artificial clay brick substrates (area 0.04 m2) placed on the river bed in triplicate at the beginning, and recovered at the end of each experimental period (A, B, C). Bricks were scraped, washings were filtered (Whatman GF/F) and chlorophyll determined by hot extraction of the filter in 3 mL methanol. Chlorophyll a concentrations were determined by absorbance at 665 nm (corrected for turbidity at 750 nm) (Talling and Driver, 1963). Algal biomass P accumulation rates (mmol P m2 day1) were calculated from molar ratios chlorophyll a:C of 1:35 (Sobczak et al., 2002) and C:P of 158:1 (Kahlert, 1998).
2.4.3.
River reach scale studies
Biotic P uptake was based on C:P stoichiometry and gross primary productivity (GPP) and ecosystem respiration (ER) similar to Hall and Tank (2003). Autotrophic production was estimated as 0.5 GPP and heterotrophic respiration (HR) as ER e 0.5 GPP. We used the following stoichiometric ratios: algal molar C:P of 158:1 (range 99:1e369:1, Kahlert, 1998), and bacterial molar C:P of 65:1 and C:P of 130:1 for the moderate (0.2) and low (0.05) heterotrophic growth efficiency (HGE) scenarios respectively (Thingstad et al., 1996; Gundersen et al., 2002; Ukpong, 2005). Heterotrophic production (HP) was calculated from: HGE ¼
HP ðHP þ HRÞ
(7)
using Solver in Excel. The average heterotrophic production (from low and moderate HGE) was used to calculate the totalbiotic uptake and partitioning the relative contribution of algae, bacteria and sediment P uptake. The relative change in
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 2 5 e4 4 3 6
ER (and similarly for total biological uptake) due to the WWTP effluent effect was calculated as follows: DERWWTP ¼
DERD DERU ERDA
(8)
with D representing change between period A (before) and C (after) effluent diversion, and subscripts D, U and A denote downstream, upstream and period A, respectively. Whole stream metabolism (GPP, ER) was calculated at 1 min time steps upstream/downstream the WWTP effluent before, during and after-effluent diversion using the diel oxygen two station approach of Marzolf et al. (1994), with corrections from Young and Huryn (1998). Tracer studies were run with NaCl to determine mean travel time and lateral inflows (here insignificant, <2%) and with propane using diffusion of micro-bubbles across the whole channel to determine the re-aeration coefficient. Due to the trapezoidal shape of the stream channels and relatively fast flowing water, the mixing zones were relatively long (50e70 m), with additional sand bags in the stream to ensure full tracer mixing. Mean travel times were 10 and 17 min for the upstream (140 m) and downstream (177 m) reach 1, respectively. The reaeration coefficients were 53 4 and 26 2 day1 in the upstream and downstream reaches, respectively.
2.5.
Data analyses
The design of the experiment was a classic BACI (before and after control experiment) and statistical analyses followed Demars and Harper (2005). Two-way ANOVAs were run (interaction site period) to test the effect of the WWTP effluent diversion on ER, GPP, and P uptake. However, the resulting probabilities should only be considered as rough indicators because GPP, ER and P uptake were all highly autocorrelated within the two periods of observations. Strictly, no probabilities should be reported until other similar experiments are reported elsewhere, allowing a meta-analysis. Increase in P uptake, relative to the upstream control, took into account the changes in the upstream control reach as well as the changes in the downstream reach. One-way ANOVAs allowed comparison between periods or sites. Uncertainties based on 1 Standard Deviation (SD) were propagated throughout the calculations as in Demars et al. (2005).
3.
Results and discussion
3.1.
Changes in-river chemistry
Means of river hydrochemistry for the three experimental periods are given in Table S2. There was some variation in concentrations upstream of the WWTP (control site) so that a one-way ANOVA showed significant ( p 0.05) differences between treatment periods A, B and C for concentrations of NO3, protons, soluble reactive, unreactive and particulate P (SRP, SUP and Part P, respectively). This was related to changes in discharge between periods (Fig. 2a) and upstream inputs. However, effluent discharge to the river increased concentrations of NH4, SRP and SUP and, to a lesser extent protons
4429
and DOC, downstream of the WWTP. River SRP was elevated ten-fold by the effluent discharge (Fig. 2b). However, there was no downstream increase in sediment, or particulate P (Part P) related to the effluent discharge. A period of stable summer baseflow preceded our experiment. However, there were several minor storm events when river discharge increased (Fig. 2a), most notably just prior to the commencement of effluent discharge (end of period A), although with no substantial river bed movement.
3.2.
Whole river P cycling
Prior to effluent discharge (period A) both upstream and downstream reaches had similar SRP cycling rates (Table 2). The nutrient spiralling studies showed that there was little exchange between the river bed and overlying water suggesting that the river bed was relatively saturated despite low background SRP concentrations. This saturation may be due to SRP pulses from upstream or from the WWTP overflow (especially during high flow and fast transit of water in the river channel (D’Angelo et al., 1991). The uptake velocities (8e41 103 mm s1) were comparable to other rivers impacted by point sources, e.g. River Erpe (4.2e69 103 mm s1, Gu¨cker and Pusch, 2006), Spavinaw Creek (2.3e7.5 103 mm s1, Haggard et al., 2001). However these uptake velocities are 10e100 times slower than those reported in agricultural (Bernot et al., 2006) or pristine streams (Doyle et al., 2003). Hence the Tarland Burn had a small P retention efficiency. The mass balance approach was adopted to calculate ecosystem (sediment and biological) uptake of SRP. The negative values for period B (Fig. 2c) in both reach 1 and reach 2 indicate SRP mass not accounted for at the downstream site (i.e. uptake of SRP in the reach). Conversely, positive values indicate net SRP release. The positive values during period A were likely to have been related to the storm event. The SRP mass balance is highly sensitive to accurate quantification of the WWTP effluent P inputs and the positive spike around 19th Aug may have occurred due to a period of unaccounted effluent P mass input when the WWTP storage tank capacity was exceeded during the storm. The SRP load (with relative uncertainties based on 1 SD; n ¼ 24 samples) associated with the effluent discharge to the river was variable at 448 mmol s1 (37%). This was derived from effluent flow rates and SRP concentrations measured over several days as 3.4 L s1 (13%) and 131 mmol L1 (35%). Although this contributes to some uncertainty in the mass balance the general patterns and magnitudes of fluxes hold. River SRP concentrations increased as soon as effluent flow commenced. Variation in-river water SRP concentrations downstream of the WWTP (10e86%; Table S2) contributed to variability in the SRP mass balance for the 229 m long reach 1 during period B. Maximum SRP uptake rates of 25e29 mmol m2 day1 occurred during several days early in period B (Fig. 2c). Subsequently, on the 24th, 27th and 28th Aug there was a net SRP release indicated. Further variability in the magnitude of the SRP mass balance for reach 1 may relate to inconsistency in effluent quantity and quality for a number of reasons associated with the WWTP operating conditions.
4430
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 2 5 e4 4 3 6
Fig. 2 e Temporal changes during the experimental periods A (prior to-), B (during-) and C (after-effluent discharge) in terms of (a) river flow, (b) SRP concentrations, and (c) SRP mass balance. Symbols are (7) U, upstream and (,) D1, downstream reach 1, D2, downstream reach 2 (-). Daily mean SRP concentrations (b) include 1 standard error bars for reach 1.
Table 2 e Background SRP concentration (CAMB), added phosphate concentration (CADD), Phosphate uptake length (SW), uptake rate (U ), and uptake velocity (vf), upstream and downstream the waste water treatment plant (WWTP) effluent before and during treatment. Negative values represent nutrient release from the stream bed towards the water column. Relative uncertainties (based on 1 SD) are reported within brackets.
Before diversion Upstream WWTP Downstream WWTP During diversion Downstream WWTP t¼2h t¼4h t¼6h t ¼ 24 h t ¼ 72 h
CAMB mmol L1
CADD mmol L1
0.472 0.486
0.512 0.839
4440 (2700 to 11 600) 1520 (930e2110)
0.666 0.747 0.778 0.821 0.682
4.054 4.632 6.518 7.417 4.947
1770 (1140e2400) 856 (703e1009) 1218 (979e1460) 3560 (2200e4920) 2300 (1470 to 3140)
SW (m)
U (mmol m2 day1)
vf (mm s1)
0.35 (0.22 to 0.90) 0.97 (0.58e1.35)
0.008 (0.005 to 0.022) 0.023 (0.014e0.032)
1.13 (0.71e1.55) 2.65 (2.13e3.16) 1.94 (1.52e2.35) 0.71 (0.42e0.97) 0.90 (0.55 to 1.23)
0.020 0.041 0.029 0.010 0.015
(0.013e0.027) (0.034e0.048) (0.023e0.035) (0.006e0.014) (0.010 to 0.021)
4431
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 2 5 e4 4 3 6
3.3.
Abiotic P sorption capacity of sediments
in-situ in the channel prior to P exchange capacity experiments) increased for sediments that had been exposed to the greater SRP concentration during the period of effluent discharge to the river. Sediment EPC0 became greater during period B (Fig. 3) downstream of the WWTP than EPC0 values for the control (upstream) site. However, the downstream increase was not large in relation to the variability over time at the control site (Table S3). Sediment EPC0 values are often compared to river SRP concentrations as an evaluation of the direction and strength of SRP adsorption/desorption (for example Jarvie et al., 2005). Sediments in the upstream control reach remained at approximate equilibrium or showed weak uptake (Fig. 3, close to the 1:1 line). For the downstream reach there was a lag prior to rising sediment EPC0 after water column SRP had already been elevated for 3 days. Subsequently, downstream sediment EPC0 increased at the end of the effluent discharge (period B), then decreased as effluent flow ceased. However, a small rise in EPC0 > SRP at the end of period C, which indicated potential for SRP desorption, was corroborated by rising FeO-test P (Table S3). This may be an artefact caused by sampling of the heterogeneous bed sediments. Kinetic batch equilibrations (using SRP concentrations matching that in the river during effluent discharge) did not show signs of P sorption saturation, i.e. the uptake rate was simply a function of the SRP concentration supplied (Fig S2a), although this was not linear (Fig S2b). There was a large initial P uptake rate with little changes in SRP concentration, and this is reflected in the P uptake velocity (sorption efficiency relative to SRP concentration, Fig S2c). Hence, the uptake rate was fairly independent of change in SRP concentrations within only the first half an hour of the study. The initial uptake rate was 5000 times greater than the averaged sorption measured by the whole river mass balance approach during the diversion of the effluent. The nutrient uptake length calculated from the Elovich parameters and stream velocity was about 0.2 m (that is 104 times smaller than what was measured with the nutrient addition experiments; Table 2).
1.4 1.2
Upstream
1
Uptake
Release
-1
EPC0 (µmol L )
Hence, there were large bounds of uncertainties associated with the SRP mass balance for period B, highlighting the difficulties of working in such systems. The average P mass balance during period B was 4.4 mmol m2 day1 (ie. uptake). The magnitude of the SRP cycling rate differed (although within the range of uncertainties) between the mass balance and the nutrient spiralling approaches. However, these techniques showed common patterns; SRP uptake after 24 h of WWTP discharge (22-Aug) and then net release of SRP after 72 h (24-Aug), due to a sudden decrease in SRP concentration coming from the WWTP effluent (Table 2, Haggard et al., 2001). It is only recently that the variability in nutrient flux from the WWTP effluent (a long known fact, e.g. Moss et al., 1988) has been explicitly acknowledged to affect the measured nutrient exchange rate between the river bed sediment and water column (net sorption or release, e.g. Haggard et al., 2005). As the effluent pipe was switched off the SRP mass balance immediately switched to become positive þ9 mmol m2 day1 (indicating release; even accounting for the degree of uncertainty of 1 to þ18 mmol m2 day1). Over the following four days there was approximately linear decrease in the rate of SRP release from the river towards an equilibrium state with respect to P release. Patterns in SRP fluxes were similar between the 2440 m reach (reach 2) and the 229 m (reach 1) immediately downstream of the WWTP, notably in the period of SRP release after treated effluent ceased. The principal difference was the greater stability in the mass balance for the longer reach, which maintained near-consistent P uptake (2.9 mmol m2 day1) averaged over period B. The slightly smaller average P uptake for the longer reach, than immediately downstream of the WWTP in period B indicated that reach SRP uptake declined with distance as river SRP concentrations driving the uptake decreased. In terms of the mass balance for the whole reach areas, of a total 542 mol of SRP entering the river during the 14 day period B only 41 mol (8%) was taken up in the 229 m reach immediately downstream, whereas 289 mol (54%) of this input load were taken up in the longer 2440 m reach. Previously reported point source P retention rates, under low flows, have varied greatly: 30% in a 55 km reach in England (House, 2003), 60% over 1e2 km lengths in England (Jarvie et al., 2002) and 63% over 4e8 km reach lengths in the U.S. (Ekka et al., 2006). The amount of P taken up in the 2.4 km downstream reach indicated the river ecosystem nearer the WWTP was unable to protect downstream reaches and buffer the P inputs. Of the P taken up in reach 1 and 2 respectively 39% and 30% was released when effluent flow ceased (period C). Therefore, the majority of the SRP taken up into these reaches was retained; becoming sequestered by sediment sorption, or incorporated into biota and made unavailable for river bed release in the short term.
Downstream 1:1 line Direction of hysteresis
0.8 0.6 0.4 0.2 0 0
1
2
3
4
5
-1
Langmuir adsorption isotherms showed consistent P saturation of approximately 1000 mmol kg1 with no significant differences temporally, or spatially (Table S3). Sediment EPC0 values (the concentration of solution SRP that would cause no net change in sorption) and native adsorbed P (Ni, i.e. P sorbed
River SRP (µmol L ) Fig. 3 e Change in sediment EPC0 and (1 standard error bars), relative to water column SRP during the experimental period for upstream and downstream sites of reach 1.
4432
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 2 5 e4 4 3 6
Of course, we must exercise caution in these batch study results since the redox conditions of these simulations differ from those of in-situ bed sediments. For example, the depth of the sediment exposed to the water SRP, diffusion and pore water concentration gradients that make such simplistic assessments difficult (e.g. House and Denison, 2002). Batch bottle EPC0 and kinetic isotherms indicate the maximum reaction rate since they describe a system where the sediment is shaken with the water, although this rate could apply if bed sediment became resuspended by a storm event. Consequently, we suggest, batch studies, despite being a common approach adopted in such studies, are inappropriate for upscaling to whole river bed P budgets and modelling (e.g. Wade et al., 2002).
3.4.
Biological P cycling
Three methods were used to quantify algal and bacterial P uptake: whole stream metabolism, algal in-situ assay and a laboratory flow-through column sorption experiment. The whole stream metabolism results (Table 3) are compared here between stable low flow periods (under which our tracer studies were conducted; Fig. S3). Uncertainties in daily ER and GPP estimates were about 42% and 2% respectively. The metabolic rates were generally lower downstream from the WWTP effluent at all time ( p < 0.002), contrary to observations in the River Erpe (Gu¨cker et al., 2006). ER (range day1) exceeded GPP (range 2.2e4.4 g C m2 2 1 0.6e1.6 g C m day ) at all times as is generally the case due to allochtonous source of matter and energy (Battin et al., 2008). The metabolic activity of Tarland Burn was relatively low (cf Battin et al., 2008), particularly compared to the highly productive River Erpe (12e22 g C m2 day1), although no uncertainties (which may easily be around 100%) were
reported in Gu¨cker et al. (2006). The metabolic activity of the water column derived from BOD5 measurements, provided by the Scottish Environmental Protection Agency, represented only about 2% of ER. Therefore, 98% of ER was from the river bed, more than in the larger River Spree (Fischer and Pusch, 2001). GPP increased over time but this was not related to the experiment ( p > 0.5), unlike from the previous descriptive observations (Gu¨cker et al., 2006). ER however increased by 67% relative to the upstream control (note decrease in ER in the upstream control), and therefore was affected by the effluent diversion ( p < 0.001). This confirmed previous observations made upstream and downstream of treated effluents (Gu¨cker et al., 2006). Heterotrophic respiration ranged from 2.3 to 3.9 g C m2 day1, similar to the total bacterial respiration estimates of the River Spree (Fischer and Pusch, 2001). Phosphorus uptake rate from autotrophic growth ranged from 0.2 (0.1e0.3) to 0.4 (0.2e0.7) mmol P m2 day1. In the future, uncertainties could be reduced by determining the C:P ratio of the primary producers both upstream and downstream the effluent, rather than relying on published studies (Bowman et al., 2005). Estimates of P uptake rates from heterotrophic growth (using low to moderate bacterial growth efficiencies with different C:P ranges) were between 0.1 and 1.2 mmol P m2 day1 (Table 3). The average total biological uptake rate increased from about 0.51 to 0.70 mmol P m2 day1 downstream of the WWTP effluent, equivalent to 50% increase relative to the upstream control ( p < 0.003) stable at 0.9 mmol P m2 day1. Hence the biota contributed to the same order of magnitude of P uptake as those measured by whole river mass balance and nutrient addition studies. Note that this method did not quantify the additional uptake by the polysaccharide matrix of the biofilm. Chlorophyll a accumulation rates on artificial substrates, placed in the channel during experimental periods, provided
Table 3 e Gross primary productivity (GPP), ecosystem respiration (ER), in g C mL2 dayL1, and P uptake rates, in mmol P mL2 dayL1, by autotrophs (mostly algae) and heterotrophs (mostly bacteria). Data compare upstream and downstream of the effluent pipe, both before and after-effluent diversion to the stream. GPP Before diversion Upstream 16 Aug 1.0 17 Aug 1.0 Downstream 16 Aug 0.6 17 Aug 0.7 After diversion Upstream 07 Sep 1.5 08 Sep 1.2 09 Sep 1.4 10 Sep 1.6 Downstream 07 Sep 1.2 08 Sep 0.8 09 Sep 1.0 10 Sep 1.1
a
Total biological P uptake
ER
Algal P uptake
Low bacterial P uptake
Moderate bacterial P uptake
4.0 4.4
0.25 0.26
0.12 0.13
1.14 1.25
0.88 0.95
2.2 2.3
0.16 0.18
0.06 0.07
0.60 0.62
0.49 0.52
3.8 3.3 3.3 3.5
0.41 0.33 0.36 0.41
0.10 0.09 0.09 0.09
0.96 0.86 0.84 0.87
0.94 0.80 0.83 0.89
3.1 2.9 2.9 2.9
0.31 0.22 0.26 0.30
0.08 0.08 0.08 0.08
0.80 0.81 0.77 0.75
0.75 0.67 0.69 0.71
a Algal P uptake þ average bacterial P uptake.
4433
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 2 5 e4 4 3 6
-2
-1
Algal P uptake (mmol m day )
0.3
0.2
Upstream
0.1
Downstream Direction of hysteresis
0 0
1
2
3
4
5
-1
River SRP (µmol L ) Fig. 4 e Change in benthic algal P uptake (1 standard error bars) as derived from the biofilm accrual experiment, relative to water column SRP during the experimental period for upstream and downstream sites of reach 1.
a direct biological measure. The algal P uptake rate (0.05e0.23 mmol P m2 day1), derived from this chlorophyll a accumulation, was remarkably comparable to the whole river GPP measurements. Downstream of the WWTP effluent algal P uptake was greatly enhanced by the elevated water SRP concentrations generated during effluent discharge (Fig. 4). However, algal P uptake continued to be stimulated aftereffluent discharge ceased, possibly due to the initial impact of river bed P release during the initial period of algal colonisation on the bricks, and was significantly greater than downstream prior to effluent discharge (t-test; p ¼ 0.02), or the
4
b
Inflow SRP concentration for uptake phase
3.5
160 Column A_uptake phase Column A_release phase Column A_abiotic adsorption Column B_uptake phase Column B_release phase Column B_abiotic adsorption
140
3
-1
Column eluant SRP (µmol L )
-1
Cumulative SRP sorption (µmol kg )
a
upstream control ( p ¼ 0.04). These results were similar to other short-term bioassays (Francoeur, 2001) but differ somewhat from the whole stream GPP which did not respond to the WWTP effluent. The river bed biofilm was most likely not limited by P as inferred by the nutrient addition studies. Mature biofilms may trap sufficient nutrient in the polysaccharide matrix (Freeman and Lock, 1995; Battin et al., 2003), however this would not be the case in the short-term algal assays. We also used a flow-through column sediment sorption system to describe uptake and release of sorbed P (Fig. 5). The column SRP uptake at 20 C (as compared to average in-situ water temperatures of 13 C) and without biocide describes the combined abiotic sorption and biological (bacterial) uptake of P for the sediment exposed to SRP concentrations equal to in-river concentrations during effluent pipe flow. The two saturated sediment columns, following pre-conditioning (see methods), were exposed to elevated concentrations of 3.8 mmol L1 SRP (Fig. 5a). The rapid increase in eluant SRP concentration on the sorption period integrates sediment SRP sorption, bacterial uptake and hydraulic interactions between column mobile and immobile flow regions. Column outflow concentrations reached a plateau at an average of 3.0 mmol L1 within 300e400 pore volumes. The assumption is then applied that the concentration difference (inflows e outflows ¼ 0.8 mmol P L1) multiplied by the flow rate indicated a biological uptake rate of 1.4 mmol P day1 per column (48 g sediment). The basis for this was the premise that (i) the concentration plateau represented a state of near maximum abiotic sorption, (ii) that any biotic (bacterial) kinetics of P uptake greatly exceeded the remaining slow kinetic abiotic P sorption, and (iii) that the difference in inflow and outflow concentrations must be due to biotic (bacterial) P uptake (at steady-state).
3
2.5
2
2
1.5
1 1
0.5
120
100
80
60
40
20
Inflow SRP concentration for release phase
0
0 0
200
400
600
Pore Volumes
800
10 0 0
0
10 0
20 0
30 0
40 0
5 00
Pore Volumes
Fig. 5 e Sediment SRP uptake and release phases for duplicate sediment laboratory flow column experiments, expressed in terms of (a) column eluant concentrations (dashed parallel lines indicate the electrolyte SRP concentrations) and (b) cumulative P uptake/release with and without subtraction of the biotic uptake rate. Numbered annotations in part (a) indicate: 1, uptake phase (square symbols); 2, release phase (triangles); 3, observed plateau in outflow concentrations and difference compared to high SRP inflow concentrations used to infer the heterotrophic P uptake rate.
4434
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 2 5 e4 4 3 6
Since the column oxygen supply (0.4 mmol O2 day1; flow rates O2 water solubility at 20 C) did not limit biological P consumption, this experimental system provided an estimate of potential heterotrophic P uptake at the surface of oxygenated bed sediments. Using a sediment density (from packed column measurements) of 32 kg per m2 of river bed for a 2 cm deep layer of sediment (as in Fischer and Pusch, 2001) infers an ‘in-river’ biological uptake rate of 0.9 mmol P m2 day1 from this column system. This is comparable to P uptake based on ER (0.1e1.2 mmol P m2 day1). Using the background electrolyte of 0.5 mmol L1 CaCl2 (matching river Ca concentrations) and with a sediment Ca content of 28 g kg1 there should be no P precipitation in the column. When the P uptake/release is expressed as cumulative mass (Fig. 5b) then only 29% of the total P uptake was released using these flow volumes. However, when we subtracted the biological (bacterial) uptake rate from the cumulative column sorption we found close agreement between the resulting abiotic component of sorptive uptake and release. This indicated that there was limited hysteresis in terms of the abiotic component of adsorption and desorption of P onto the sediment exchange surfaces. In future, a more realistic open column flow experiments could be based on intact sediment cores (e.g. Fischer et al., 2002). The depth of the sediment core could be based on the depth profile of redox or oxygen concentrations (including/excluding the anoxic zone to test the triple zone model; House and Denison, 2002, and monitoring competitive metabolic pathways; Mermillod-Blondin et al., 2005).
3.5. Relative contributions to P cycling during effluent diversion The most accurate and relevant measurements to calculate the contribution of algae, bacteria and abiotic P uptake were the whole stream metabolism (average of before and aftereffluent diversion) and nutrient spiralling studies (average of first 24 h) as measured downstream of the WWTP. Based on these estimates, net P uptake by algae, bacteria and sediment were about 0.2 (0.1), 0.4 (0.3), and 1.0 (0.9) mmol m2 day1. This assumes that abiotic uptake ¼ total uptake e (algal þ bacterial uptake). Note again, however, that some of the abiotic uptake can come from the polysaccharide matrix of the biofilm. Therefore it is likely that biotic and abiotic uptakes were more equivalent than our results suggest as our methodology may have ascribed some biological P uptake in with the in-situ sediment uptake value. This would explain the small amount of P released in the recovery period (most likely due to fast abiotic P release).
4.
Conclusions
Whole ecosystem P behaviour as measured in-situ differs markedly from closed system studies. The batch studies showed unrealistic results and their use should probably be limited to the EPC0 values for suspended sediment or water column studies (rather than bed sedimentewater interactions). Where in-situ whole river studies (nutrient additions, stream metabolism) are not practically feasible (e.g. large river) then, open flow-through columns or flumes should be
used. In practice the whole system mass balance study is the simplest approach but it requires continuous monitoring of the WWTP effluent (discharge and pollutant concentration) to reduce the uncertainties reported here. We have shown that biotic and abiotic uptake was of the same order of magnitude with both bioassays and whole river studies based on metabolic activities. Further work should be carried out to reduce our reported uncertainties. Differences in behaviour were probably related to differences in acclimation and history of the systems (e.g. new versus old biofilm, dry versus wet sediment). Despite some inherent limitations in the present study, it is the first experiment involving a WWTP that has been able to separate the role of the biota from the role of abiotic river bed by combining complementary methods. These independent methods also provided good agreement. We have shown that, even in response to relatively small effluent point sources, ecosystem P retention occurs over several kilometres and requires considerably longer than exposure times for P rerelease under low stable flow conditions. The internal cycling of P involves autotrophic and heterotrophic components. The prolonged period of P retention following even short exposure times to elevated P means (i) limited potential for periods when ecosystem P saturation may decline and (ii) decreasing downstream buffering for cumulative sources. Periods of sediment scour by storms perhaps represent one of few opportunities for reducing internal P status and show the importance of managing and maintaining appropriate fluvial dynamics. However, such entrained sediment P also has implications for downstream water quality.
Acknowledgements We thank the Scottish Government’s Rural Environment Research and Analysis Directorate for funding, Scottish Water for manipulation of the waste water treatment plant, the Scottish Environmental Protection Agency for providing information, Y. Cook, C. Taylor, H. Watson, L. Clark, R. Gwatkin and S. Richards for assistance in field and laboratory investigations. We also thank two anonymous reviewers for suggesting valuable ways to improve the communication of our findings.
Appendix. Supplementary material Supplementary material can be found, in the online version, at doi:10.1016/j.watres.2010.06.014.
references
Allen, J.A., Scaife, P.H., 1966. The Elovich equation and chemisorption kinetics. Aust. J. Chem. 19, 2015e2023. Ball, R.C., Hooper, F.F., 1963. Translocation of Phosphorus in a Trout Stream Ecosystem. In: Schultz, V., Klement, A.W.J. (Eds.), Radioecology. Reinhold, New York, pp. 217e228.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 2 5 e4 4 3 6
Battin, T.J., Kaplan, L.A., Newbold, J.D., Hansen, C.M.E., 2003. Contributions of microbial biofilms to ecosystem processes in stream mesocosms. Nature 426, 439e442. Battin, T.J., Kaplan, L.A., Findlay, S., Hopkinson, C.S., Marti, E., Packman, A.I., Newbold, J.D., Sabater, F., 2008. Biophysical controls on organic carbon fluxes in fluvial networks. Nat. Geosci. 1, 95e100. Bernot, M.J., Tank, J.L., Royer, T.V., David, M.B., 2006. Nutrient uptake in streams draining agricultural catchments of the midwestern United States. Freshwater Biol. 51, 499e509. Bowman, M.F., Chambers, P.A., Schindler, D.W., 2005. Changes in stoichiometric constraints on epilithon and benthic macroinvertebrates in response to slight nutrient enrichment of mountain rivers. Freshwater Biol. 50, 1836e1852. CEC, 1991. Council of European Communities. Directive concerning urban waste water treatment (91/271/EEC). Off. J. L135/40. Chardon, W.J., Menon, R.G., Chien, S.H., 1996. Iron oxide impregnated filter paper (Pi test): a review of its development and methodological research. Nut. Cycling Agroecosyst. 46, 41e51. D’Angelo, D.J., Webster, J.R., Benfield, E.F., 1991. Mechanisms of stream phosphorus retention e an experimental-study. J.N. Am. Benthol. Soc. 10, 225e237. Demars, B.O.L., 2008. Whole-stream phosphorus cycling: testing methods to assess the effect of saturation of sorption capacity on nutrient uptake length measurements. Water Res. 42, 2507e2516. Demars, B.O.L., Harper, D.M., 2005. Water column and sediment phosphorus in a calcareous lowland river and their differential response to point source control measures. Water Air. Soil. Pollut. 167, 273e293. Demars, B.O.L., Harper, D.M., Pitt, J.-A., Slaughter, R., 2005. Impact of phosphorus control measures on in-river phosphorus retention associated with point source pollution. Hydrol. Earth Syst. Sci. 9, 43e55. Dorioz, J.M., Pilleboue, E., Ferhi, A., 1989. Phosphorus dynamics in watersheds: role of trapping processes in sediments. Water Res. 23, 147e158. Doyle, M.W., Stanley, E.H., Harbor, J.M., 2003. Hydrogeomorphic controls on phosphorus retention in streams. Water Resour. Res. 39, 1147. Ekka, S.A., Haggard, B.E., Matlock, M.D., Chaubey, I., 2006. Dissolved phosphorus concentrations and sediment interactions in effluent-dominated Ozark streams. Ecol. Eng 26, 375e391. Fischer, H., Pusch, M., 2001. Comparison of bacterial production in sediments, epiphyton and the pelagic zone of a lowland river. Freshwater Biol. 46, 1335e1348. Fischer, H., Sachse, A., Steinberg, C.E.W., Pusch, M., 2002. Differential retention and utilization of dissolved organic carbon by bacteria in river sediments. Limnol. Oceanogr. 47, 1702e1711. Francoeur, S.N., 2001. Meta-analysis of lotic nutrient amendment experiments: detecting and quantifying subtle responses. J.N. Am. Benthol. Soc. 20, 358e368. Freeman, C., Lock, M.A., 1995. The biofilm polysaccharide matrix: a buffer against changing organic substrate supply? Limnol. Oceanogr 40, 273e278. Gainswin, B.E., House, W.A., Leadbeater, B.S.C., Armitage, P.D., Patten, J., 2006. The effects of sediment size fraction and associated algal biofilms on the kinetics of phosphorus release. Sci. Total Environ. 360, 142e157. Gu¨cker, B., Brauns, M., Pusch, M.T., 2006. Effects of wastewater treatment plant discharge on ecosystem structure and function of lowland streams. J.N. Am. Benthol. Soc. 25, 313e329. Gu¨cker, B., Pusch, M.T., 2006. Regulation of nutrient uptake in eutrophic lowland streams. Limnol. Oceanogr 51, 1443e1453.
4435
Gundersen, K., Heldal, M., Norland, S., Purdie, D.A., Knap, A.H., 2002. Elemental C, N, and P cell content of individual bacteria collected at the Bermuda Atlantic Time-Series Study, BATS) site. Limnol. Oceanogr 47, 1525e1530. Haggard, B.E., Stanley, E.H., Hyler, R., 1999. Sediment-phosphorus relationships in three northcentral Oklahoma streams. Trans. ASAE. 42, 1709e1714. Haggard, B.E., Storm, D.E., Stanley, E.H., 2001. Effect of a point source input on stream nutrient retention. J. Am. Water Resour. Ass 37, 1291e1299. Haggard, B.E., Stanley, E.H., Storm, D.E., 2005. Nutrient retention in a point-source-enriched stream. J.N. Am. Benthol. Soc. 24, 29e47. Hall, R.O., Tank, J.L., 2003. Ecosystem metabolism controls nitrogen uptake in streams in Grand Teton National Park. Wyoming. Limnol. Oceanogr 48, 1120e1128. House, W.A., 2003. Geochemical cycling of phosphorus in rivers. Appl. Geochem. 18, 739e748. House, W.A., Denison, F.H., Armitage, P.D., 1995. Comparison of the uptake of inorganic phosphorus to a suspended and stream bed-sediment. Water Res. 29, 767e779. House, W.A., Denison, F.H., 2002. Exchange of inorganic phosphate between river waters and bed-sediments. Environ. Sci. Technol. 36, 4295e4301. Jarvie, H.P., Ju¨rgens, M.D., Williams, R.J., Neal, C., Davies, J.J.L., Barrett, C., White, J., 2005. Role of river bed sediments as sources and sinks of phosphorus across two major eutrophic UK river basins: the Hampshire Avon and Herefordshire Wye. J. Hydrol. 304, 51e74. Jarvie, H.P., Neal, C., Williams, R.J., Sutton, E.J., Neal, M., Wickham, H.D., et al., 2002. Phosphorus sources, speciation and dynamics in the lowland eutrophic River Kennet, UK. Sci. Total. Environ. 282e283, 175e203. Jarvie, H.P., Neal, C., Withers, P.J.A., 2006. Sewage effluent phosphorus: a greater risk to river eutrophication than agricultural phosphorus? Sci. Total. Environ. 360, 246e253. Kahlert, M., 1998. C:N:P ratios of freshwater benthic algae. Arc. Hydrobiol. Adv. Limnol 51, 105e114. Lottig, N.R., Stanley, E.H., 2007. Benthic sediment influence on dissolved phosphorus concentrations in a headwater stream. Biogeochemistry 84, 297e309. Marzolf, E.R., Mulholland, P.J., Steinman, A.D., 1994. Improvements to the diurnal upstreamedownstream dissolved oxygen change technique for determining wholestream metabolism in small streams. Can. J. Fish. Aquat. Sci. 51, 1591e1599. McDowell, R.W., 2003. Sediment phosphorus chemistry and microbial biomass along a lowland New Zealand stream. Aquat. Geochem. 9, 19e40. Mermillod-Blondin, F., Mauclaire, L., Montuelle, B., 2005. Use of slow filtration columns to assess oxygen respiration, consumption of dissolved organic carbon, nitrogen transformations and microbial parameters in hyporheic sediments. Water Res. 39, 1687e1698. Moss, B., Balls, H., Booker, I., Manson, K., Timms, M., 1988. Problems in the construction of a nutrient budget for the River Bure and its Broads (Norfolk) prior to its restoration from eutrophication. In: Round, F.E. (Ed.), Algae and the Aquatic Environment. Biopress, Bristol, pp. 327e353. Mulholland, P.J., Helton, A.M., Poole, G.C., Hall, R.O., Hamilton, S.K., Peterson, B.J., Tank, J.L., Ashkenas, L.R., Cooper, L.W., Dahm, C.N., Dodds, W.K., Findlay, S.E.G., Gregory, S.V., Grimm, N.B., Johnson, S.L., McDowell, W.H., Meyer, J.L., Valett, H.M., Webster, J.R., Arango, C.P., Beaulieu, J.J., Bernot, M.J., Burgin, A.J., Crenshaw, C.L., Johnson, L.T., Niederlehner, B.R., O’Brien, J.M., Potter, J.D., Sheibley, R.W., Sobota, D.J., Thomas, S.M., 2008. Stream denitrification across
4436
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 2 5 e4 4 3 6
biomes and its response to anthropogenic nitrate loading. Nature 452, 202e205. Newbold, J.D., Elwood, J.W., O’Neill, R.V., Sheldon, A.L., 1983. Phosphorus dynamics in a woodland stream ecosystem: a study of nutrient spiralling. Ecology 64, 1249e1265. Newbold, J.D., Bott, T.L., Kaplan, L.A., Dow, C.L., Jackson, J.K., Aufdenkampe, A.K., Martin, L.A., Van Horn, D.J., de Long, A.A., 2006. Uptake of nutrients and organic C in streams in New York City drinking-water-supply watersheds. J.N. Am. Benthol. Soc. 25, 998e1017. Rodda, J.C., 2007. Sustaining water resources in South East England. Atmos. Sci. Lett. 7, 73e77. Slavik, K., Peterson, B.J., Deegan, L.A., Bowden, W.B., Hershey, A.E., Hobbie, J.E., 2004. Long-term responses of the Kuparuk River ecosystem to phosphorus fertilization. Ecology 85, 939e954. Small, G.E., Helton, A.M., Kazanci, C., 2009. Can consumer stoichiometric regulation control nutrient spiraling in streams? J.N. Am. Benthol. Soc. 28, 747e765. Sobczak, W.V., Cloern, J.E., Jassby, A.D., Mu¨ller-Solger, A.B., 2002. Bioavailability of organic matter in a highly disturbed estuary: the role of algal and detrital sources. Proc. Natl. Acad. Sci. U S A 99, 8101e8105. Stutter, M.I., Langan, S.J., Cooper, R.J., 2008. Spatial and temporal dynamics of stream water and particulate N, P and C forms along a catchment transect, NE Scotland. J. Hydrol. 350, 187e202. Talling, J.F., Driver, D., 1963. Some problems in the estimation of chlorophyll-a in phytoplankton. TID-7633. In: Proceedings, conference of primary productivity measurement, marine and freshwater, Hawaii, 1961. U.S. Atomic Energy Commission, pp. 142e146.
Taylor, A.W., Kunishi, H.M., 1971. Phosphate equilibria on stream sediment and soil in a watershed draining an agricultural region. J. Agric. Food Chem. 19, 827e831. Thingstad, T.F., Riemann, B., Havskum, H., Garde, K., 1996. Incorporation rates and biomass content of C and P in phytoplankton and bacteria in the Bay of Aarhus (Denmark) June 1992. J. Plankton Res. 18, 97e121. Ukpong, M., 2005. Element homeostasis in bacteria. McNair Res. J. Summer, 28e29. Van Vliet, M.T.H., Zwolsman, J.J.G., 2008. Impact of summer droughts on the water quality of the Meuse river. J. Hydrol. 353, 1e17. Wade, A.J., Whitehead, P.G., Hornberger, G.M., Jarvie, H.P., Flynn, N., 2002. On modelling the impact of phosphorus stripping at sewage works on in-stream phosphorus and macrophyte/epiphyte dynamics: a case study for the River Kennet. Sci. Total. Environ. 282e283, 395e416. Withers, P.J.A., Jarvie, H.P., 2008. Delivery and cycling of phosphorus in rivers: a review. Sci. Total Environ. 400, 379e395. Webster, J.R., Valett, T.P., 2006. Solute Dynamics. In: Hauer, F.R., Lamberti, G.A. (Eds.), Methods in Stream Ecology, second ed. Academic Press, San Diego, pp. 145e160. Webster, J.R., Newbold, J.D., Thomas, S.A., Valett, H.M., Mulholland, P.J., 2009. Nutrient uptake and mineralization during leaf decay in streams e a model simulation. Int. Rev. Hydrobiol. 94, 372e390. Young, R.G., Huryn, A.D., 1998. Comment: improvements to the diurnal upstreamedownstream dissolved oxygen change technique for determining whole-stream metabolism in small streams. Can. J. Fish. Aquat. Sci. 55, 1784e1785.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 0 5 e4 5 1 6
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Role of extracellular polymeric substances in bioflocculation of activated sludge microorganisms under glucose-controlled conditions Appala R. Badireddy a, Shankararaman Chellam a,b,*, Paul L. Gassman c, Mark H. Engelhard c, Alan S. Lea c, Kevin M. Rosso c a
Department of Civil and Environmental Engineering, University of Houston, Houston, TX 77204-4003, USA Department of Chemical and Biomolecular Engineering, University of Houston, Houston, TX 77204-4004, USA c Environmental Molecular Sciences Laboratory, Pacific Northwest National Laboratory, Richland, WA 99352, USA b
article info
abstract
Article history:
Extracellular polymeric substances (EPS) secreted by suspended cultures of microorganisms
Received 29 December 2009
from an activated sludge plant in the presence of glucose were characterized in detail using
Received in revised form
colorimetry, X-ray photoelectron spectroscopy (XPS) and Fourier transform infrared (FTIR)
7 June 2010
spectroscopy. EPS produced by the multi-species community were similar to literature
Accepted 8 June 2010
reports of pure cultures in terms of functionalities with respect to C and O but differed subtly
Available online 16 June 2010
in terms of N and P. Hence, it appears that EPS produced by different microorganisms maybe homologous in major chemical constituents but may differ in minor components such as
Keywords:
lipids and phosphodiesters. The role of specific EPS constituents on microbial aggregation
Extracellular polymeric substances
was also determined. The weak tendency of microorganisms to bioflocculate during the
Wastewater treatment
exponential growth phase was attributed to electrostatic repulsion when EPS concentration
Activated sludge
was low and acidic in nature (higher fraction of uronic acids to total EPS) as well as reduced
Biofilms
polymer bridging. However, during the stationary phase, polymeric interactions overwhelmed electrostatic interactions (lower fraction of uronic acids to total EPS) resulting in improved bioflocculation. More specifically, microorganisms appeared to aggregate in the presence of protein secondary structures including aggregated strands, b-sheets, a- and 3turn helical structures. Bioflocculation was also favored by increasing O-acetylated carbohydrates and overall Ce(O,N) and O] C eOH þ O] C eOR functionalities. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
Extracellular polymeric substances (EPS) are biosynthetic polymers of microbial origin, present on or outside the cell. A multi-species community of microorganisms is often embedded in an EPS matrix in technological systems such as membrane bioreactors and wastewater treatment processes
primarily because they can better acquire substrates and nutrients, maintain hydration, and protect themselves from environmental stresses and predation (Wingender et al., 1999). Proteins and polysaccharides constitute the major EPS components, with nucleic acids, humic acids, lectins, lipids and other polymers being present in lower concentrations.
* Corresponding author. Department of Civil and Environmental Engineering, University of Houston, Houston, TX 77204-4003, USA. Tel.: þ1 (713) 743 4265; fax: þ1 (713) 743 4260. E-mail address:
[email protected] (S. Chellam). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.024
4506
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 0 5 e4 5 1 6
Biodegradation of organic compounds present in municipal and industrial effluents is most commonly accomplished using the activated sludge process where microorganisms exist as large aggregates held together by EPS (Bossier and Verstraete, 1996; Raszka et al., 2006). Interestingly, limited available data suggests that in some instances, excessive amounts of EPS might be detrimental to bioflocculation in the activated sludge process resulting in reduced efficiency of liquidesolid separation in the secondary settling tank or clarifier (Wile´n et al., 2003a; Li and Yang, 2007) suggesting that gross EPS concentrations alone do not determine microorganism flocculation. Therefore, it is necessary to determine the role of individual EPS moieties in controlling bioflocculation (Sobeck and Higgins, 2002; Geesey and Van Ommen Kloeke, 2005; Park et al., 2008; Ras et al., 2008; Ni et al., 2009) since growth of non-settleable microorganisms increases effluent turbidity from the secondary clarifier and reduces biochemical oxygen demand (BOD) removal. Experiments with pure cultures suggest that cell surface characteristics such as surface charge and hydrophobicity determine bacterial aggregation prior to significant EPS secretion (Dufreˆne and Rouxhet, 1996; Zita and Hermansson, 1997). Certain protein secondary structures and acylated polysaccharides present in EPS have also shown to play important roles in bioflocculation and adhesion (Beech et al., 1999; Kalaji and Neal, 2000; Nivens et al., 2001; Sutherland, 2001; Omoike and Chorover, 2004, 2006; Tielen et al., 2005; Parikh and Chorover, 2006; Eboigbodin et al., 2007; Eboigbodin and Biggs, 2008). For example, b-sheets were reduced and unordered secondary structures were increased in Pseudomonas sp. biofilms compared with planktonic bacteria (Beech et al., 1999; Kalaji and Neal, 2000). Solution pH is an important determinant of EPS protein conformation with a-helices dominating at acidic pH and random coils dominating at neutral pH during Bacillus subtilis adsorption onto goethite (Omoike and Chorover, 2004, 2006). Another study has shown that surface proteins of a Gram-positive bacteria (B. subtilis) adopted a ahelical conformation whereas those of Gram-negative bacteria (Shewanella oneidensis and Pseudomonas aeruginosa) assumed random coil structures when adsorbed on to iron oxide (Parikh and Chorover, 2006). Hence, results from pure cultures cannot be directly extended to aggregation in environmental systems which incorporate mixed communities. Aggregation of activated sludge microorganisms is often only related to gross measurements of total proteins and polysaccharides (Frølund et al., 1996; Wile´n et al., 2000, 2003a, b). A limited number of studies have also been performed characterizing the molecular weight and size distribution and functionalities of activated sludge EPS including specific amino acids, lectins, subfamilies of proteins and polysaccharides, as well as bridging by divalent cations (Higgins and Novak, 1997; Go¨rner et al., 2003; Garnier et al., 2005; Park et al., 2008; Ras et al., 2008; Ni et al., 2009). These studies suggest the role of specific EPS moieties including proteins and carbohydrate functionalities on aggregation. Consequently, we undertook this study to rigorously and systematically characterize EPS from an activated sludge culture and determine potential links between its specific components and bioaggregation with the ultimate goal of developing a molecular level understanding of bioflocculation.
This work is based on the hypothesis that bioflocculation of activated sludge microorganisms is controlled not only by the total amount of EPS but also by individual EPS chemical functionalities and structures. The specific objective was to perform a detailed molecular and elemental characterization of EPS and correlate them with separate measurements of bioflocculation ability. EPS were extracted from a multi-species community of microorganisms obtained from a local wastewater treatment plant that were grown over a period of 3 days in filtered and sterilized wastewater enriched with glucose. The effects of growth phase on EPS molecular composition and its relationship to bioflocculation ability were determined. Several measurements were made over the 3-day growth period to relate EPS and bioflocculation ability. Total carbohydrates, proteins, and uronic acids were quantified using colorimetric assays. Fourier transform infrared (FTIR) spectroscopy was used to obtain in-depth information on EPS functionalities including O-acetylated and acidic carbohydrates and protein secondary structures (through second derivative analysis of the amide I region). The elemental composition (C, O, and N molar ratios) and associated functionalities were obtained using X-ray photoelectron spectroscopy (XPS). Flocs were also observed directly using phase contrast microscopy and scanning electron microscopy.
2.
Methods
2.1.
Activated sludge and microbial growth
A 4 L activated sludge sample was obtained from the secondary clarifier treating both industrial and municipal effluents (sludge age approx. 30 days) from the City of Houston’s 69th Street Wastewater Treatment Plant, which is designed for nitrogen removal. Activated sludge was sampled (dipped-anddrawn using a 1 L sterile plastic bottle tethered to the end of a rope) at a depth of approximately 1 m from the surface of the secondary clarifier. The sludge was brought to the laboratory within 30 min after sampling and analyzed immediately for total suspended solids (SM2540 D) volatile suspended solids (SM2540 E), conductivity, and pH which were measured as 4 g/L and 1.5 g/L, 883 mS cm1, and 7.1, respectively. A portion of sludge was allowed to settle for 2 h at 4 C. The supernatant was collected, autoclaved, cooled to room temperature, and then filtered through a Whatman filter (Grade No. 41) to remove suspended particles. The filtrate was similar in conductivity (885 mS cm1) and pH (7.2) to the original wastewater. To simplify laboratory work, batch experiments were performed rather than using continuous flow systems. Since microorganisms from the secondary clarifier in real-world activated sludge units are generally under near-starvation conditions, an electron donor was necessary to stimulate short-term heterotrophic activity and to obtain actively respiring microorganisms. The choice of carbon source (e.g. acetate, glucose, lactate, or propionate) and concentration influences microbial activity and consequently bioflocculation (Wile´n et al., 2000; Li and Yang, 2007). A common and easily assimilated carbon substrate, glucose was employed as the electron donor in this research (Magbanua and Bowers, 2006; Eboigbodin et al., 2007). 1 mL of activated sludge (pipetted
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 0 5 e4 5 1 6
4507
Fig. 1 e Representative images of activated sludge flocs. Phase contrast image (a), epifluorescence micrograph (b), and scanning electron micrograph (c) are shown. Red color is from 5-cyano-2,3-ditolyl tetrazolium chloride (CTC) and the blue color is from 40 ,6-diamidino-2-phenylindole (DAPI). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
with a tip cut to 4 mm diameter) was added to 499 mL of the sterilized water described in the previous paragraph enriched with 0.5 g/L glucose and shaken continuously at 200 rpm for 3 days at 25 C. 100 mL of the culture was collected daily to extract EPS and measure bioflocculation ability. Specific oxygen uptake rate was determined using SM2710 B (APHA et al., 1995). Results reported in this manuscript are from three separate batch experiments conducted in parallel.
Fig. 2 e Concentration of EPS carbohydrates, proteins, and uronic acids. Error bars correspond to standard deviations of colorimetric measurements of three independent samples obtained from separate flasks set up to grow microorganisms.
Fig. 3 e XPS wide survey scans of EPS during 3 days of growth.
4508 2.2.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 0 5 e4 5 1 6
EPS extraction
EPS was extracted following 0, 1, 2, and 3 days of growth using a cation exchange resin (Frølund et al., 1996). 75 mg of the Naþ form of a polystyrene divinylbenzene microporous ion exchange resin (Dowex 50WX8, 20e50 mesh, Sigma Aldrich)/g
volatile suspended solids (VSS) was added to 50 mL of sample and shaken at 900 rpm for 4 h at 4 C. A two-step centrifugation at 4 C (first at 5000g for 15 min and then at 12,000g for 30 min) followed by filtration using a 0.45 mm cellulose acetate membrane was used to remove resin, microorganisms, and residual debris to obtain an EPS sample for further analysis. The daily EPS samples were stored at 20 C until further analysis.
2.3.
Colorimetry
A portion of purified EPS sample was dialyzed (at 4 C) using regenerated cellulose 2000 Da molecular weight cut-off membranes (Spectra/Por7, Spectrum Chemicals, Gardena, CA) against three changes of ultrapure water per day for 3 days to remove potentially interfering ions such as Kþ. Total proteins and humic compounds were together quantitated by the Modified Lowry Protein Assay Kit (Pierce Biotechnology, Rockford, IL) with bovine serum albumin standards. Humic compounds were determined using the original Lowry method without CuSO4. Protein concentrations were obtained by subtracting humic substances (Frølund et al., 1995). Carbohydrates were measured using the phenolesulfuric acid method against glucose standards. Uronic acids were determined by first adding a 40 mL aliquot of EPS to the wells in a microtiter plate (van den Hoogen et al., 1998). Next, 200 mL of concentrated sulfuric acid (96% (w/w)) containing 120 mM sodium tetraborate was added and mixed thoroughly by pipetting in and out three times. After incubating at 80 C for 1 h, the plate was brought to room temperature and the absorbance at 540 nm was obtained as the background reading. To the above mixture, 100 mL of mhydroxydiphenyl reagent (freshly prepared by mixing 100 mL dimethyl sulfoxide (100 mg/mL) with 4.9 mL of 80% (w/w) sulfuric acid) was added, mixed well, and kept at room temperature for 15 min. Once the samples turned pink their absorbance at 540 nm was measured again. Following background subtraction, uronic acids were quantitated using a standard curve prepared using D-glucuronic acid.
2.4.
Fig. 4 e High resolution 1s XPS spectra of carbon (a), oxygen (b), and nitrogen (c) from EPS after 1-day growth.
X-ray photoelectron spectroscopy (XPS)
XPS (Physical Electronics Quantum, 2000 Scanning ESCA Microprobe) was utilized to estimate the elemental composition of EPS samples and assess the functionalities associated with carbon and oxygen. Similar to our recent report (Badireddy et al., 2008b), measurements were performed on 100 mL of air-dried EPS sample deposited on sputter cleaned 1 cm2 pieces of a highly polished silicon wafer. The 100 mL application and drying procedure yielded a film of sufficient thickness so as to minimize the interferences from the silicon substrate. This system uses a focused monochromatic Al Ka X-ray (1486.7 eV) source and a spherical section analyzer equipped with a 16-element multi-channel detector. A 100 W, 100 mm diameter X-ray beam was rastered over a 1.3 mm by 0.2 mm rectangle on the sample. The incident X-ray beam was normal to the sample and the photoelectron detector was at 45 and 90 off-normal with an analyzer angular acceptance width of 20 20 . Wide scan (step size 1 eV) data was collected using a pass energy of 117.4 eV. For the Ag3d5/2 line, these conditions produced the full width at half-maximum (FWHM) of better than 1.6 eV. The high energy resolution (step size 0.1 eV) photoemission spectra were collected using a pass
4509
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 0 5 e4 5 1 6
Table 1 e O and N atomic ratios and functional groups with respect to total C obtained from high resolution XPS spectra of EPS extracted from mixed cultures. Day
0 1 2 3
Elemental composition (molar ratio vs. total carbon)
Chemical functional groups (molar ratio with respect to total carbon)
(O þ N)/C
O/C
N/C
284.8 eV C eðC; HÞ
286.2 eV C eðO; NÞ
287.8 eV C ]O;Oe C eO
289.2 eV O] C eOH;O] C eOR
531.3 eV O ]C
532.7 eV H O eC;Ce O eC
0.884 0.959 1.36 1.16
0.764 0.880 1.28 1.07
0.120 0.079 0.075 0.086
0.497 0.462 0.400 0.404
0.293 0.310 0.393 0.418
0.182 0.179 0.151 0.118
0.028 0.049 0.056 0.060
0.397 0.465 0.660 0.449
0.367 0.415 0.623 0.623
energy of 25.35 eV. For the Ag3d5/2 line, these conditions produced a FWHM of better than 0.88 eV. The binding energy (BE) scale was calibrated using the Cu2p3/2 feature at 932.62 0.05 eV and Au 4f at 83.96 0.05 eV for known standards. Because samples experienced variable degrees of charging, low energy electrons at w1 eV, 20 mA and low energy Arþ ions were used to minimize this charging (Beamson and Briggs, 1992). XPS spectra depicted in this manuscript are the average of at least 10 spectra for each sample.
2.5.
Fourier transform infrared (FTIR) spectroscopy
FTIR spectra of EPS were collected using a Bruker IFS 66v/S FTIR spectrometer, equipped with a Globar source, KBr beam splitter, MCT detector and OPUS (v5.0) operating software similar to our recent publications (Badireddy et al., 2008a,b). All spectra consist of 512 co-added scans collected at 4 cm1 resolution, with 16 cm1 phase resolution, and a zero-filling factor of 2, using a Blackman Harris three-term apodization and Mertz phase correction. A 16 kHz low-pass filter was used to prevent aliasing. Before acquiring spectra, the spectrometer was evacuated and stabilized at 7 mbar vacuum for 3 min to minimize the interferences from atmospheric CO2 and water vapor. 2 mL aliquots of purified EPS were vacuum dried to a pellet in a centrifuge tube overnight at 4 C. This EPS pellet was ground with infrared grade KBr and molded into a disc before collecting spectra in transmission mode. Absorbance spectra were calculated as the ratio of the sample spectrum against the open beam configuration. Each FTIR spectrum depicted in this manuscript is an average of two spectra obtained from separate samples, each of which was obtained by averaging 5 spectra. Peak assignments between 1800 and 600 cm1 were based on vibration bands reported for model organic compounds, biomolecules, and bacteria (see Table 3). Through second derivative analysis and 9-point SavitzkyeGolay
Table 2 e Ratio of proteins, carbohydrates, uronic acids, and hydrocarbons to total carbon in EPS during 3 days of growth. Day 0 1 2 3
Total Carbohydrates Uronic Hydrocarbon-like proteins acids compounds 0.428 0.282 0.268 0.307
0.145 0.183 0.262 0.272
0.028 0.049 0.056 0.060
0.399 0.486 0.414 0.361
smoothing, the amide I region was further resolved into component peaks arising due to various protein secondary structures (Buijs et al., 1996; Beech et al., 1999; Stuart, 2004). A Lorentzian shape was assumed for the amide I band prior to curve fitting the original spectra.
2.6.
Microscopy
Microbial flocs were imaged using an Olympus BX51 phase contrast microscope equipped with a charge-coupled device connected to a personal computer. Prior to examination, a drop of microbial suspension was placed on a glass slide and the excess water was removed slowly and carefully using tissue paper. Samples were dried with utmost care to avoid loss of flocs and minimize changes in floc morphology. Microbial flocs were also placed on conductive carbon tape, sputter coated with a thin (10 nm) layer of carbon, and visualized using a field emission scanning electron microscope (FEI Quanta 400 FEG, Hillsboro, OR). Respiratory activity in the original activated sludge sample was assessed using epifluorescence microscopy using our recently reported procedure (Badireddy et al., 2008b) using 5-cyano-2,3-ditolyl tetrazolium chloride (CTC) and 40 ,6diamidino-2-phenylindole (DAPI).
2.7.
Bioflocculation ability
A previously reported procedure was adopted to measure the tendency of disrupted flocs to reflocculate (Jorand et al., 1994; Wile´n et al., 2003a,b). First, a 50 mL aliquot of microbial cultures in a glass bottle was sonicated for 2 min to disrupt the flocs. Immediately after sonication, a 6 mL sample was centrifuged at 1300g for 2 min at 4 C and the absorbance of the supernatant at 650 nm was recorded (OD0 min). The remaining 44 mL of the sonicated sample was stirred for 15 min allowing microorganisms to reflocculate. As before, 6 mL of sample was taken, centrifuged at 1300g for 2 min at 4 C, and the absorbance of the supernatant (OD15 min) was measured. The “bioflocculation ability” was estimated using (OD0 min OD15 min)\(OD0 min) 100.
3.
Results and discussion
3.1.
Activated sludge microorganisms
As seen in Fig. 1a, biological flocs in the original activated sludge sample were irregularly shaped and contained
4510
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 0 5 e4 5 1 6
filamentous organisms. Epifluorescence microscopy showed that the majority of microorganisms in the initial sample were actively respiring as indicated by the red color (CTC formazan crystals) in Fig. 1b. Examination at a higher magnification revealed that flocs were held together by a dense EPS matrix interspersed between the microorganisms (Fig. 1c).
3.2.
EPS characterization
Activated sludge EPS from short-term batch experiments simulating both nutrient-rich and nutrient-poor conditions were characterized in detail as described below.
3.2.1.
Colorimetric analysis
Fig. 2 demonstrates that during the 3 days of monitoring (i) protein content of EPS was always higher than carbohydrates, and (ii) microbial secretions of proteins, polysaccharides, and uronic acids increased during the growth phase (2 days) but remained relatively constant during stationary phase (2e3 days). The growth curve is shown in Appendix Figure A1. The measured protein to carbohydrate ratios were comparable to other activated sludge samples (Frølund et al., 1996; Wile´n et al., 2003a,b; Park et al., 2008). These ratios decreased monotonically from 3.5 (12 h) to 1.8 (1 day) to 1.4 (36 h) and 1.2 (2 days) during the exponential growth phase but remained constant during the stationary phase (1.2 on day 3). Similarly, the ratio of uronic acids to carbohydrates decreased monotonically from 0.34 (12 h) to 0.31 (1 day) to 0.28 (36 h) and 0.24 (2 days) during the growth phase but remained constant during the stationary phase (0.24 on day 3). These trends are attributed to progressively higher secretion of neutral carbohydrates such as acetals, hemiacetals, and alcohols with glucose depletion (Eboigbodin et al., 2007). The specific oxygen uptake rate increased from 30 to 84 mg DO/h/g VSS demonstrating respiratory activity during the growth phase of our experiments.
3.2.2.
Fig. 5 e FTIR spectra of EPS from activated sludge microorganisms grown for 3 days (a) and second derivative spectrum of a day 1 EPS sample (b). The curve fitted amide I peak of 1-day EPS and corresponding protein secondary structures (c).
XPS analysis
Fig. 3 depicts wide survey XPS spectra, over the energy range of 0e1200 eV of EPS obtained over 3 days of growth. An XPS spectrum is the plot of the number of electrons detected per unit time versus the binding energy of the electrons from the elements present in the material being analyzed. Each peak corresponds to electrons with a characteristic binding energy from a particular element; the peak intensities being proportional to the relative elemental abundances (Beamson and Briggs, 1992). The 0-day EPS was comprised of 52.7% C, 40.3% O, 6.3% N and 0.1% N (nitrate) and 0.6% K and that of 3-day sample 45.4% C, 48.6% O, 3.9% N and 1.3% N (nitrate) and 0.7% K. Even though samples were handled carefully there is a likelihood of a small amount of surface adventitious hydrocarbons (predominantly from C eðC; HÞ). Note the absence of divalent cations (such as Ca and Mg) that are known to cross-link charged EPS moieties and keep microorganisms in an aggregated state in the activated sludge (Sobeck and Higgins, 2002; Park et al., 2008) and the presence of strong Na signals in Fig. 3. This is because the strongly acidic cationic resin exchanged Naþ with Caþþ and Mgþþ thereby weakening the biological flocs allowing efficient EPS extraction. Total elemental nitrogen is separated into N and nitrate to emphasize that XPS spectra are sensitive to both organic and
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 0 5 e4 5 1 6
4511
Table 3 e Band assignments for FTIR spectral features (cmL1) of bound and free EPS (Lin-Vien et al., 1991; Naumann et al., 1991; Schmitt and Flemming, 1998; Barth and Zscherp, 2002; Maquelin et al., 2002; Stuart, 2004). Bound EPS
Free EPS
1741 1633 e 1457 1403 1384 1363 1326 1236 1209
1739 1643 1550 1454 1405 1384 1357 1326 1241 1205
1157 1114, 1080 1043 956 916 858
1155 1112, 1078 1035 952 916 862
Band assignments C]O stretch, esters, O-alkyl group ysC]O stretch (amide I) associated with proteins; NH2 scissors of primary amines dNeH and ysCeN stretches (amide II) associated with proteins dCeOH and dsCH3, dsCH2 possibly associated with proteins ysCOO stretches associated with amino acids ysCeO stretching of COO groups ysCeO and dCeH stretches possibly associated with amino acids dCeH stretches associated with amines and lipids ysCeN stretch associated with secondary amides of proteins (amide III) yasP]O from either nucleic acids or phosphorylated proteins; may also be due to dNeH, dCeH, ysCeN, and ysCeC dCeOH, dCeO, and yCeO possibly associated with amino acids Ring vibrations yP]O, yCeOeC, yCeOeP as in phosphodiesters and polysaccharides yCeOH of phosphorylated proteins and associated alcohols yasOePeO stretches associated with nucleic acids ysCeOeC associated with phosphodiesters Ring “breathing” associated with yCeC and yCeOH
inorganic sources. Proteins in the activated sludge EPS are the primary source for the elemental nitrogen, which is indicated as N. Since microorganisms were obtained from an activated sludge plant designed for nitrification, the nitrate content is indicated separately. Changes in C 1s, N 1s, and O 1s functionalities during 3 days of growth were investigated with high resolution scans to obtain more detailed information on the chemical bonding states of the elements (1 eV step size). The resulting peaks were later resolved into their individual components. Representative narrow scans are shown only for day 1 (see Fig. 4) since XPS spectra were largely homologous in terms of functionalities of EPS constituents for all 3 days and consequently peak decomposition results looked very similar. The C 1s peak was resolved into four component peaks: (i) 284.8 eV ðC eðC; HÞ mainly from hydrocarbonsÞ, (ii) 286.2 eV ðC eðO; NÞfrom proteins and alcoholsÞ; (iii) 287.8 eV (C ]O or Oe C eO, as in carboxylate, carbonyl, amide, acetals, or hemiacetals), and (iv) 289.2 eV (O]CeOH and O]CeOR from uronic acids). The O 1s peak was decomposed into two peaks at 531.3 eV (O ¼ C as in carboxylate, carbonyl, ester, or amide) and 532.7 eV ðO eðC; HÞas in hydroxide; acetals or hemiacetalsÞ. Similar to a study of B. subtilis (Omoike and Chorover, 2004), only non-protonated nitrogen compounds were detected in EPS samples indicating that amides (peptides) were dominant in our experiments. In all cases, the P 2p peak was absent demonstrating that nucleic acids and phospholipids were
Table 4 e Amide I, carbohydrate, and O-acetyl ester band areas of EPS from mixed cultures grown for 3 days (area units). Day 0 1 2 3
Amide I
Carbohydrates
O-Acetyl esters
11.8 35.3 46.2 51.4
24.5 41.0 55.1 62.2
0.21 0.42 0.51 0.54
below the XPS detection limit (total P < 0.1%) and suggesting that cell lysis was insignificant during EPS extraction. C 1s and O 1s functionalities observed in our mixed community have also been reported for pure cultures of B. subtilis (Omoike and Chorover, 2004), Serratia marcescens (Ahimou et al., 2007), Azospirillum brasilense (Dufreˆne and Rouxhet, 1996), P. aeruginosa (Go´mez-Sua´rez et al., 2002), and Lactobacillus helveticus (Rouxhet et al., 1994) as well as for a mixed population from a laboratory-scale sequencing batch reactor operating on synthetic wastewater (Sun et al., 2009). However, in contrast to our current findings, both protonated and non-protonated nitrogen compounds have been detected in EPS extracted from S. marcescens (Badireddy et al., 2008b) and directly on the surface (bound EPS) of B. subtilis (Ahimou et al., 2007). Additionally, earlier XPS studies have reported P 2p in EPS extracted from pure cultures (Chan et al., 2002; Go´mez-Sua´rez et al., 2002; Omoike and Chorover, 2004; Badireddy et al., 2008b). These results suggest similarities in EPS secreted by dissimilar microorganisms in terms of overall chemical composition (functionalities with respect to C and O) but that subtle differences exist in minor components especially with respect to N and P. Several trends can be observed in Table 1, which summarizes all C 1s and O 1s peaks. The total of O and N-containing compounds increased during the exponential growth phase but decreased during the stationary phase. Protein, alcohol (C286.2), and uronic acid (C289.2) contents of EPS increased over time. Hydrocarbon-like compounds (C284.8) decreased over time. Overall EPS acidity (C289.2 and O531.3) due to contributions from uronic acids and carboxylate groups increased sharply during exponential growth and decreased (or remained constant) during stationary phase due to increased contributions from hemiacetals, acetals, and alcohols (O532.7) as well as acetylated esters (C287.8 and O532.7). C286.2 and C287.8 values in Table 1 are in the range of values reported for proteins and polysaccharides (Rouxhet et al., 1994; Bruinsma et al., 2001; Latrache et al., 2002; Ahimou et al., 2007). Concentrations of total proteins (CPR), carbohydrates (CPS), uronic acids (CUA), and hydrocarbon-like compounds (CHC)
4512
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 0 5 e4 5 1 6
Table 5 e Effect of growth phase on EPS protein secondary structures from mixed cultures during course of experimentation. Secondary structures Aggregated strands b-Sheet Random coil a-Helix 3-Turn helix Antiparallel b-sheet/ aggregated strands
Wavenumber (cm1)
0 Day (area units)
1 Day (area units)
2 Days (area units)
3 Days (area units)
1625e1610 1640e1630 1645e1640 1657e1648 1666e1659 1680e1695
5.31 e 6.54 e e e
0.08 6.92 11.6 1.13 7.24 8.37
3.37 6.80 1.94 8.32 24.6 1.20
7.66 8.94 2.62 23.1 8.22 0.87
with respect to total carbon were estimated using CPR ¼ 3.57 (N/C); CPS ¼ C286.2 (N/C) CUA; and CHC ¼ 1 CPS CPR CUA (Rouxhet et al., 1994; Dufreˆne et al., 1999; Go´mez-Sua´rez et al., 2002; Ahimou et al., 2007) and summarized in Table 2. These calculations showed that EPS samples were richer in proteins followed by carbohydrates and uronic acids validating colorimetric results. The decrease in hydrocarbon-like compounds from day 1 onwards was probably due to assimilation by microorganisms.
3.2.3.
FTIR spectroscopy
Similar to XPS, the dominant effect observed with FTIR spectra was that absorbance intensities increased over the 3-day period indicating greater amounts of EPS (Fig. 5a). Spectra essentially contained similar features suggesting only minor differences in EPS constituents over time. Spectra showed four predominant bands containing several characteristic functional groups: 1700e1600 cm1 (amide I region), 1500e1300 cm1 (carboxylic group-containing and hydrocarbon-like compounds such as lipids), 1200e900 cm1 (carbohydrates and nucleic acids), and 900e600 cm1 (fingerprint region). The band at 1633 cm1 was specifically assigned to b-structures of proteins. The peaks at 1450 cm1 and 1385 cm1 are mainly from asymmetric bending and symmetric deformations of methyl groups. The less intense peak at 1398 cm1 corresponds to the ysCOO of carboxylate groups attributed to the presence of uronic acids. The weak bands located in the regions 1310e1240 cm1 and 12501220 cm1 were contributions from amide III and yasP] O of POO of phosphodiesters, respectively. The weak vibrations in between 1084 and 1080 cm1 region correspond to ysP]O of POO of phosphodiesters. The band region 1200e900 cm1 corresponds to yCeC, yCeO, dCeOeH, dCeOeC, CeOeP, PeOeP vibrations from carbohydrates and nucleic acids. The fingerprint region between 900 and 600 cm1 exhibited ring vibrations from aromatic amino acids (e.g. Phe, Trp, and Tyr) and nucleotides. Subtle differences in protein secondary structures induced over time were interpreted by resolving the amide I region (1700e1600 cm1) into component peaks using second derivative analysis and curve-fitting procedure (Buijs et al., 1996). Fig. 5b and c depicts representative second derivative spectra and amide I band resolved into secondary structures respectively. Peak areas of O-acetyl esters, amide I proteins, and carbohydrates and nucleic acids (1200e900 cm1) were estimated by choosing appropriate baselines assuming Lorentzian-shaped curves. A summary of peak assignments
based on available vibrational frequencies for model organic compounds, biomolecules, and bacteria is given in Table 3. Similar to previously published FTIR spectra of EPS directly extracted from real-world wastewater treatment plants, Fig. 5a depicts strong vibrational signals for amide I, carbohydrates, as well as for methyl, methylene, and carboxylates in the mixed region (Go¨rner et al., 2003; Garnier et al., 2005) suggesting that our batch experiments captured important characteristics inherent to the activated sludge process. EPS spectra exhibited weak absorbances for amide II (1550 cm1) and amide III (1350e1200 cm1), demonstrating that C]O stretching was predominant compared with CeN stretching, NeH bending, and O]CeN bending. The band centered at 1400 cm1 is due to carboxylate groups, indicating the acidic nature of the EPS components, and is consistent with the XPS results. The presence of O-acetylated carbohydrates was inferred from a band in the region 1740e1720 cm1 (O-acetyl groups) and the broad ester band at 1270e1230 cm1 consistent with their suggested role in cell aggregation, adhesion, and biofilm integrity (Nivens et al., 2001; Sutherland, 2001; Tielen et al., 2005). Weak vibrations from phosphodiester bonds were detected at 951 and 991 cm1 indicating low amounts of nucleic acids, which were undetected in XPS spectra (see Table 3). Peaks at 1205e1209, 1114, 1080e1078, 956e952, and 916 cm1 denote the contribution from extracellular nucleic acids. While their role has not yet been clearly identified, it could serve as an indicator for quality of the sludge in terms of the health of the microbial population and biomass, source of wastewater, and may also influence bioflocculation and settling (Raszka et al., 2006). The peak areas of amide I, carbohydrate (1150e1000 cm1), and O-acetyl ester (1740e1720 cm1) are summarized in Table 4 after normalization by the total band area having two peaks centered at 2960 cm1 and 2930 cm1 corresponding to the methyl and methylene band region (3000e2880 cm1) to remove small variations in sample amount and KBr pellet thickness. Monotonic increases in each of the band areas with time observed in Table 4 are consistent with colorimetric and XPS results for total proteins and carbohydrates. Peak assignments for secondary structures were based on spectral analysis of model peptides and proteins of known structures (Stuart, 2004) and are summarized in Table 5. Initially, EPS was mainly comprised of aggregated strands and random coils. As cells progressed through the exponential phase, as total EPS increased, (i) antiparallel b-sheets/aggregated strands (1680e1695 cm1) were more prevalent than aggregated strands (1625e1610 cm1) likely due to changes in
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 0 5 e4 5 1 6
4513
the inter/intra-molecular hydrogen bonding (Stuart, 2004), and (ii) b-sheets, random coils, and a- and 3-turn helical structures increased. In the stationary phase, (i) aggregated stands with significant infrared absorbances in the 1625e1610 cm1 region increased relative to antiparallel bsheets/aggregated strands, (ii) b-sheets increased only slightly, (iii) random coils were present at low levels, and (iv) the total helical content (sum of a- and 3-turn) was high and remained constant. As observed in Table 5, the measured bsheets content changed only in a small range (6.80e8.94 area units) indicating they were stable irrespective of growth phase. These results demonstrate that the growth phase influenced protein secondary structure contents suggesting variations in bioflocculation ability between the exponential and stationary phase as discussed in Section 3.3. It is emphasized that XPS spectra and protein secondary structures were largely compared with available results from pure cultures due to the paucity of detailed information on activated sludge EPS. Hence, several additional measurements are necessary before our results can be extended to other wastewater treatment systems especially those operating under different nutrient regimes, water quality, and design parameters.
3.3. Relationships between EPS and bioflocculation ability 3.3.1.
Role of total EPS
Fig. 6a illustrates that bioflocculation ability improved with total carbohydrates, proteins, uronic acids, and total EPS (carbohydrates þ proteins þ uronic acids) during the 3 days of experimentation. Reflocculation of microorganisms appeared to be inhibited during the exponential growth phase when EPS content was low and acidic in nature. This is attributed to a combination of reduced polymeric interactions and electrostatic repulsion. During the exponential phase, a higher fraction of uronic acids to total EPS suggests that microorganisms were predominantly negatively charged at the pH of our experiments, thereby reducing their ability to bioflocculate (Tsuneda et al., 2003; Wile´n et al., 2003b). In contrast, bioflocculation ability was enhanced during the stationary growth phase since copious amounts of EPS were present and the fraction of uronic acids with respect to total EPS was low. Better bioflocculation at higher concentrations of proteins and carbohydrates (and total EPS) is attributed to polymeric interactions overwhelming electrostatic interactions in the stationary phase, similar to bacterial adhesion to surfaces (e.g. formation of hooks-like structures from alginates, polymer entanglement, etc.) (van Loosdrecht et al., 1990; Go´mez-Sua´rez et al., 2002; Tsuneda et al., 2003; Wile´n et al., 2003a,b; Tielen et al., 2005; Eboigbodin and Biggs, 2008).
3.3.2. Role of functional groups and protein secondary structures on bioflocculation ability A more detailed analysis of the role of specific functionalities and protein conformation on bioflocculation is presented in Fig. 6b and c. Fig. 6b shows that the presence of Ce(O,N) and O] C eOH þ O] C eOR containing compounds such as proteins, carbohydrates, and alcohols promoted microorganism flocculation whereas compounds with C eðC; HÞ and
Fig. 6 e Effects of overall EPS components, specific carbon functionalities, and protein secondary structures on bioflocculation. Role of carbohydrates, proteins, uronic acids, and total EPS is shown in (a). Detailed analysis of the role of carbon functionalities obtained using XPS and protein secondary structures obtained using FTIR are shown in (b) and (c), respectively.
4514
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 0 5 e4 5 1 6
Fig. 7 e Phase contrast images of typical microbial flocs during 3 days of growth in the presence of glucose.
C ]O þ Oe C eO functionalities such as carbonyl, carboxylate, acetal, or hemiacetal and hydrocarbons hampered bioflocculation. Additionally, certain protein secondary structures e.g. aggregated strands, b-sheets, a- and 3-turn helices promoted bioflocculation and antiparallel b-structures, and random coils decreased bioflocculation ability (Fig. 6c). This is consistent with recent findings that aggregation, adsorption, and biofilm formation have been stimulated by a-helices and reduced in the presence of b-structures (Beech et al., 1999; Omoike and Chorover, 2004). In summary,
protein secondary structures and functional groups were found to be affected by the changes in EPS concentration during growth phase consequently influencing microbial aggregation. Visual evidence also corroborates the role of EPS on flocculation. Fig. 7 depicts typical phase contrast images, which show that in the exponential phase when EPS concentrations were low, microorganisms were loosely connected to each other via a network of EPS resulting in porous, irregular shaped flocs. In contrast, during the stationary growth phase,
Fig. 8 e Scanning electron micrographs showing microbial flocs embedded in a copious EPS matrix.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 0 5 e4 5 1 6
flocs were more compact, denser, and more spherical in the presence of higher EPS amounts. Inspection of flocs at higher magnification using scanning electron microscopy revealed that in all cases, flocs comprised a variety of cocci, rod-shaped, and filamentous microorganisms that were embedded in a rich EPS matrix (two representative images showing different types of microorganisms are shown in Fig. 8).
4.
Conclusions
Results reported herein demonstrate the importance of EPS functional constituents such as O-acetylated and acidic carbohydrates and individual protein secondary structures on bioflocculation in the activated sludge processes. Subtle differences in EPS composition at the molecular level observed between exponential and stationary growth phase in our batch study (e.g. a- and 3-turn helices, (hemi)acetals, alcohols, and acetylated esters were more prevalent during the stationary phase whereas uronic acids, carboxylate groups, b-structures, and random coils were predominant in the exponential phase) may explain differences in bioflocculation with retention time in activated sludge units. Further, even though the current study correlates concentrations of certain EPS moieties with bioflocculation ability, more work is necessary to understand the fundamental mechanisms by which certain secondary structures and functionalized carbohydrates induce flocculation. Also, the choice of substrate (glucose in this manuscript) can potentially influence aggregation kinetics. Finally, these results provide the basis for designing longer term laboratory experiments using continuous flow reactors to study the role of retention time and microbial activity on bioflocculation in oligotrophic environments.
Acknowledgments We appreciate the assistance of Mr. Gurdip Hyare, Managing Engineer of the City of Houston’s Wastewater Operations Branch during activated sludge sampling. This research has been funded by grants from the National Science Foundation CAREER Program (CBET-0134301) and the Texas Hazardous Waste Research Center. A portion of the research was performed using EMSL, a national scientific user facility sponsored by the Department of Energy’s Office of Biological and Environmental Research located at Pacific Northwest National Laboratory. The contents do not necessarily reflect the views and policies of the sponsors nor does the mention of trade names or commercial products constitute endorsement or recommendation for use.
Appendix. Supplementary data The supplementary data associated with this article can be found in the on-line version at doi:10.1016/j.watres.2010.06. 024.
4515
references
Ahimou, F., Boonaert, C.J.P., Adriaensen, Y., Jacques, P., Thonart, P., Paquot, M., Rouxhet, P.G., 2007. XPS analysis of chemical functions at the surface of Bacillus subtilis. Journal of Colloid and Interface Science 309, 49e55. APHA, AWWA, WEF, 1995. Standard Methods for the Examination of Water and Wastewater. American Public Health Association, Baltimore, MD. Badireddy, A.R., Chellam, S., Yanina, S., Gassman, P.L., Rosso, K. M., 2008a. Bismuth dimercaptopropanol (BisBAL) inhibits the expression of extracellular polysaccharides and proteins by Brevundimonas diminuta: implications for membrane microfiltration. Biotechnology and Bioengineering 99 (3), 634e643. Badireddy, A.R., Korpol, B., Chellam, S., Gassman, P.L., Engelhard, M.H., Lea, A.S., Rosso, K.M., 2008b. Spectroscopic characterization of extracellular polymeric substances from Escherichia coli and Serratia marcescens: suppression using subinhibitory concentrations of bismuth thiols. Biomacromolecules 9, 3079e3089. Barth, A., Zscherp, C., 2002. What vibrations tell us about proteins. Quarterly Reviews of Biophysics 35 (4), 369e430. Beamson, C., Briggs, D., 1992. High Resolution XPS of Organic Polymers: the Scienta ESCA300 Database. John Wiley & Sons, Chichester, UK. Beech, I., Hanjagsit, L., Kalaji, M., Neal, A.L., Zinkevich, V., 1999. Chemical and structural characterization of exopolymers produced by Pseudomonas sp. NCIMB 2021 in continuous culture. Microbiology 145, 1491e1497. Bossier, P., Verstraete, W., 1996. Triggers for microbial aggregation in activated sludge? Applied Microbiology and Biotechnology 45, 1e6. Bruinsma, G.M., van der Mei, H.C., Busscher, H.J., 2001. Bacterial adhesion to surface hydrophilic and hydrophobic contact lenses. Biomaterials 22 (24), 3217e3224. Buijs, J., Norde, W., Lichtenbelt, J.W.T., 1996. Changes in the secondary structures of adsorbed IgG and F(ab0 )2 studied by FTIR spectroscopy. Langmuir 12, 1605e1613. Chan, K.-Y., Xu, L.-C., Fang, H.H.P., 2002. Anaerobic electrochemical corrosion of mild steel in the presence of extracellular polymeric substances produced by a culture enriched in sulfate-reducing bacteria. Environmental Science & Technology 36, 1720e1727. Dufreˆne, Y.F., Boonaert, C.J.P., Rouxhet, P.G., 1999. Surface Analysis by X-ray Photoelectron Spectroscopy in Study of Bioadhesion and Biofilms. Academic Press, London, UK. Dufreˆne, Y.F., Rouxhet, P.G., 1996. X-ray photoelectron spectroscopy analysis of the surface composition of Azospirillum brasilense in relation to growth conditions. Colloids and Surfaces B: Biointerfaces 7, 271e279. Eboigbodin, K.E., Biggs, C.A., 2008. Characterization of the extracellular polymeric substances produced by Escherichia coli using infrared spectroscopic, proteomic, and aggregation studies. Biomacromolecules 9 (2), 686e695. Eboigbodin, K.E., Ojeda, J.J., Biggs, C.A., 2007. Investigating the surface properties of Escherichia coli under glucose controlled conditions and its effect on aggregation. Langmuir 23, 6691e6697. Frølund, B., Griebe, T., Nielsen, P.H., 1995. Enzymatic activity in the activated-sludge floc matrix. Applied Microbiology and Biotechnology 43, 755e761. Frølund, B., Palmgren, R., Keiding, K., Nielsen, P.H., 1996. Extraction of extracellular polymers from activated sludge using a cation exchange resin. Water Research 30 (8), 1749e1758. Garnier, C., Go¨rner, T., Lartiges, B.S., Abdelouhab, S., de Donato, P., 2005. Characterization of activated sludge exoploymers from
4516
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 0 5 e4 5 1 6
various origins: a combined size-exclusion chromatography and infrared microscopy study. Water Research 39, 3044e3054. Geesey, G.G., Van Ommen Kloeke, F., 2005. Extracellular enzymes associated with microbial flocs from activated sludge of wastewater treatment systems. In: Droppo, I.G., Leppard, G.G., Liss, S.N., Milligan, T.G. (Eds.), Flocculation in Natural and Engineered Environmental Systems. CRC Press, Boca Raton, pp. 295e315. Go´mez-Sua´rez, C., Pasma, J., van der Borden, A.J., Wingender, J., Flemming, H.-C., Busscher, H.J., van der Mei, H.C., 2002. Influence of extracellular polymeric substances on deposition and redeposition of Pseudomonas aeruginosa to surfaces. Microbiology 148 (4), 1161e1169. Go¨rner, T., de Donato, P., Ameil, M.-H., Montarges-Pelletier, E., Lartiges, B.S., 2003. Activated sludge exopolymers: separation and identification using size exclusion chromatography and infrared micro-spectroscopy. Water Research 37, 2388e2393. Higgins, M.J., Novak, J.T., 1997. Characterization of exocellular protein and its role in bioflocculation. Journal of Environmental Engineering (New York, N.Y.) 123 (5), 479e485. van den Hoogen, B.M., Rene´ van Weeren, P., Lopes-Cardozo, M., van Golde, L.M.G., Barneveld, A., van de Lest, C.H.A., 1998. A microtiter plate assay for the determination of uronic acids. Analytical Biochemistry 257, 107e111. Jorand, F.J., Guicherd, P., Urbain, V., Manem, J., Block, J.C., 1994. Hydrophobicity of activated sludge flocs and laboratory-grown bacteria. Water Science and Technology 30 (11), 211e218. Kalaji, M., Neal, A.L., 2000. IR study of self-assembly of capsular exopolymers from Pseudomonas sp. NCIMB 2021 on hydrophilic and hydrophobic surfaces. Biopolymers 57 (1), 43e50. Latrache, H., El Ghmari, A., Karroua, M., Hakkou, A., Mousse, H.A., El Bouadili, A., Bourlioux, P., 2002. Relations between hydrophobicity tested by three methods and surface chemical composition of Escherichia coli. New Microbiologica 25, 75e82. Li, X.Y., Yang, S.F., 2007. Influence of loosely bound extracellular polymeric substances (EPS) on the flocculation, sedimentation and dewaterability of activated sludge. Water Research 41, 1022e1030. Lin-Vien, D., Colthup, N.B., Fateley, W.G., Grasselli, J.G., 1991. The Handbook of Infrared and Raman Characteristic Frequencies of Organic Molecules. Academic Press, Inc, San Diego. van Loosdrecht, M., Norde, W., Lyklema, L., Zehnder, J., 1990. Hydrophobic and electrostatic parameters in bacterial adhesion. Aquatic Sciences 51, 103e114. Magbanua Jr., B.S., Bowers, A.R., 2006. Characterization of soluble microbial products (SMP) derived from glucose and phenol in dual substrate activated sludge bioreactors. Biotechnology and Bioengineering 93 (5), 862e870. Maquelin, K., Kirschner, C., Choo-Smith, L.-P., Ngo-Thi, N.A., Naumann, D., Puppels, G.J., 2002. Vibrational Spectroscopic Studies of Microorganisms. John Wiley & Sons, Chichester. Naumann, D., Helm, D., Labischinski, H., Gieshbrecht, P., 1991. The characterization of microorganisms by Fourier-transform infrared spectroscopy (FT-IR). In: Nelson, W.H. (Ed.), Modern Techniques for Rapid Microbiological Analysis. VCH, New York, pp. 43e96. Ni, B.-J., Fang, F., Xie, W.-M., Sun, M., Sheng, G.-P., Li, W.-H., Yu, H.-Q., 2009. Characterization of extracellular polymeric substances produced by mixed microorganisms in activated sludge with gel-permeating chromatography, excitationeemission matrix fluorescence spectroscopy measurement and kinetic modeling. Water Research 43 (5), 1350e1358. Nivens, D.E., Ohman, D.E., Williams, J., Franklin, M.J., 2001. Role of alginate and its O acetylation in formation of Pseudomonas
aeruginosa microcolonies and biofilms. Journal of Bacteriology 183 (3), 1047e1057. Omoike, A., Chorover, J., 2004. Spectroscopic study of extracellular polymeric substances from Bacillus subtilis: aqueous chemistry and adsorption effects. Biomacromolecules 5, 1219e1230. Omoike, A., Chorover, J., 2006. Adsorption to goethite of extracellular polymeric substances from Bacillus subtilis. Geochimica et Cosmochimica Acta 70, 827e838. Parikh, S.J., Chorover, J., 2006. ATR-FTIR spectroscopy reveals bond formation during bacterial adhesion to iron oxide. Langmuir 22 (20), 8492e8500. Park, C., Novak, J.T., Helm, R.F., Ahn, Y.-O., Esen, A., 2008. Extraction of the extracellular proteins in full-scale activated sludges. Water Research 42, 3879e3889. Ras, M., Girbal-Neuhauser, E., Paul, E., Spe´randio, M., Lefebvre, D., 2008. Protein extraction from activated sludge: an analytical approach. Water Research 42 (8e9), 1867e1878. Raszka, A., Chorvatova, M., Wanner, J., 2006. The role and significance of extracellular polymers in activated sludge. Part I: literature review. Acta Hydrochimica et Hydrobiologica 35 (5), 411e424. Rouxhet, P.G., Mozes, N., Dengis, P.B., Dufreˆne, Y.F., Gerin, P.A., Genet, M.J., 1994. Application of X-ray photoelectron spectroscopy to microorganisms. Colloids and Surfaces B: Biointerfaces 2 (1e3), 347e369. Schmitt, J., Flemming, H.-C., 1998. FTIR-spectroscopy on microbial and material analysis. International Biodeterioration & Biodegradation 41, 1e11. Sobeck, D.C., Higgins, M.J., 2002. Examination of three theories for mechanims of cation-induced bioflocculation. Water Research 36, 527e538. Stuart, B.H., 2004. Infrared Spectroscopy: Fundamentals and Applications. John Wiley & Sons Ltd, Chichester, West Sussex. Sun, X.-F., Wang, S.-G., Zhang, X.-M., Chen, J.P., Li, X.-M., Gao, B.-Y., Ma, Y., 2009. Spectroscopic study of Zn2þ and Co2þ binding to extracellular polymeric substances (EPS) from aerobic granules. Journal of Colloid and Interface Science 335, 11e17. Sutherland, I.W., 2001. Exopolysaccharides in biofilms, flocs and related structures. Water Science and Technology 43 (6), 77e86. Tielen, P., Strathmann, M., Jaeger, K.-E., Flemming, H.-C., Wingender, J., 2005. Alginate acetylation influences intial surface colonization by mucoid Pseudomonas aeruginosa. Microbiological Research 160, 165e176. Tsuneda, S., Aikawa, H., Hayashi, H., Yuasa, A., Hirata, A., 2003. Extracellular polymeric substances responsible for bacterial adhesion onto solid surface. FEMS Microbiology Letters 223, 287e292. Wile´n, B.-M., Nielsen, J.L., Keiding, K., Nielsen, P.H., 2000. Influence of microbial activity on the stability of activated sludge flocs. Colloids and Surfaces B: Biointerfaces 18, 145e156. Wile´n, B.-M., Jin, B., Lant, P., 2003a. The influence of key chemical constituents in activated sludge on surface and flocculating properties. Water Research 37, 2127e2139. Wile´n, B.-M., Jin, B., Lant, P., 2003b. Relationship between flocculation of activated sludge and composition of extracellular polymeric substances. Water Science and Technology 47 (12), 95e103. Wingender, J., Neu, T.R., Flemming, H.-C., 1999. Microbial Extracellular Polymeric Substances: Characterization, Structure and Function. Springer, Berlin, Heidelberg. Zita, A., Hermansson, M., 1997. Effects of bacterial cell surface structure and hydrophobicity on attachment to activated sludge flocs. Applied and Environmental Microbiology 63, 1168e1170.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 3 7 e4 4 5 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Toxicity removal efficiency of decentralised sequencing batch reactor and ultra-filtration membrane bioreactors Giovanni Libralato*, Annamaria Volpi Ghirardini, Francesco Avezzu` Environmental Sciences Department, University of Venice Ca` Foscari, Campo della Celestia 2737/b, I-30122 Venice, Italy
article info
abstract
Article history:
As a consequence of the Water Framework Directive and Marine Strategy Framework
Received 31 March 2010
Directive, there is now more focus on discharges from wastewater treatment plants both to
Received in revised form
transitional and marine-coastal waters. The constraint to encourage sustainable water
31 May 2010
policy to prevent water deterioration and reduce or stop discharges has entailed new
Accepted 4 June 2010
requirements for existing wastewater treatment plants in the form of advanced waste-
Available online 11 June 2010
water treatment technologies as further suggested by the Integrated Pollution and Prevention Control Bureau. A whole toolbox of physico-chemical and ecotoxicological
Keywords:
parameters to investigate commercial and mixed domestic and industrial discharges was
AS-SBR
considered to check the efficiency of an Activated-Sludge Sequencing Batch Reactor (AS-
UF-MBR
SBR) and two Ultra-Filtration Membrane Biological Reactors (UF-MBRs) on a small scale
Discharge quality
decentralised basis. All discharges were conveyed into Venice lagoon (Italy), one of the
Ecotoxicity
widest impacted Mediterranean transitional environment. The UF-MBRs were able to
Saltwater species
provide good quality effluents potentially suitable for non-potable reuse, as well as reducing specific inorganic micro-pollutants concentration (e.g. metals). Conversely, the AS-SBR showed unpredictable and discontinuous removal abilities. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
The importance has recently been evidenced of producing higher quality treated wastewater within the perspective of zero emissions (OSPAR, 2007) along with the precautionary principle (Harremoe¨s, 2000), not only to protect the receiving water body, but also to further support water recycling and reuse, covering end-of-pipe technologies for treatment of pollution immediately after it has been generated. In the European Union, the Water Framework Directive (2000/60/EC) (WFD, 2000) and the more recent Marine Strategy Framework Directive (2008/56/EC) (MSFD, 2008) suggested the adoption of a sustainable water policy to prevent water deterioration and reduce or stop discharges, emissions and losses of hazardous substances. Treated discharges from WasteWater Treatment Plants
(WWTPs) must comply with Environmental Quality Standards defined under the WFD, entailing new requirements for existing WWTPs in the form of advanced wastewater treatment technologies. Various aspects must therefore be checked before selecting the optimal advanced treatment technology at a specific WWTP, including not only technical and economic values, but also environmental targets (i.e. physical, chemical and ecotoxicological goals) to be met that may play a leading role in the selection process (Høibye et al., 2008). Sustainable development is at the forefront of today’s policy agendas for technology developers who are involved in wastewater treatment. As indicated by the Integrated Pollution and Prevention Control directive (IPPC, 2008), recent Best Available Techniques (BAT) in wastewater management are oriented to water recycling as well as nutrients (N and P) and
* Corresponding author. Tel.: þ39 0412347737/8596; fax: þ39 0415281494. E-mail address:
[email protected] (G. Libralato). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.06.006
4438
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 3 7 e4 4 5 0
organics removal and, potentially, toxicity lowering. Moreover, it is increasingly expected the reduction/removal of growing concern inorganic (e.g. metals) and organic (e.g. pharmaceuticals) micro-pollutants (Verliefde et al., 2007; Abegglen et al., 2009; Santos et al., 2010). The general trend is to make river and sea outflows redundant with the reduction of requirements for large pipes, mostly by supporting onsite treatments and decentralisation procedures (Maurer et al., 2006; Weber et al., 2007). In particular, decentralised on-site wastewater treatment plants are spreading not only in rural and suburban communities, where sewerage systems are not available, but also in industrial, commercial and residential areas where water consumption rates necessitate considering alternative wastewater treatment scenarios to improve economic, social and environmental aspects related to water conservation and reclamation (Bakir, 2001; Ho and Anda, 2004). In Italy, the best example of decentralisation is in the city of Venice. The city has no sewage system due to its geographical situation and historical characteristics, so untreated wastewater has been discharged directly into the surrounding Lagoon. In 1990, policy-makers and local authorities decided to urgently improve water quality and prevent pollution phenomena by requiring on-site WWTPs installation principally for arts and craft businesses, hospitals, tourist-related structures and restaurants. As a consequence of this 4493 WWTPs now exist, mainly septic tanks (80%), even if Activated-Sludge Sequencing Batch Reactor (AS-SBR) (z1%) and Ultra-Filtration Membrane Biological Reactor (UF-MBR) (z1%) facilities are increasing (MAV, 2007). The integrated assessment of wastewater ecotoxicological effects has been recognised to be of major importance besides the physico-chemical characterisation, thus ecotoxicity testing is referred to as a useful way to identify potential environmental impacts to the receiving water environment (Mendonc¸a et al., 2008). Current legislation including the WFD (2000), MSFD (2008), IPPC (2008) and Registration, Evaluation, Authorisation and restriction of CHemicals (REACH, 2006) as well as the Whole Effluent Toxicity (WET) approach (USEPA, 2004) and the Whole Effluent Assessment (WEA) (OSPAR, 2007) indicates that ecotoxicity testing is an integral part of the toolbox to investigate discharges in order to define a realistic assessment and management strategy. The aim of this research was to check the efficiency of two advanced small scale decentralised wastewater treatment technologies, AS-SBR (Celis et al., 2008; Ben et al., 2009) and et al., 2009), to UF-MBR (Nosenzo et al., 2005; Radjenovic increase the physico-chemical and ecotoxicological quality of effluents to be discharged into Venice lagoon, that is one of the widest Mediterranean transitional environment, boosting at the same time the general level of sustainability within the perspective of treated wastewater reclamation and reuse. Both commercial and mixed domestic and industrial (i.e. contaminated by metal and metallic micro-pollutants) wastewater samples were taken into consideration. Saltwater testing species were selected within the most widespread organisms already used in scientific literature for wastewater monitoring as well as required by national and international legislations. Bioluminescent bacteria (Vibrio fischeri) (Gutie´rrez et al., 2002; ISO, 2007; Ricco et al., 2004) and two bivalve
molluscs (Crassostrea gigas and Mytilus galloprovincialis), in order to allow the comparison of their relative sensitivities, were considered for this purpose (USEPA, 1995; RIKZ, 1999; SEPA, 2003; ASTM, 2004; OSPAR, 2007). Finally, traditional physico-chemical parameters were compared to toxicity data elaborated on the basis of the Libralato et al. (in press) scoring system and wastewater toxicity index to provide a whole integrated assessment of samples.
2.
Materials and methods
2.1.
Wastewater treatment plants
This research focused on three on-site decentralised WWTPs (AS-SBR, UF-MBR1 and UF-MBR2) located in Venice (Italy) historical centre, with the Venice lagoon as target receiving water body. The AS-SBR was installed in 1998, whereas the UF-MBR1 and UF-MBR2, in 2004 and 2005, respectively. Specifically, AS-SBR and UF-MBR1 were placed next to San Marco’s square in the core of Venice, whereas UF-MBR2 was sited in Murano island that is a worldwide recognised district for its artistic glass production. AS-SBR and UF-MBR1 treated commercial wastewater characterised by sudden variations in influent load, while UF-MBR2 mixed domestic and metal-rich wastewater. The main specifications of the considered WWTPs are provided in Table 1. In addition, it must be said that all WWTPs are periodically required to manage and dispose excess sludge. The AS-SBR operates on the basis of five sequential steps including feeding, mixing, aerobic reaction, settling and drawing (Metcalf and Eddy Inc., 2003), before the final discharge of treated wastewater as reported in Fig. 1. The UFMBR still works on the principle of the activated-sludge process, but the secondary clarifier is replaced by a UFmembrane filtration system consisting of PolyVinyliDene Fluoride (PVDF) tubular membranes with a 0.12 mm particle cut off (A19, PCI, UK). The UF-MBR1 as shown in Fig. 2 has two interconnected aeration basins (named A and B) working simultaneously with two independent UF units, named A and B, respectively. The retentate is recirculated in the oxidation basin while the permeate is accumulated in the effluent tank, before the final discharge. The UF-MBR2 carries out the treatment process including screening and grinding, denitrification and aerobic oxidation prior to UF on PVDF membranes as reported in Fig. 3. The retentate is recirculated both in the denitrification and oxidation basins, while the permeate is sent to an activated carbon column to further improve effluent quality with special regard to colour and residual trace metals content, before the final discharge. The industrial component of the mixed wastewater was mainly composed of glass factory effluents that were rich in trace metal and metallic species. Before entering the equalisation basin, industrial wastewater was generally physico-chemically pretreated as a first step of a larger multi-purpose plant.
2.2.
Sample collection and handling
Wastewater samples were collected manually according to USEPA (2004) general guidelines. Influent was sampled in
4439
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 3 7 e4 4 5 0
Table 1 e Wastewater treatment plants main characteristics. Specification
Units
3
Values
1
AS-SBR
UF-MBR1
UF-MBR2
General
Daily flow rate Mixed liquor dissolved oxygen MLSS MLVSS Sludge retention time Sludge production Operating temperature Remote control
m day mg l1 g l1 g l1 Day kg MLSS (kg CODi)1 C
120 1.9e3.2 6e8 4e6 30e40 0.4e0.6 15e25 Yes
150 2.4e3.1 8e10 6e8 50e75 0.09e0.12 18e30 Yes
80e120 2.5e5.2 9e23 7e13 150e300 0.06e0.09 16e35 Yes
Denitrification
Basin area Minimum volume Maximum volume Working volume Hydraulic retention time
m2 m3 m3 m3 h
e e e e e
e e e e e
18 e 90a 90 18e27
Reaction/aeration
Basin area Minimum total volume Maximum total volume Working volume Hydraulic retention time
m2 m3 m3 m3 h
32 þ 32 80 92 84 18
38 þ 38 91b 106b 100b 16
73 87 145 109 22e33
i ¼ Influent. a Always operating at the maximum volume. b Total volume.
WWTP feed tanks, whereas effluent was sampled after the final treatment and downstream of all entering wastewater before the final discharge. In the case of UF-MBR2, wastewater samples were also collected immediately after UF-membrane filtration. Every sample was the result of 3 grab samples collected over a period of time not exceeding 6 h and homogenised to obtain a composite sample in order to reduce the variability of wastewater according to a time composite sampling procedure. Non-reactive pre-cleaned polyethylenterephtalate containers were completely filled (1 l), leaving no air-space between the content and the lid. Wastewater samples were cooled to 4 C 1 C for transport from the sampling site to the laboratory to minimise physico-chemical and biological changes. In the laboratory, discrete samples were mixed to produce composite samples. Wastewater aliquots for physical and chemical analyses were not processed further and stored at 4 C 1 C, providing their full characterisation 24e36 h after collecting. Conversely, ecotoxicological evaluations were carried out on salinity adjusted samples (OSPAR, 2007) by means of hypersaline brine addition, in order to simulate the
potential adverse effects on the receiving saltwater environment (USEPA, 1995; Libralato et al., 2009). Samples were named by a combination of the WWTP collection site identification letter (X ¼ AS-SBR, Y ¼ UF-MBR1 or Z ¼ UF-MBR2), the treatment stage (i ¼ influent, e ¼ effluent only for X, p ¼ permeate for both Y and Z and ac ¼ activated carbon only for Z ) and an integer number indicating the sequence in specimen collection. AS-SBR and UF-MBR1 were monitored for 8 weeks (from April to May) consecutively, whereas UF-MBR for 21 weeks (from January to August). Both monitoring periods provided one integrated sample per week considering influent, permeate and effluent after activated carbon filtration on a case-by-case basis.
2.3.
Chemical analyses
The pH was measured via pHmeter HI 9025 Microcomputer (HANNA Instrument, Beverly, MA, USA), the salinity was checked with a refractometer (Atago, Japan) and the Dissolved Oxygen (DO) by a WTW multi-parametric device (Nova Analytics, Weilheim, Germany).
Reaction/ Sedimentation Basin A
Raw domestic Wastewater
Effluent Tank
Feed Tank Reaction/ Sedimentation Basin B
excess sludge disposal
Fig. 1 e Flow chart of AS-SBR plant.
Lagoon of Venice (IT)
4440
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 3 7 e4 4 5 0
Excess sludge disposal UF unit
Aeration basin A
Permeate
A Raw domestic Wastewater
Effluent Tank
Retentate
Feed Tank
UF unit
Aeration basin B
B
Lagoon of Venice (IT)
Permeate
Excess sludge disposal
Fig. 2 e Flow chart of UF-MBR1 plant.
The determination of ionic species, chloride (Cl), nitrite þ (NeNO 2 ), nitrate (NeNO3 ), ammonia (NeNH4 ), phosphate 3 2 (PePO4 ) and sulphate (SeSO4 ), was performed using an Ion Chromatograph (IC) system (column Metrohm Metrosep A Supp 5150 4 mm, Metrohm 761 Compact IC, Switzerland) according to APHA (1998) methods. Chemical Oxygen Demand (COD), Total Kjeldahl Nitrogen (TKN), total phosphorus (PTOT), Mixed Liquor Suspended Solids (MLSS), Mixed Liquor Volatile Suspended Solids (MLVSS) and raw wastewater Suspended Solids (SS) were analysed according to APHA (1998) methods. The determination of metal and metallic elements such as aluminium (Al), arsenic (As), barium (Ba), cadmium (Cd), cobalt (Co), total chromium (Crtot), copper (Cu), iron (Fe), manganese (Mn), nickel (Ni), molybdenum (Sb), selenium (Se), vanadium (V) and zinc (Zn) was carried out according to USEPA (1992) and APHA (1998) methods using an Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES Spectroflame Compact E, Spectro Analytical Instruments, Kleve, Germany). An ICP-OES multi-element standard solution (Merck 10580) was used for calibration and Quality Assurance/Quality Control (QA/QC) procedures. Only UFMBR2 samples were checked for metal and metallic elements due to the origin of the treated wastewater.
2.4.
Toxicity tests
Microtox tests were performed using Gram-negative marine bioluminescent bacteria V. fischeri. The Azur Environmental (1998) 100% protocol was followed using Microtox Model 500 Test System. This protocol allowed measurement of light outputs at a wavelength of 490 nm with readings after 5-, 15 and 30-min time exposure at 15 C 1 C to samples serial dilutions. Specifically, in this study only the 5-min data were taken into account to consider the effects derived from the maximum contact time. The light loss as a consequence of bacteria exposure to the toxic samples was the endpoint. Three replicates were performed for every sample dilution, including the control (dilution water) and reference toxicant. Light emission was recorded and the output data analysed using MicrotoxOmni software Version 1.18 (Azur Environmental, 1998). The bioassays with C. gigas and M. galloprovincialis, based on embryo-larval development abnormalities, were performed according to the methods proposed by ASTM (2004) modified to use gametes pools (Volpi Ghirardini et al., 2005; Libralato et al., in press). Conditioned adult oysters were purchased ready to use from the Guernsey Sea Farm Ltd
Glass factory Wastewater
Retentate
Aerobic reaction basin
Feed Tank
Denitrification
Raw municipal Wastewater
Screening and grinding
Physicochemical treatment
UF unit
Permeate
Activated Carbon column
Excess sludge
Leachate Filter-press
Sludge to disposal
Fig. 3 e Flow chart of UF-MBR2 plant.
Effluent
Lagoon of Venice (IT)
4441
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 3 7 e4 4 5 0
Table 2 e Species-specific Toxicity Scores (TS) organised in five classes for C. gigas and M. galloprovincialis embryotoxicity tests and V. fischeri 5-min luminescence inhibition test modified from Libralato et al. (in press).
Toxicity scores
Test organisms
TS Toxicity classes
V. fischeri 5-min (A)
C. gigas (B)
M. galloprovincialis (C)
S ≥ TL
S > TL
S > TL
50 < S ≤ TL or TU50 < 1.22
50 < S ≤ TL or TU50 < 2.13
50 < S ≤ TL or TU50 < 2.48
Low (1)
1.22 ≤ TU50 < 2.09
2.13 ≤ TU50 < 32.57
2.48 ≤ TU50 < 18.08
Medium (2)
2.09 ≤ TU50 < 15.87
32.57 ≤ TU50 < 105.63
18.08 ≤ TU50 < 41.76
High (3)
TU50 ≥ 15.87
TU50 ≥ 105.63
TU50 ≥ 41.76
Very high (4)
hatchery (Guernsey, UK), while wild mussels were collected during the breeding season (OctobereApril) from the Adriatic sea. Sterile polystyrene micro-plates with lids (Iwaki Brand, Asahi Techno Glass Corporation, Tokyo, Japan) with 24 wells (3 ml) were used as test chambers. Dilution water was artificial seawater reconstituted according to ASTM (2004) at a salinity of 34. Toxicity tests were conducted in triplicate using at least five geometrically-scaled dilutions including the control (dilution water) and reference toxicant as reported in Libralato et al. (in press).
2.5.
Absent (0)
Data analysis and statistics
Microtox EC50 values were obtained by linear regression between sample concentration and the fraction of light loss to light remaining (G) in a logarithmic scale where the EC50 corresponds to the sample concentration matching G ¼ 1 with 95% confidence limits. Bivalves toxicity data were expressed as EC50 values based on the Percentages of Effect (i.e. percentage of not normally developed larvae) (PE). EC50 values with 95% confidence limits
Table 3 e AS-SBR (X) physico-chemical results. Parameters
Units
pH
Sample
i e
AS-SBR (X)
DR 24/08/1995
1
2
3
4
5
6
7
8
7.77 7.89
7.81 7.92
7.91 7.22
7.92 7.61
7.42 7.60
8.06 7.55
7.84 7.61
7.40 7.45
DM 12/06/2003
6.0e9.5
SS
mg l1
i e
176 6
184 272
148 104
216 44
252 68
112 68
125 66
115 100
50%a
10
COD
mg l1
i e
368 11
287 338
357 287
437 209
502 209
225 266
390 256
352 42
75%a
100
Cl
mg l1
i e
40 55
88 70
33 91
38 71
27 76
169 69
37 63
31 93
250
TKN
mg l1
i e
37 3
31 11
37 23
35 27
33 27
37 24
25 22
37 4
15
NeNHþ 4
mg l1
i e
20 1
14 10
13 10
23 12
24 19
22 20
22 18
20 4
2
NeNO 2
mg l1
i e
0.12 0.33
<0.01 <0.01
<0.01 <0.01
<0.01 <0.01
0.06 <0.01
<0.01 <0.01
0.10 <0.01
0.77 <0.01
e
NeNO 3
mg l1
i e
0.20 13.90
0.01 0.03
0.15 <0.01
<0.01 <0.01
1.07 <0.01
1.73 <0.01
<0.01 <0.01
<0.01 15.08
e
PePO3 4
mg l1
i e
1.75 1.77
2.33 1.52
1.56 2.80
0.70 1.34
2.33 2.26
9.88 2.29
1.05 2.37
1.23 1.86
e
PTOT
mg l1
i e
6 2
3 6
4 6
6 3
5 3
6 6
4 5
4 4
10
SeSO2 4
mg l1
i e
8 10
9 8
12 4
6 5
9 4
3 4
9 4
9 11
500
i ¼ Influent, e ¼ effluent. DR 24th August 1995 is about discharge limits from urban individual WWTPs in Venice lagoon. DM 12th June 2003 is about treated wastewater reuse limits for non-potable purposes. a Required decrease in the parameter concentration at the discharge compared to raw wastewater.
4442
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 3 7 e4 4 5 0
Table 4 e UF-MBR1 (Y) physico-chemical results. Parameters Units Sample
UF-MBR1 (Y) 1
2
3
4
7.75 7.89
7.83 7.79
8.00 7.92
7.98 7.85
DR 24/08/1995 DM 12/06/2003
5
6
7
8.12 7.86
7.95 7.79
7.89 7.83
480 <0.001
416 <0.001
8
pH
i p
8.32 7.77
SS
mg l1 i p
392 224 668 376 <0.001 <0.001 <0.001 <0.001
COD
mg l1 i p
769 4
500 7
903 11
797 9
1170 11
1530 6
1344 6
500 7
Cl
mg l1 i p
54 67
31 56
85 234
55 44
50 44
62 60
61 61
56 56
250
TKN
mg l1 i p
87 2
33 3
71 4
64 4
47 5
26 2
36 5
30 3
15
NeNHþ 4
mg l1 i p
17 1.1
33 2.2
14 1.3
37 0.7
20 1.6
19 1.0
22 2.0
24 2.3
2
NeNO 2
mg l1 i p
0.09 <0.01
0.37 0.01
0.13 0.03
<0.01 <0.01
<0.01 <0.01
<0.01 <0.01
<0.01 <0.01
0.42 <0.01
e
NeNO 3
mg l1 i p
0.73 15.61
0.02 12.89
0.19 2.70
<0.01 3.32
0.04 1.86
0.03 0.36
<0.01 0.15
0.06 0.99
e
PePO3 4
mg l1 i p
3.87 3.29
3.86 4.29
4.22 3.05
35.24 4.76
4.97 4.76
13.44 3.46
7.17 3.53
2.38 3.65
e
PTOT
mg l1 i p
9 4
7 5
12 4
44 5
37 4
8 5
8 4
8 4
10
SeSO2 4
mg l1 i p
7 11
10 10
10 16
9 11
9 11
14 12
9 12
10 10
500
560 304 <0.001 <0.001
6.0e9.5 50%a
10
75%a
100
i ¼ Influent, p ¼ permeate. DR 24th August 1995 is about discharge limits from urban individual WWTPs in Venice lagoon. DM 12th June 2003 is about treated wastewater reuse limits for non-potable purposes. a Required decrease in the parameter concentration at the discharge compared to raw wastewater.
were calculated by Trimmed SpearmaneKarber statistical method (ASTM, 2004). Toxic Unit at 50% of the population exhibiting a response (TU50) was determined as 100/EC50 to provide values directly correlated to the toxicity magnitude. The Abbott’s formula (ASTM, 2004) was considered to correct the responses for each treatment due to the effects in control tests. Moreover, in order to test the null hypothesis that the different treatments had no effect on larval development, the percentages of normal larvae at each concentration were compared to the controls using a one-way ANOVA after conducting Cochran’s test for homogeneity of variance. If the data failed this test, an arcsin P½ transformation was applied to the data to achieve homoschedasticity. Toxicity data were elaborated according to Libralato et al. (in press) scoring system based on species-specific Toxicity Scores (TSs) and a final Wastewater Toxicity Index (WTI). The TSs have been defined in relation to (1) a separate-variance t test to verify if there is a significant difference ( p < 0.05) in the mean organism response between the sample and the negative laboratory control and (2) the 90th-percentile of the Minimum Significant Difference (MSD) distribution according to Phillips et al. (2001). V. fischeri, C. gigas and M. galloprovincialis TSs have been displayed in Table 2. The WTI presents a five-class structure, each toxicity class is identified by a colour, a range of scores (0e4z, where z ¼ number of TS available) and a synthetic judgement: absent (blue, 0), low
(green, 1 z), medium (yellow, z þ 1 2z), high (orange, 2z þ 1 3z) and very high (red, 3z þ 1 4z), suggesting, in addition, the timing of the actions to be undertaken to improve the quality of the monitored discharge (from no action to urgency). The WTI is calculated as the sum of single species-specific TS values as follows: WTI ¼ A þ B þ C þ c, where A ¼ 0e4, scoring from V. fischeri 5-min toxicity score, B ¼ 0e4, scoring from C. gigas toxicity score, C ¼ 0e4, scoring from M. galloprovincialis toxicity score, and c is an adjustment coefficient (if A s 0 hence c ¼ 0, whilst if A ¼ 0 and B ¼ {2,3,4} and C ¼ {2,3,4} hence c(B,C) ¼ 2; if only B or C is available, c[(B) or (C)] ¼ 1). The application of WTI was performed considering both all toxicity data (V. fischeri, C. gigas and M. galloprovincialis) and just V. fischeri coupled one time with C. gigas and one time with M. galloprovincialis to observe how the presence of one or more than one sub-chronic endpoint as an index component would influence the final output. The relationships between variables and the variation present in the dataset matrix were accounted via biplotting both the ordination component scores and the variable loading coefficients through Principal Component Analysis (PCA) based on the Pearson’s correlation matrix, in order to identify the major discriminating variables associated with a given principal component. Normality of data and homogeneity of variance were previously checked. XLSTAT software, version 2008.4.01, a data analysis and statistical
2 2 2 2 2 3 0 3 3 4 3 2 2 3 3 1 3 3 2 3 3 3 - 3 0 2 1 1 2 3 0 0 0 0 1 0 0 1 0 0 1 0 2 4 4 2 3 2 1 2 2 0 0 0 0 1 0 0 0 0 0 1 0 i = influent; e = effluent; p = permeate; ac = after activated carbon
X = AS-SBR; Y = UF-MBR1; Z = UF-MBR2
i 1 0 3 2 1 2 0 3 3 3 3 1 3 3 3 3 4 4 0 4 3 p 1 - 1 1 0 0 1 0 1 1 0 1 0 1 0 1 0 0 1 1 1 ac 1 0 0 1 0 1 0 0 1 1 0 0 0 0 1 1 0 1 0 1 0 Z
2 2 2 2 3 3 0 2 2 4 4 2 3 3 3 3 4 4 2 4 3 3 - 3 0 2 2 2 2 3 0 0 0 0 0 0 0 0 0 2 1 0 2 4 2 1 3 2 2 2 3 2 0 0 1 1 0 0 1 1 0 0 0
-
-
-
-
-
-
-
-
-
-
i 2 2 2 2 1 1 2 2 p 0 0 0 0 0 0 0 0 Y
3 3 3 2 3 3 3 3 0 0 0 0 0 0 0 0
-
-
-
-
-
-
-
-
-
-
-
-
-
3 3 3 4 3 3 3 4 0 0 0 0 0 0 0 0
-
-
-
-
-
-
-
-
-
-
10 11 12 13 14 15 16 17 18 19 20 21 9
2 3 3 2 2 3 2 3 0 2 3 2 3 3 2 1 2 2 2 3 3 3 3 3 1 2 2 2 3 3 3 1 -
-
-
-
-
-
-
-
-
i 2 1 1 2 1 3 2 2 e 0 2 2 2 3 3 2 0 X
-
M. galloprovincialis
8 7 6 5 4 3 2 1 10 11 12 13 14 15 16 17 18 19 20 21
C. gigas
9 8 7 6 5 4 3 2 1 10 11 12 13 14 15 16 17 18 19 20 21 9
V. fischeri 5- min
8 7 6 5 4 3 2 1
Table 5 e Libralato et al. (in press) species-specific toxicity scores assessment to X (AS-SBR), Y (UF-MBR1) and Z (UF-MBR2) wastewater samples.
application available for Microsoft Excel, was used for data elaboration.
3.
Results and discussion
3.1.
AS-SBR vs UF-MBR1
AS-SBR and UF-MBR1 physico-chemical results for raw wastewater and final discharge are provided in Tables 3 and 4, respectively. Moreover, ecotoxicological data are shown in Table 5, as species-specific toxicity score judgements and as WTI in Table 6AeC. Both series of raw commercial wastewater samples presented similar physico-chemical characteristics, except for COD and SS that presented higher values in UFMBR1. Nevertheless, the efficiency of these two treatment facilities could be suitably compared anyway. Indeed, the COD and SS were much better removed by UFMBR1 compared to AS-SBR. UF-MBR1 consistently provided high efficiency levels throughout the monitoring period, reducing COD by 99% and SS by 99.9%, whereas AS-SBR lowered both of them by less than 50% on average, as for ammonia and TKN. Phosphate and total phosphorus discharge concentrations were also improved better by UFMBR1 rather than AS-SBR. The assessment of toxicity data from Tables 5 and 6AeC revealed that AS-SBR and UF-MBR1 presented similar raw wastewater ecotoxicological characteristics, although the latter was slightly more toxic. From Table 5, the raw wastewater toxicity was identified in the range 2e3 and 3e4 for C. gigas and M. galloprovincialis, respectively, whereas in the range 1e3 for V. fischeri. Nevertheless, it was highlighted that UF-MBR1 effluent samples presented no toxicity according to each and every one testing species during all the monitoring period. On the contrary, the AS-SBR discharged effluents presenting toxicity in the range 0e3 (from no toxic to highly toxic). Sometimes, it has been evidenced that the discharged effluent presented the same or higher levels of toxicity than the corresponding untreated wastewater. The integration of species-specific toxicity judgements resulting in WTI, as shown in Table 6AeC, provided the final synthetic values stating the presence or absence of toxicity and its relative estimated magnitude. According to Table 6A summarising the integration of all toxicity data, the AS-SBR was shown to be less efficient than UF-MBR1 in toxicity reduction throughout all the monitoring period, with substantial unpredictable removal rates and some residual toxicity at the discharge (i.e. equal or higher that the influent). Conversely, UF-MBR1 removed toxicity in a continuous and efficient way, supporting the possibility for treated wastewater reclamation and reuse. The comparison of integrated toxicity data from Table 6B and C, where only one sub-chronic endpoint was considered at a time, confirmed the judgements expressed from Table 6A. The correlation analysis between the toxicity results from Table 6A and B indicated that there was no significant difference ( p < 0.01) between C. gigas and M. galloprovincialis sensitivities towards the tested commercial wastewater (X(i,e) 1e8 and Y(i,p)1e8). Moreover, the AS-SBR and UF-MBR1 performances were also compared to regulatory limits about effluent discharge
4444
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 3 7 e4 4 5 0
Table 6A e WTI application to X (AS-SBR), Y (UF-MBR1) and Z (UF-MBR2) samples considering all toxicity data (V. fischeri, C. gigas and M. galloprovincialis).
WTI V. fischeri 5-min + C. giga s + M. galloprovincialis (z = 3) 8
9
X
i 6 6 6 7 6 9 7 8 e 1 6 7 6 9 9 7 2
1
2
3
4
5
6
7
-
10 11 12 13 14 15 16 17 18 19 20 21
-
-
-
-
-
-
-
-
-
-
-
-
Y
i 8 8 8 8 7 7 8 9 p 0 0 0 0 0 0 0 0
-
-
-
-
-
-
-
-
-
-
-
-
-
Z
i 5 6 7 6 6 8 0 8 8 11 10 5 8 9 9 7 11 11 6 11 9 p 7 - 7 1 6 3 4 6 7 1 0 1 0 2 0 1 1 0 3 3 1 ac 5 10 9 4 8 5 3 6 6 3 0 0 1 2 1 1 1 2 0 2 0
X = AS-SBR; Y = UF-MBR1; Z = UF-MBR2
i = influent; e = effluent; p = permeate; ac = after activated carbon z = number of toxicity scores # = adjusted with c = 2
from urban individual WWTPs in Venice lagoon (DR, 1995) and treated wastewater reuse for non-potable purposes (DM, 2003). Both limits have been displayed in the last two columns of Tables 3 and 4. The only two parameters taken into consideration by DR (1995) are SS and COD that are required to be removed from raw wastewater not less than 50% and 75%, respectively. The AS-SBR for both SS and COD did not always guarantee the above-mentioned removal rates, whereas the UF-MBR1 provided an outstanding performance. Considering the DM (2003) about treated wastewater reuse for non-potable purposes, it could be observed that the relative regulatory limits for SS, COD, TKN and NeNHþ 4 were not always respected during the AS-SBR monitoring period. On the contrary, the UF-MBR1 complied with the regulatory limits for effluent reuse, except for samples Yp2,7,8 for NeNHþ 4 that presented a slight greater value than the respective threshold. In order to prevent future similar events, it was suggested the implementation of an activated carbon column (Long et al., 2008).
A biplot summarising the PCA results on chemical data for AS-SBR and UF-MBR1 wastewater samples weighted on WTI values from Table 6A is shown in Fig. 4. The first two principal components accounted for 48.60% and 18.10% of the variation, respectively. Thus 66.69% of the variation can be depicted by a two-axis ordination diagram. The biplot regarding components loadings suggested that the first component (F1) scores are influenced by the values of SS, COD, PTOT, PePO3 4 , TKN and pH with positive loadings on the first axis. The second 3 component (F2) was mainly influenced by NeNHþ 4 and PePO4 concentrations. The ordination plot of component scores present in the F1eF2 biplot, as shown in Fig. 4, clustered wastewater samples in two main groups: all permeates (Yp1e8), Xe1e2 and Xe8 at the bottom left, AS-SBR effluents (Xe3e7) at the top left. Raw wastewater samples are scattered mostly on the right side of the plot, probably due to the high variability of their intrinsic characteristics. In accordance with WTI, the bottom left group consisted of good quality discharges from UF-MBR1, except for
Table 6B e WTI application to X (AS-SBR), Y (UF-MBR1) and Z (UF-MBR2) samples considering only V. fischeri and C. gigas toxicity data.
WTI V. fischeri 5-min + C. giga s (z = 2) 1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21
4 3 3 5 4 6 5 5 1 4 4 4 6 6 5 1 -
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
X
i e
Y
i 5 5 5 4 4 4 5 5 p 0 0 0 0 0 0 0 0
Z
i 3 3 5 4 4 5 0 5 5 7 7 3 6 6 6 6 8 8 3 8 6 p 4 - 4 1 3 2 3 3 4 1 0 1 0 1 0 1 0 0 3 2 1 ac 3 5 3 2 4 3 2 3 4 3 0 0 1 1 1 1 1 2 0 1 0
X = AS-SBR; Y = UF-MBR1; Z = UF-MBR2
i = influent; e = effluent; p = permeate; ac = after activated carbon z = number of toxicity scores # = adjusted with c = 1
4445
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 3 7 e4 4 5 0
Table 6C e WTI application to X (AS-SBR), Y (UF-MBR1) and Z (UF-MBR2) samples considering only V. fischeri and M. galloprovincialis toxicity data.
WTI V. fischeri 5-min + M. galloprovincialis (z = 2) 1
2
3
4
5
6
7
8
9
10 11 12 13 14 15 16 17 18 19 20 21
4 4 4 4 3 6 4 5 0 4 5 4 6 6 4 1 -
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
X
i e
Y
i 5 5 5 6 4 4 5 6 p 0 0 0 0 0 0 0 0
Z
i 3 3 5 4 3 5 0 6 6 7 6 3 5 6 6 4 7 7 3 7 6 p 4 - 4 1 3 1 2 3 4 1 0 1 0 2 0 1 1 0 1 2 1 ac 3 5 5 3 4 3 1 3 3 1 0 0 0 1 1 1 0 1 0 2 0
X = AS-SBR; Y = UF-MBR1; Z = UF-MBR2
i = influent; e = effluent; p = permeate; ac = after activated carbon z = number of toxicity scores # = adjusted with c = 1
Xe2, whereas the top left one had discharges with residual toxicity even after treatment from AS-SBR.
3.1.1.
UF-MBR2
UF-MBR2 physico-chemical data for mixed domestic and industrial influent, permeate and final discharge are summarised in Table 7. UF-MBR2 removed most of the COD (95%) during the biological treatment process, and to a lesser extent by activated carbon filtering, throughout all the monitoring period (1e21). The SS were also always completely removed (99.9%) from the final discharge (<0.01 mg l1), thanks to UFmembrane physical barrier. Regarding N-based compounds, the ionised ammonia, that was reduced on average by 42%, evidenced the existence of two distinct treatment periods efficiency identified by two groups of
Biplot (axis F1 and F2: 66.69 %) 5 Yi4 4
F2 (18.10 %)
3 Xi6 N-NH4+ P-PO4-
2 Xe6 Xe5 Xe3 Xe7 1
Xe4
PTOT Xi8
Yi2 Xi5 Xi1 Xi4 Xi7
TKN
Yi5
0 -1 -2
Yp2 Yp8 Yp5 Yp4 Yp7 Yp6 Yp1 Xe8 Xe1
Xi2
pH
Xe2
Xi3 COD SS S-SO4-
Yp3 -5 -4 -3 -2 -1
Yi1 Yi8
0
1
2
3
Yi7 Yi6 Yi3 4
5
6
7
8
9
10 11 12
F1 (48.60 %)
Fig. 4 e Principal component analysis biplot of chemical data with loadings and scores in the coordinates of the first two principal components (F1 and F2) weighed on WTI values according to Libralato et al. (in press) for X (AS-SBR) and Y (UF-MBR1) commercial wastewater considering influent (i), effluent (e, only for X) and permeate (p, only for Y) samples.
samples: Z(p,ac)1e9 and Z(p,ac)10e21. This was probably due to the multi-purpose plant start-up period that occurred exactly in the 1e9 sampling. During the second period (Z(p,ac)10e21), NeNHþ 4 concentration was reduced more efficiently (79%) than in the first one. The TKN concentration indicated that organic nitrogen was more than halved (63% average removal), even though the efficiency was lower (49%) during the first period (Z (p,ac)1e9) than in the second (73%). The concentration of metal and metallic contaminants presented the same trend as ammonia especially for Al, Ba, Cd, Mn, Ni and Zn, whereas the concentrations of As, Cr, Sb, Se and V were frequently under the relative detection limit values. The values of Co and V are not reported because their values were always below the relative detection limits, <7 mg l1 and <2 mg l1, respectively. In particular, Cd and Ni concentrations were higher in the final effluent after activated carbon filtering than in the raw wastewater in the Zac1e9 period because, as discovered subsequently, the activated carbon filter was saturated and required backwashing. Indeed, Cd and Ni residual concentration in the following period, Zac10e21, after activated carbon cleaning, was significantly improved, as also occurred for Al, Ba, Fe, Mn and Zn. The toxicity data elaborated according to Libralato et al. (in press) for mixed domestic and industrial wastewater samples were shown in Table 5 and Table 6AeC, as species-specific toxicity score results and WTI values, respectively. As for the chemical data, all three toxicity scores and WTI identified the existence of two distinct efficiency periods, summarising all chemical instances at one time. In particular, it was shown that after the starting up period (Z1e9), the UF-MBR2 provided very good quality effluents considering both permeate after UF filtration and permeate after activated carbon filtration, although the general toxicity level of untreated wastewater increased till the maximum upper value. On the basis of Table 5, untreated wastewater ranged between scores 0 and 4 for all toxicity bioassays considered. Scores for the starting up period (Z1e9) for both permeate and permeate samples after activated carbon filtration were particularly high, showing that sometimes the effluent toxicity was greater than the relative untreated wastewater specimen, thus correlating with chemical data presented above.
Parameters
Units Sample
1
pH
SS
i p ac mg l
1
i p ac
COD
mg l1 i p ac 1
mg l
TKN
mg l1 i
NeNO 2
mg l
mg l
1
1
i p ac
2
6.45 7.57 e 7.10 e
8.02
e
34 56 85
e
214
4 8.40 7.10 6.17 9800
<0.01 <0.01
<0.01 <0.01
6
7
8
9
8.37 6.94 7.83
7.63 6.28 7.02
7.88 6.71 7.18
8.31 7.70 6.41
8.70 6.06 6.65
516 <0.01 <0.01
204 <0.01 <0.01
76 <0.01 <0.01
238 <0.01 <0.01
430 2103 2032
13
20
149
54
23
16
64
51
16
18 10
2 7
32 30
14 13
12 10
27 25
27 14
e
i p ac
26 10.8 11.5
i p ac
<0.01 <0.01 <0.01 e <0.01 <0.01
8 e 14.8
283 46 30
277 31 13
640 47 15
437 1499 878 1668 <0.01 1648
2845 928 1015
2222 2232 2144
269 1268 1299
<0.01 <0.01
3320 e 2171
18 13
370 44 21
196
9
35
1092 33 15
5
146 25 19
e
p ac
1 NeNO i 3 mg l
7.89 7.81 7.88
7.94
<0.01 e <0.01 <0.01 216 24 11
3
336 46 16
10 8.45 7.99 7.10 238 <0.01 <0.01 628 65 25
11 8.46 7.73 7.63 242 <0.01 <0.01 906 76 45
12 8.30 7.45 7.20 324
13 8.47 7.89 7.68 104
<0.01 <0.01 <0.01 <0.01 563 39 28
472 55 32
86 <0.01 <0.01
<0.01 <0.01 <0.01 <0.01 <0.01 <0.01
69
82
30
7 5
7 1
6 2
<0.01 <0.01 49 <0.01 <0.01
417 85 <0.01 998 <0.01 1010
53
571 41 27 222 <0.01 <0.01
1076 <0.01 <0.01 1024 29 17
94 <0.01 <0.01 512 40 27
8.41 7.70 7.63 208
18 8.30 7.45 6.59 234
<0.01 <0.01 <0.01 <0.01 577 36 26
532 36 30
19 7.86 8.24 7.41 732 <0.01 <0.01 943 10 11 493 <0.01 <0.01
20 7.42 8.24 7.50 202 <0.01 <0.01 519 32 24 73 <0.01 <0.01 63
68
5 5
5 2
5 4
65 2.5 2.1
91 2.1 1.4
72 1.5 2.6
53 1.7 1.9
60 2.8 2.4
1 0.9 3.0
61 5.2 1.9
62 1.2 2.2
2
2
<0.01 <0.01 <0.01
0.3
e
0.34 <0.01 <0.01
0.46 <0.01 <0.01
0.30 <0.01 <0.01
0.70 <0.01 0.10 <0.01 <0.01 <0.01
<0.01 <0.01 <0.01
<0.01 <0.01 <0.01
<0.01 <0.01 <0.01
0.47 0.00 <0.01
0.40 0.10 0.09
<0.01 <0.01 <0.01
<0.01 <0.01 <0.01
<0.01
<0.01
0.33 <0.01
2.88 4.38
5.19 6.37
<0.01 <0.01
2.86 3.52
PePO3 mg l1 i 4 p ac
<0.01 <0.01 <0.01 e <0.01 <0.01
<0.01 <0.01 <0.01
<0.01 <0.01 <0.01
<0.01 <0.01 <0.01
<0.01 <0.01 <0.01
<0.01 <0.01 <0.01
3.67 <0.01 <0.01
2.54 <0.01 <0.01
4.05 <0.01 <0.01
mg l1 i p ac
<0.01 <0.01 <0.01 e <0.01 <0.01
<0.01 <0.01 <0.01
<0.01 <0.01 <0.01
<0.01 <0.01 <0.01
<0.01 <0.01 <0.01
<0.01 <0.01 <0.01
5.64 <0.01 <0.01
3.20 <0.01 <0.01
4.06 <0.01 <0.01
15
2
7 2
<0.01 <0.01 <0.01
0.15
10
73
<0.01 <0.01 <0.01
1.38 1.87
250
2 5
<0.01 1.85 1.92
<0.01
300
71
<0.01 <0.01 <0.01
8.05 7.55
110 <0.01 <0.01
2 3
<0.01 <0.01 <0.01
<0.01
100
118
0.93 <0.01 <0.01
4.51 2.81
120
1 1
28 2.2 1.6
12.15
525 20 10
226
<0.01 <0.01 <0.01
0.06 1.78
10
5 5
53 5.2 1.8
0.15
35
<0.01 <0.01
83
39 22.7 11.5
30.10 30.35
324
<0.01
<0.01
<0.01
<0.01
0.31
<0.01
<0.01
2.55 1.41
1.88 1.72
2.47 1.95
<0.01 <0.01
3.27 4.90
0.13 0.14
3.45 <0.01
<0.01 <0.01
<0.01 <0.01
3.50 <0.01 <0.01
<0.01 3.09 <0.01 0.77 <0.01 <0.01
4.72 <0.01 0.07
<0.01 <0.01 0.06
1.83 0.17 <0.01
3.39 <0.01 <0.01
4.00 <0.01 <0.01
<0.01 <0.01 <0.01
11.40 0.44 1.40
4.61 0.45 4.04
4.72 0.43 1.43
3.56 1.17 1.12
6.0e9.0 6.0e9.5
88 91 <0.01 <0.01 <0.01 <0.01
47 22.1 23.5
<0.01
7.89 8.00 8.10
114 <0.01 <0.01
3 10.1 11.0
10.60 11.20
21
619 <0.01 <0.01
4 12.9 11.2
16.14
1
<0.01 <0.01
8.59 8.02 7.48
17
30 29.0 28.8
4.02
SeSO2 mg l1 i 4
212
8.37 7.62 7.12
16
31 1.0 0.8
13.01 e 20.60 12.73
PTOT
8.77 6.86 7.63
15
7 9.1 9.2
p ac
0.66
14
DM 30/07/ DM 12/ 1999 06/2003
5.26 0.07 0.88
1.60 0.042 0.23
4.72 0.02 0.09
5.71 0.15 0.45
4.80 0.52 0.72
3.68 3.33 <0.01 0.051 0.02 <0.01 3.99 0.36 0.64
4.00 0.28 0.28
0.312 e
87
22
21
16
26
38
15
12
12
10
21
22
11
7
11
12
14
6
6
21
p ac
175 e <0.01 54
53 48
39 41
34 34
26 26
34 35
28 29
23 25
27 27
<0.01 <0.01
22 22
20 21
18 18
19 19
17 17
22 22
21 21
27 29
27 29
33 33
e
0.5
e
1
10
500
500
Al
mg l1
i p ac
e e e
151 e 12276
247 10296 10813
6674 31 406
81 207 6941
164 5 117
632 47 45
852 3 2185
1037 754 1440
1129 5 315
1507 5 54
430 3 65
67 9 47
1153 2 62
202 8 17
1054 10 23
1451 15 38
894 12 36
304 3 12
1047 17 48
1171 31 46
e
1000
As
mg l1
i p ac
e e e
5
6 <1 <1
10 <1 <1
5 <1 <1
5 <1 <1
6 <1 <1
<1 <1 <1
5 <1 <1
5 <1 <1
<1 <1 <1
8 <1 <1
8 <1 <1
7 <1 <1
16 <1 <1
5 <1 <1
5 <1 <1
5 <1 <1
9 <1 <1
10 <1 <1
11 <1 <1
1
20
e
i
e
91
161
1032
80
111
307
127
113
75
105
122
71
83
49
116
104
790
183
62
219
e
10,000
p ac
e e
126
162 109
31 28
57 50
42 30
75 506
44 51
32 193
40 29
41 42
52 46
35 40
210 49
74 92
68 66
60 63
51 43
59 62
46 48
55 53
Ba
mg l1
<1
e
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 3 7 e4 4 5 0
Cl
NeNHþ 4
UF-MBR2 (Z)
4446
Table 7 e UF-MBR2 (Z) physico-chemical results.
Cd
Crtot
Cu
Fe
mg l1
mg l
1
1
mg l
1
mg l
1
mg l
Ni
mg l1
Sb
Se
Zn
mg l1
mg l1
mg l1
e e e
e
<1
<1 11 12
4 <1 <1
<1 <1 4
<1 <1 <1
1 <1 2
1 <1 4
2 <1 4
<1 <1 <1
<1 <1 <1
<1 <1 <1
<1 <1 <1
<1 <1 <1
<1 <1 <1
<1 <1 <1
<1 <1 <1
<1 <1 <1
1 <1 <1
<1 <1 <1
1 <1 <1
1
5
i p ac
e e e
e
50 <1 <1
1 <1 <1
2 <1 <1
5 <1 <1
25 <1 <1
4 <1 <1
4 <1 <1
10 <1 <1
4 <1 <1
1 <1 <1
4 <1 <1
6 <1 <1
6 1 <1
4 <1 <1
4 <1 <1
5 <1 <1
3 <1 <1
9 <1 <1
1
100
1
5 <1 <1
i
e
p ac
e e
22
8
67
7
14
30
50
39
43
35
88
14
41
18
39
32
42
20
37
87
50
1000
<2
8 <2
<2 <2
8 <2
<2 <2
<2 <2
9 <2
<2 9
<2 <2
<2 <2
<2 <2
<2 <2
<2 <2
<2 <2
<2 <2
<2 <2
<2 <2
<2 <2
<2 <2
<2 <2
i p ac
e e e
e
28835 27 22
700 33 21
1990 13 12
7963 13 2
3620 19 8
1206 14 124
293 28 20
385 40 25
1096 16 10
381 16 6
523 4 6
450 18 8
275 13 6
1152 16 6
320 11 3
6978 10 9
800 21 5
7311 16 9
500
2000
41
1352 96 39
i p ac
e e e
70 e 2397
27 2039 1896
113 127 157
84 <2 385
115 37 51
166 694 1422
50 157 309
87 117 1162
20 62 72
22 4 15
94 <2 11
66 3 4
26 <2 3
407 16 80
22 14 47
24 <2 23
22 32 6
79 <2 44
18 34 28
24 29 27
500
200
i
e
p ac
e e
18
7
21
17
19
17
6
15
4
3
12
8
4
9
4
4
4
19
4
7
100
200
162
117 150
13 17
19 77
14 15
29 30
19 110
61 184
24 29
31 35
31 34
11 18
34 15
25 20
28 22
18 21
16 17
28 17
12 16
17 19
i p ac
e e e
20 <5 <5
<5 7 <5
<5 <5 <5
<5 <5 <5
<5 <5 <5
20 <5 <5
20 <5 <5
20 <5 <5
20 <5 <5
24 11 <5
20 12 <5
20 <5 <5
21 <5 <5
20 15 <5
20 <5 <5
21 <5 <5
20 <5 <5
<5 <5 <5
<5 13 11
e
e
i p ac
e e e
41 <5 <5
<5 <5 <5
<5 <5 <5
<5 <5 <5
<5 <5 8
10 <5 <5
41 <5 9
10 <5 <5
10 11 14
12 10 8
10 12 15
10 <5 14
10 <5 7
10 11 7
10 13 13
10 18 15
10 14 16
<5 15 19
<5 17 18
10
10
i
237
134
p ac
e e
81
1157
46
36
263
105
207
104
191
95
46
104
45
108
122
121
95
75
236
250
500
e 10978
7260 9141
80 386
146 1779
22 98
175 565
20 1620
695 1643
22 193
14 72
21 62
9 40
63 41
17 56
12 42
14 33
4 7
9 21
3 15
8 17
13 3
e
200
e
<5 e <5 <5 e <5
i ¼ Influent, p ¼ permeate, ac ¼ after activated carbon. DM 30th July 1999 is about discharge limits in Venice lagoon. DM 12th June 2003 is about treated wastewater reuse limits for non-potable purposes.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 3 7 e4 4 5 0
Mn
i p ac
4447
4448
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 3 7 e4 4 5 0
Biplot (axis F1 and F2: 58.28 %) 22
Zac2
20
Zac3 Zp3 Zn Ni MnCo
18
Cd Al
16
F2 (27.08 % )
14 12 10 8 6 4 2 0 -2 -4 -6 -8 -10
Zac9
Zi4 Ba
Cl-
SS Fe
Zac5 Zac8 Zac7
Cr
Zp7 Zp9 Zp5 Zi2 Zp14 Zi6
Zi7 Cu
Zi5 Zi21 V Zac4 Zi19 Zi18 Zp4 Zp8 Zp6 Zac6 Zp10 Zp15 As COD Zi8 Zp11 Zi20 Zac1 Zp17 Zp12 Zp19 Zi11 Zp1 pH Zp18 Zi9 Zi12 N-NH4+ Zp20 Zp16 Zp13 Zp21 Zi3 Zi14 Zi17 Zi10 Zi16 Zi15 CH3COOZi13 Se Sb
-5
0
5
10
15
20
25
30
F1 (31.21 %)
Fig. 5 e Principal component analysis biplot of chemical data with loadings and scores in the coordinates of the first two principal components (F1 and F2) weighed on WTI values according to Libralato et al. (in press) for Z (UFMBR2) mixed metal and industries wastewater considering influent (i), permeate (p) and permeate after activated carbon filtration (ac) samples.
The integration of all species-specific toxicity judgements resulting in WTI, as displayed in Table 6A, provided interesting results. Indeed, it can be observed that there is a manifest difference in the efficiency of UF-MBR2 toxicity removal, allowing its exact distinction in the two previously mentioned performance periods. In particular, the first period was characterised by medium and highly toxic discharges, whereas the second one presented most of the time effluents with no or low toxicity levels. Thus, the UF-MBR2 after the starting up period significantly improved the final quality of the discharge. Considering the WTI defined on the basis of two toxicity endpoints (i.e. V. fischeri and C. gigas; V. fischeri and M. galloprovincialis) as displayed in Table 6B and C, it could be observed that there is a general similarity between their sensitivities: the correlation analysis between the toxicity results elaborated on the two WTIs indicated that there was no significant difference ( p < 0.01) between C. gigas and M. galloprovincialis sensitivities (91% of correlation) towards the tested mixed domestic and industrial wastewater samples (Z(i,p,ac)1e21). Moreover, the UF-MBR2 performances were also compared to the very strict regulatory limits about effluent discharge from mixed domestic and industrial WWTPs in Venice lagoon (DM, 1999) and treated wastewater reuse for non-potable purposes (DM, 2003). Due to the above-mentioned problems occurred during the WWTP start-up, it was decided to comment results just from Zac10e21 samples. About DM (1999), it could be observed that all parameters complied with the relative regulatory limits except for NeNHþ 4 for Zac16,18,19,21, for PTOT for Zac13,19,20 and for Se for Zac11,13,14,17e21, even though in a not meaningful way as stated by the absence/low toxicity of effluents in the second monitoring period (10e21). The same problems were found about the compliance with DM (2003) for treated wastewater
reuse, where the same limits of DM (1999) are applied for both for NeNHþ 4 and Se. In conclusion, it could be said that UFMBR2 effluents are not immediately suitable for discharge and/or reuse, but some actions should be taken to obtain the full regulatory compliance. Indeed, one of the main reasons of excess Se concentration in the final effluent might be associated to the fact that saturated activated carbon were not substituted, but only backwashed, as well as to the presence of potential great fluctuations in raw wastewater loads treated by the WWTP. The complete substitution of activated carbon filter will allow to improve Se removal as well as the slight excess of NeNHþ 4 concentrations compared to the relative regulatory limits (Jegadeesan et al., 2003; Long et al., 2008). Previous unpublished studies evidenced that regulatory requirements could fully complied via substituting the activated carbon column with a reverse osmosis treatment stage. A biplot summarising PCA results concerning chemical data for UF-MBR2 wastewater samples weighted on WTI values from Table 6A is shown in Fig. 5. The first two principal components accounted for 31.21% and 27.08% of the variation, respectively. Therefore 58.28% of the variation can be depicted by a two-axis ordination diagram. The biplot regarding components loadings suggested that the F1 scores are influenced by high values of COD, total concentration of Cr, Cu, As, Fe, SS, NeNHþ 4 , Ba and Sb in ascending order, which are clustered together and have positive loadings on the first axis. In addition, the loading of Cd, Zn, Al, Co, Ni, Mn on the F2 suggested that the second component scores could reflect the concentrations of these compounds in the samples. Looking at the ordination plot of component scores in the F1eF2 biplot, it was found that wastewater samples could be clustered in three main groups: at the bottom left the good quality permeates (Zp10e21) and effluents after activated carbon filtering from the second monitoring period (Zac10e21), at the bottom right the raw wastewater specimen (Zi1e21), while at the top left the contaminated permeates and activated carbon filtered permeates from the first monitoring period (Zp1e9 and Zac1e9). This last group showed to be mainly influenced by the presence of some heavy metals (Ni, Mn and Zn) and chlorine.
4.
Conclusions
This research focused on how to provide useful information to support innovation in the field of wastewater treatment to comply with most recent legislative trend, which makes the assessment of advanced wastewater treatment technologies a key issue for water sustainability and its potential for reuse. It has been evidenced that: - ecotoxicological tools may be successfully used to discriminate between wastewater treatment technologies efficiency; - the combined use of physico-chemical analyses and ecotoxicological issues might support potential effluent reclamation and reuse with the final aim of approaching the zero emissions discharge; - it is worth to consider tools and approaches providing strategic integrated results on whole wastewater samples
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 3 7 e4 4 5 0
and to select them on the basis of specific relative targets such as those related to receiving water body characteristics (e.g. transitional and marine-coastal waters); - bacteria and bivalves toxicity outputs elaborated in the form of species-specific scoring and wastewater toxicity index offered immediate interesting tips for discriminating between the efficiency of AS-SBR and UF-MBR wastewater treatment technologies; - UF-MBR technologies applied both to commercial and mixed domestic and metal industries wastewaters have shown to be able to provide superior quality effluents, as confirmed by physico-chemical analyses, even if some of the very strict regulatory limits were sporadically slightly exceeded (i.e. substantially unworthy under the ecotoxicological viewpoint); - conversely, the AS-SBR facility did not attain the same level of efficiency of UF-MBR, displaying unpredictable and discontinuous performance in the final effluent quality.
Acknowledgements This study was partly supported by a grant from Ingegnerie e Tecnologie Ecologiche srl (In.T.Ec. srl). The authors thank Andrea Scandella for chemical analyses.
references
Abegglen, C., Joss, A., McArdell, C.S., Fink, G., Schlu¨sener, M.P., Ternes, T.A., Siegrist, H., 2009. The fate of selected micropollutants in a single-house MBR. Water Research 43 (7), 2036e2046. APHA, American Public Health Association, 1998. Standard Methods for the Examination of Water and Wastewater, twentieth ed.. Washington, USA. ASTM, American Society for Testing and Materials, 2004. Standard Guide for Conducting Static Acute Toxicity Tests Starting with Embryos of Four Species of Saltwater Bivalve Molluscs E724/2004. Azur Environmental, 1998. Microtox Acute Toxicity Test Guide, Users Manual. Bakir, H.A., 2001. Sustainable wastewater management for small communities in the Middle East and North Africa. Journal of Environmental Management 61 (4), 319e328. Ben, W., Qiang, Z., Pan, X., Chen, M., 2009. Removal of veterinary antibiotics from sequencing batch reactor (SBR) pre-treated swine wastewater by Fenton’s reagent. Water Research 43 (17), 4392e4402. Celis, E., Elefsiniotis, P., Singhal, N., 2008. Biodegradation of agricultural herbicides in sequencing batch reactors under aerobic or anaerobic conditions. Water Research 42 (12), 3218e3224. DM, 30th July 1999. Ministry of Environment Decree about discharge limits in Venice lagoon (Limiti allo scarico in laguna di Venezia e nei corpi del bacino scolante). DM, 12th June 2003. Ministry of Environment Decree about treated water reuse limits for non potable purposes (Regolamento recante norme tecniche per il riutilizzo delle acque reflue in attuazione dell’articolo 26, comma 2, del D.Lgs 11 maggio 1999, n. 152).
4449
DR, 24th August 1995. Bollettino Ufficiale della regione del Veneto (Italy) n. 4287, Veneto Region Decree about effluent discharge from urban individual WWTPs in Venice lagoon. Gutie´rrez, M., Etxebarria, J., de las Fuentes, L., 2002. Evaluation of wastewater toxicity: comparative study between Microtox and activated sludge oxygen uptake inhibition. Water Research 36 (4), 919e924. Harremoe¨s, P., 2000. Advanced water treatment as a tool in water scarcity management. Water Science & Technology 42 (12), 73e92. Høibye, L., Clauson-Kaas, J., Wenzel, H., Larsen, H.F., Jacobsen, B. N., Dalgaard, O., 2008. Sustainability assessment of advanced wastewater treatment technologies. Water Science & Technology 58 (5), 963e968. Ho, G., Anda, M., 2004. Centralised versus decentralised wastewater systems in an urban context: the sustainability dimension. 2nd International Water Association Leading Edge Conference on Sustainability: Sustainability in Water-Limited Environments, Sydney, Australia, November 2004. In: Olsson, G. (Ed.), Proceedings of Second IWA Leading-Edge Conference on Sustainability: Sustainability in Water-Limited Environments. IWA, London, UK. IPPC, Integrated Pollution Prevention and Control, 2008. Directive 2008/1/EC of the European Parliament and of the Council of 15 January 2008 concerning integrated pollution prevention and control. ISO, International Organization for Standardization, 2007. Water Quality e Determination of the Inhibitory Effect of Water Samples on the Light Emission of Vibrio fischeri (Luminescent Bacteria Test) e Part 3: Method Using Freezedried Bacteria. International Organisation for Standardisation, ISO 11348-3. Jegadeesan, G., Mondal, K., Lalvani, S.B., 2003. Comparative study of selenite adsorption on carbon based adsorbents and activated alumina. Environmental Technology 24 (8), 1049e1059. Libralato, G., Losso, C., Avezzu`, F., Volpi Ghirardini, A., 2009. Influence of salinity adjustment methods, salts and brine, on the toxicity of wastewater samples to mussels embryos. Environmental Technology 30 (1), 85e91. Libralato, G., Avezzu`, F., Volpi Ghirardini, A. How toxic is toxic? A proposal for wastewater toxicity hazard assessment. Ecotoxicology and Environmental Safety, in press, doi:10.1016/ j.ecoenv.2010.03.007. Long, X., Cheng, H., Xin, Z., Xiao, W., Li, W., Yuan, W., 2008. Adsorption of ammonia on activated carbon from aqueous solutions. Environmental Progress 27 (2), 225e233. Maurer, M., Rothenberger, D., Larsen, T.A., 2006. Decentralised wastewater treatment technologies from a national perspective: at what cost are they competitive? Water Science & Technology 5 (6), 145e154. MAV, Magistrato alle Acque di Venezia, 2007. Venice, an obliged choice. Individual wastewater treatments. (Venezia, una scelta obbligata. I trattamenti individuali di depurazione). In: Ferrari, G., Tromellini, E. (Eds.). Marsilio, Venice, Italy. Mendonc¸a, E., Picado, A., Paixao, S.M., Silva, L., Cunha, M.A., Leitao, S., Moura, I., Cortez, C., Brito, F., 2008. Ecotoxicity tests in environmental analysis of wastewater treatment plants: case study in Portugal. Journal of Hazardous Materials 163 (2e3), 665e670. Metcalf and Eddy Inc., 2003. Wastewater Engineering: Treatment and Reuse, fourth ed. McGraw-Hill, New York, USA. MSFD, Marine Strategy Framework Directive, 2008. Directive 2008/56/EC Establishing a Framework for Community Action in the Field of Marine Environmental Policy. Nosenzo, G., Gualdi, A., Mignani, M., Bellini, G., 2005. Industrial removal of micropollutants from water of varying quality by
4450
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 4 3 7 e4 4 5 0
FLAMEC flat sheet polymeric membranes cassettes. Desalination 185 (1e3), 167e183. OSPAR, 2007. Practical Guidance Document on Whole Effluent Assessment. 316/2007, ISBN 978-1-905859-55-9. Phillips, B., Hunt, J.W., Anderson, B.S., Puckett, H.M., Fairey, R., Wilson, C.J., Tjeerdema, R., 2001. Statistical significance of sediment toxicity test results: threshold values derived by the detectable significance approach. Environmental Toxicology & Chemistry 20 (2), 371e373. , J., Petrovic , M., Barcelo´, D., 2009. Fate and Radjenovic distribution of pharmaceuticals in wastewater and sewage sludge of the conventional activated sludge (CAS) and advanced membrane bioreactor (MBR) treatment. Water Research 43 (3), 831e841. REACH, 2006. EC Regulation N. 1907/2006 of the European Parliament and of the Council. Ricco, G., Tomei, M.C., Ramadori, R., Laera, G., 2004. Toxicity assessment of common xenobiotic compounds on municipal activated sludge: comparison between respirometry and Microtox. Water Research 38 (8), 2103e2110. RIKZ, 1999. Handbook for toxicity testing of dredged material. In: Schipper, C.A., Stronkhorst, J. (Eds.). National Institute for Coastal and Marine Management/RIKZ, The Hague (NL) Report RIKZ-99.012. Santos, L.H.M.L.M., Arau´jo, A.N., Fachini, A., Pena, A., DelerueMatos, C., Montenegro, M.C.B.S.M., 2010. Ecotoxicological aspects related to the presence of pharmaceuticals in the aquatic environment. Journal of Hazardous Materials 175 (1e3), 45e95.
SEPA, Scottish Environmental Protection Agency, 2003. Technical Guidance Manual for Licensing Discharges to Water e Direct Toxicity Assessment 03-DLM-COPA-EA3, pp. 1e24. USEPA, 1992. Methods for Determination of Metals in Environmental Sample. CRC Press Inc., Boca Raton, USA. USEPA, 1995. Short Method for Estimating the Chronic Toxicity of Effluents and Receiving Waters to Coastal Marine and Estuarine Organisms. EPA/600/R-95-136, Cincinnati, USA, pp. 209e258. USEPA, 2004. NPDES Compliance Inspection Manual. EPA 305-X03-004 Cincinnati, USA. Verliefde, A., Cornelissen, E., Amy, G., Van der Bruggen, B., van Dijk, H., 2007. Priority organic micropollutants in water sources in Flanders and the Netherlands and assessment of removal possibilities with nanofiltration. Environmental Pollution 146 (1), 281e289. Volpi Ghirardini, A., Losso, C., Arizzi Novelli, A., Baugrave, A., His, E., Ghetti, P.F., 2005. Mytilus galloprovincialis as bioindicator in embryotoxicity testing to evaluate sediment quality in the lagoon of Venice (Italy). Chemistry & Ecology 21 (6), 455e463. Weber, B., Cornel, P., Wagner, M., 2007. Semi-centralised supply and treatment systems for (fast growing) urban areas. Water Science & Technology 55 (1e2), 349e356. WFD, Water Framework Directive, 2000. Directive 2000/60/EC of the European Parliament and of the Council of 23 October 2000 Establishing a Framework for Community Action in the Field of Water Policy.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 9 0 e4 6 0 0
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Using silicone passive samplers to detect polycyclic aromatic hydrocarbons from wildfires in streams and potential acute effects for invertebrate communities Ralf Bernhard Scha¨fer a,*, Laurence Hearn b, Ben J. Kefford a,c, Jochen F. Mueller b, Dayanthi Nugegoda a a
School of Applied Sciences, RMIT University, PO Box 71, Bundoora, VIC 3083, Australia National Research Centre for Environmental Toxicology (Entox), University of Queensland, 39 Kessels Road, Coopers Plains, QLD 4108, Australia c Centre for Environmental Sustainability, University of Technology Sydney, PO Box 123, Broadway, NSW 2007, Australia b
article info
abstract
Article history:
Silicone rubber passive samplers spiked with 4 deuterated performance reference
Received 13 December 2009
compounds were deployed for 29e33 days to estimate the concentrations of 16 polycyclic
Received in revised form
aromatic hydrocarbons (PAHs) in 9 streams in Victoria, Australia, following a wildfire.
10 May 2010
Silicone rubber strips of 2 thicknesses were used to obtain information on the status of
Accepted 27 May 2010
uptake of the chemicals of interest at retrieval. In addition, we monitored the stream
Available online 2 June 2010
macroinvertebrate community for potential effects of PAHs or other fire organics. All selected PAHs were detected in the passive samplers and the sampling rates ranged from 0.5
Keywords:
to 50 L/day significantly varying between sites but not compounds, presumably due to
Passive sampling
differences in current velocity. The estimated water concentrations were 0.1e10 ng/L for
Field study
total PAHs with phenanthrene, pyrene and fluoranthene accounting for 91% of the total
Toxic effects
concentration. All PAHs were a factor of 1000 or more below the reported 48-h median lethal
Stream invertebrates
concentrations (48-h LC50) for Daphnia magna. Two sites located closest to the fires exhibited
Freshwater
elevated concentrations compared to the other sites and the passive samplers in these sites
Organic toxicants
remained in the integrative uptake regime for all compounds, suggesting precipitationassociated PAH input. No acute toxic effects of PAHs or other fire organics on the invertebrate community were detected using a biotic index for organic toxicants (SPEAR), whereas a non-specific biotic index (SIGNAL) decreased in two sites indicating impacts from changes in other environmental parameters. We conclude (1) that silicone-based passive samplers with two different area-to-volume ratios represent a promising tool for determining organic toxicants and (2) that PAHs from wildfires are unlikely to be a common main cause for firerelated ecological effects in streams adjacent to burnt regions. ª 2010 Elsevier Ltd. All rights reserved.
1.
Introduction
In several regions of the world such as South Europe, parts of America and Australia wildfires represent an integral part of
the landscape and many species are tolerant of or even dependent on recurring wildfire events (Bradstock, 2008). Fires can have impacts on stream ecosystems. While the direct effects of fires on streams such as higher temperatures are in
* Corresponding author. Tel.: þ61 (03) 9925 7105; fax: þ61 (03) 9925 7110. E-mail address:
[email protected] (R.B. Scha¨fer). 0043-1354/$ e see front matter ª 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.watres.2010.05.044
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 9 0 e4 6 0 0
most cases rather negligible, subsequent indirect effects may be more severe (Minshall, 2003). Indirect effects include a decrease in dissolved oxygen, a decrease in allochthonous energy, an increase in nutrients and an increase in hydrodynamic stress resulting from the input of sediments and ash slurry, as well as channel and vegetational alterations (Minshall, 2003; Hall and Lombardozzi, 2008). Especially strong rain events following fires have been identified as a trigger for post-fire effects since they can wash considerable ash and sediments into the aquatic system. Declines of fish populations and changes in invertebrate communities have been attributed to the input of sediments and ash after fires (Earl and Blinn, 2003; Lyon and O’Connor, 2008). Since the input of ash and sediments simultaneously alters several water quality parameters (including turbidity, hydrological conditions and dissolved oxygen) it is in most cases unclear as to which factor or combination of factors is responsible for observed biological effects (Vieira et al., 2004). The potential contribution of polycyclic aromatic hydrocarbons (PAHs) and other fire organics that are released into the environment during fires resulting from the combustion of biomass (Hays et al., 2005; Yuan et al., 2008) has been little studied (VilaEscale et al., 2007). These substances may enter freshwater systems adsorbed to ash, sediments or via dry deposition. To date, only two studies have been published on the input of PAHs in freshwater systems following wildfires, both conducted in Spain (Olivella et al., 2006; Vila-Escale et al., 2007). Passive sampling has gained growing attention for the continuous sampling of pollutants in the environment and has been effective in catching episodic pulse exposures in streams (Scha¨fer et al., 2008b; Shaw and Mueller, 2009). Polydimethylsiloxane (PDMS) as a receiving phase represents a promising approach for the sampling of a wide-range of organic pollutants in terms of polarity including PAHs (Smedes, 2007; Bauer, 2008; ter Laak et al., 2008), whereas most existing receiving phases are relatively selective for compounds of a certain polarity. Several studies successfully employed PDMS fibres or rods as a receiving phase (Vrana et al., 2001; Wennrich et al., 2003; Ouyang et al., 2007). Due to the relatively small volume of the receiving phases used in these studies the samplers reach equilibrium with the sampling phase within a few days or use a membrane to slowdown the uptake rate, which may compromise the sampling of episodic events (Scha¨fer et al., 2008a). An alternative represents the use of large-volume PDMS rubber sheets without membrane, but it has only been applied rarely for passive sampling in the aquatic environment (Smedes, 2007; Bauer, 2008). For PDMS type samplers where the chemicals partition with the sampling phase, it is typically assumed that uptake (and release) follows first order kinetics. Thus a time integrated exposure concentration, the so-called timeweighted average (TWA) concentration, can be obtained provided that clearance of the chemical is low relative to the uptake i.e. sampling in the linear uptake phase. Depending on the volume of the sampler, the surface area exposed, the capacity of the sampler-to-water partition coefficient Ksw and factors that affect the kinetics, samplers will exit the linear uptake phase and enter the curvilinear uptake phase more or less rapidly i.e. from minutes to years (Vrana et al., 2005). For an adequate estimate of the water concentration of an analyte
4591
after field deployment it is crucial to know whether the passive sampler remained in the integrative uptake regime (i.e. linear uptake phase) or approached equilibrium within the deployment time (Vrana et al., 2005). Bartkow et al. (2004) employed two receiving phases with different thicknesses in air passive sampling for PAHs to determine the uptake regime (integrative or equilibrium) after deployment. Different thicknesses result in a different area-to-volume ratio of the receiving phases, which influences the equilibration times with the sampling phase. Consequently, the ratio of the analyte mass between receiving phases of different thicknesses changes until equilibrium is reached in both phases and allows for a derivation of the uptake regime (Bartkow et al., 2004). In water, this approach has been used only in one study employing polyethylene strips as receiving phase (Mu¨ller et al., 2001). During the summer 2008/2009 Victoria, Australia was subject to the lowest precipitation (0 mm between 1.1.2009 and 28.2.2009) and highest temperatures on record in 120 years across several regions (BOM, 2009a,b). These conditions, on top of an extended drought, promoted the outbreak of several large forest fires in Victoria. The largest fire represented the Kilmore EasteMurrurundi Complex (KEMC) fire to the north-east of Melbourne that burned approximately 260 000 ha of land (BOM, 2009c). We initiated an ad-hoc study that had the following objectives: (1) to determine the input of PAHs in streams during the fires and associated with rain events after the fires, (2) to assess the suitability of PDMS passive samplers with different thicknesses for the determination of the uptake regime (3) to assess potential impacts of PAHs or other fire organics released by the KEMC fire on invertebrates in streams adjacent and in the vicinity of the burned area.
2.
Materials and methods
2.1.
Study design and rain events
The study area was located within 200 km east of Melbourne, Victoria, Australia and comprised nine sampling sites in first, second and third-order streams (Fig. 1). The sites were selected (1) to represent a potential gradient in exposure from ash and smoke during the wildfires and potential runoff events and (2) based on safe accessibility after the outbreak of the fires (Fig. 1). The site code corresponded to the site code of the respective site in a larger study. No known industrial facilities were present in the catchments that could account for significant PAH discharges. Nevertheless, except for the Sites 14 and 15, which were located in forested areas, most of the sites were in the vicinity of highways or roads so that exhaust fumes represent a potential source of PAHs. The passive samplers were deployed approximately 10 days after the outbreak of the fires (19th to 21st of February), which was in the middle of the period of the KEMC fire (9th February to 5th of March) but before the first rain event. The samplers were recovered between 29 and 33 days after deployment (22nd to 24th of March). The invertebrate fauna was monitored between the 15th and 20th of February (Supplementary material, Table S1). Two precipitation events
4592
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 9 0 e4 6 0 0
Fig. 1 e Location of the sampling sites (dots with numbers) in streams (black lines) in relation to the wildfires (grey area). The figure was generated using data on the hydrological network, wildfire areas and the base map for Victoria provided by the Department of Sustainability and Environment, Victoria. The exact positions of the sampling sites are given in Table S1 (Supplementary material).
of 15 and 25 mm within 24 h were recorded (4th to 15th of March, respectively) within the study period, which represented the first precipitation events after the outbreak of the KEMC fire (BOM, 2009a,c).
2.2. Preparation, deployment time and extraction of the passive samplers Strips of 5 cm width, 62 cm length and two different thicknesses (0.5 and 1.5 mm, hereafter thin and thick) were cut from large-volume polydimethylsiloxane (PDMS) sheets (Purple pig, Notting Hill, Victoria, Australia), placed in a 0.5 L jar and twice pre-extracted before usage for 48 h with 500 mL of GC-grade n-hexane/acetone 3:1 (Ajax, Taren Point, NSW, Australia) to remove impurities. Rubber strips were then dried
under nitrogen and returned to the jars for spiking with four deuterated compounds (purity of 98.5% or higher) acenaphthene-d10, phenanthrene-d10, chrysene-d12 and perylened12 (Accustandard, New Haven, CT, USA) and served as performance reference compounds (PRCs) (Huckins et al., 2002). The spiking was done by exposing the strips to 400 mL of a 20:80 (vol./vol.) solution of HPLC-grade MeOH: bi-distilled H2O containing 100 mL of a stock solution (4 mg/100 mL) of the deuterated compounds in HPLC-grade MeOH (Merck, Darmstadt, Germany). The jars were shaken for 60 h as this time was sufficient in another study to reach equilibrium between the receiving phase and the standard in solution (Booij et al., 2002). A procedural blank and a method blank were processed concurrently and used to detect potential contamination of the samplers during the pre-extraction step and during
4593
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 9 0 e4 6 0 0
spiking. The spiking of PRCs allows for the in-situ determination of sampling rates, which is used in calculating water concentrations, and accounts for between-site variation in environmental parameters such as current velocity or temperature (Huckins et al., 2002). Each passive sampler consisted of a thin and thick PDMS strip placed in a steel (1 cm) mesh. One sampler was fixed approximately 10e30 cm above the bottom of the stream at each site. A field blank was exposed to the air during deployment and retrieval of the samplers to account for airborne contamination during field handling. On recovery, the PDMS strips were cleaned with analytical-grade ethanol (Ajax, Taren Point, NSW, Australia) to eliminate biofilms. Samplers were then dried using paper tissue and stored at 4 C in a glass jar. In the laboratory, all blanks and exposed samplers were thoroughly cleaned with bi-distilled water, subsequently rinsed with HPLC-grade acetone (Merck, Darmstadt, Germany) to remove traces of water and air dried for 10 min in a fume hood. Elution was conducted twice for 24 h each with 250 mL GC-grade n-hexane (Ajax, Taren Point, NSW, Australia). The eluate was gently blown down to 1 mL at 50 C in a water bath in a nitrogen stream using Mini-vap evaporators (Sigma-Aldrich, Melbourne, Australia) and then transferred to a vial. Subsequently, 10 mL of triphenylphosphate (TPP) (100 ng/mL) (Merck, Darmstadt, Germany) was added as internal standard (IS).
2.3.
Chemical analysis
The chemical analysis was conducted using high-resolution gas chromatography/high-resolution mass spectrometry (HRGC/HRMS, HP 5890 II GC coupled with VG AutoSpec); splitless injection; injector temperature 220 C. The target
PAHs (Table 1) were separated on a J & W Scientific DB-1701 column (30 m 0.25 mm id., 0.25 mm film thickness) with ultra-high purity helium carrier gas; temperature program 110 C for 2 min, 10 C min1 to 170 C, 5 C min1 to 300 C and held 6 min; total run-time 40 min. The mass spectrometer operating conditions were: ion source and transfer line temperatures 265 C; ionization energy 38 eV; electron multiplier voltage set to produce a gain of 106. Resolution was maintained at 5000 (10% valley definition) throughout the sample sequence. Selective ion recording (SIR) experiments were performed in the electron impact (EI) mode monitoring the quantitative ion for each target analyte, including the recovery standard TPP, which was added to the samples prior to HRGC/HRMS analysis. Quantitation of PAHs was performed using an external standard calibration (ng/vial) with criteria for positive identification: (i) retention time within 1 s of the standard retention time (ii) signal to noise ratio > than 3:1 and (iii) signal > limit of reporting (LOR) referring to the average noise level in the field blanks. The limit of determination (LOD) was 0.5 ng/mL, the method detection limits (MDL) are given in Table 2.
2.4.
Estimation of PAH concentrations in water
The mass of PAHs determined in the field blanks was subtracted from the field samples to correct for contamination during sample handling and processing. The method of calculation of the water concentrations was selected based on the kinetic regime of a compound (integrative or equilibrium) during field exposure. This was assessed using the ratio of the compound mass in the thick and thin samplers (Bartkow et al., 2004). During integrative uptake, the ratio of the mass in both samplers is approximately 1 and declines to the ratio of the
Table 1 e PAH target compounds and labelled standards with physicochemical properties. Compound Acenaphthene-d10 Phenanthrene-d10 Chrysene-d12 Perylene-d12 Phenanthrene Fluoranthene Pyrene Chrysene Benzo (a) anthracene Perylene Benzo (e) pyrene Benzo (k) fluoranthene Benzo (b) fluoranthene Benzo (a) pyrene Indeno (1,2,3cd) pyrene Benzo (ghi) perylene
Abbreviation
Number of rings
Quantitation ion
log10 Kswa
log10 Kowb
48-h LC50 (mg L1)c
Ace-d10 Phe-d10 Chr-d12 Per-d12 Phe Flu Pyr Chr B(a)A Per B(e)P B(k)F B(b)F B(a)P I(c,d)P B(g,h,i)P
3 3 4 5 3 4 4 4 4 5 5 5 5 5 6 6
164.1410 188.1410 240.1692 264.1692 178.0782 202.0782 202.0782 228.0939 228.0939 252.0939 252.0939 252.0939 252.0939 252.0939 276.0939 276.0939
3.35 3.61 4.91 5.38 3.89 4.38 4.44 4.97 5.06 5.44 5.45 5.51 5.51 5.52 5.72 5.92
3.92 e e e 4.52 5.2 5 5.86 5.91 6.25 6.44 6.11 5.78 6.35 6.58 6.9
e e e e 699 11.38 4.33 NT 1.48 e 1.43 e e 1.62 e 1.04
a taken from Smedes (2007), except for phenanthrene taken from Bauer (2008) and values in italics estimated with regression imputation (log Ksw ¼ 0.85 log Kow þ 0.12; r2 ¼ 0.95). b Sangster, J., 2009. LOGKOW. A databank of evaluated octanolewater partition coefficients (Log P). URL: http://logkow.cisti.nrc.ca/logkow/ intro.html (accessed 27.10.09). c 48-h median lethal concentration (48-h LC50) for Daphnia magna as given in Lampi et al. (2006). NT ¼ Nontoxic at concentration levels below maximum water solubility.
4594
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 9 0 e4 6 0 0
Table 2 e Derived sampling rates (L/day) at the different sites for analytes (abbreviations see Table 1) which remained in integrative uptake regime during field deployment (see text for details). Site 3 5 10 11 13 14 15 23
Phe
Flu
Pyr
B(a)A þ Chr
3.9
0.5
0.5
1.4 1.2 1.4
8.9
7.6
7.2
4.2
Per
B(e)P
B(b)F þ B(k)F
B(a)P
I(c,d)P
B(g,h,i)P
1.4 3.8 3.2 3.8 16 3.7 20 12
1.2 3.9 3.2 3.9 17 2.3 20 9.9
1.4 4.4 3.7 4.4 16 2.6 19 11
1.4 4.5 3.8 4.5 16 2.1 19 10
2.0 7.2 6.0 19 17 12 18 8.4
3.2 11 9.6 23 19 16 20 13
volumes of thick and thin samplers (mean: 0.29; standard deviation (sd): 0.03) when reaching equilibrium with the sampling phase. We assumed integrative uptake for a compound if two conditions were met: (1) mass ratio of thin to thick samplers >0.65 and (2) if the majority of compounds with a higher sampler-to-water partitioning coefficient (Ksw) value (Table 1) than the compound also remained in the integrative uptake regime, else equilibrium with the water phase was assumed. For integrative uptake, the TWA water concentration CTWA was calculated according to: CTWA ¼
ms ðtÞ Ksw Vke t
(1)
where ms(t) is the mass of compound s in the receiving phase after the deployment time t, V is the volume of the receiving phase and ke is the exchange rate constant. The product of Ksw, ke and V is also called the substance-specific sampling rate RS. The values of ke for the different compounds were derived from the relationship between Ksw and the ke values of the four PRCs that were calculated according to:
ke ¼
ms ðtÞ lnðm s ð0Þ t
(2)
where ms(0) is the initial mass of compound s in the receiving phase after spiking, which was determined using the method blanks. The relationship between Ksw and ke was modelled using a four parametric log-logistic regression using maximum and minimum ke values as upper and lower limit, respectively. Only the results for the PRCs of thick samplers were included as the PRCs of the thin samplers exhibited complete loss of most PRCs. In case of equilibrium, the water concentration Ceq was calculated according to: Ceq ¼
ms ðtÞ Ksw V
(3)
2.5. Monitoring of physicochemical variables and macroinvertebrates and computation of biotic indices D-Opto optical dissolved oxygen loggers (Zebra-Tech, Nelson, New Zealand) were deployed concurrently with the passive samplers at three sites (Fig. 1; Sites 5, 14 and 15) to measure a potential drop in dissolved oxygen from ash and sediment inputs (Minshall, 2003). Physicochemical parameters (temperature, pH, conductivity, dissolved oxygen and turbidity) were measured at the deployment and retrieval of the passive samplers (Hanna Instruments, Melbourne,
Australia). Current velocity was estimated based on the time an object needed to travel 1 m stream distance. The invertebrate fauna from edge/pool habitat was monitored using a rapid bioassessment method that comprised sweep sampling of representative habitats in the streams and field picking of the macroinvertebrates (Chessman, 1995; EPA, 2003). Since two family-level biotic indices were employed for data analysis, the taxa were identified to family-level or lower in the laboratory. We used the SPEARorganic biotic index to detect potential changes in the macroinvertebrate community composition due to toxic effects of PAHs or other fire organics as this index demonstrated high selectivity in its response towards organic toxicants in two other studies (Beketov and Liess, 2008; Schletterer et al., 2010). Secondly, an established Australian biotic index, SIGNAL scores, that responds to a variety of stressors was employed to identify general effects on the invertebrate community from fire-associated stressors (Chessman, 1995; EPA, 2003). Both indices are calculated by averaging sensitivity values of all taxa in a sample. For SPEARorganic and SIGNAL scores, the sensitivity value represents the calculated relative sensitivity of a taxon to organic toxicants (Sorganic) (von der Ohe and Liess, 2004) and an assigned general pollution sensitivity grade (Chessman, 1995), respectively. The taxa found in the sampling, their sensitivity values and details on the calculation of Sorganic can be found in the Supporting material, Table S2. For both indices we examined changes from the beginning of the fires to the postfire period including the first rainfall event by computing the ratio of the index values.
2.6.
Data analysis
Although the thick and thin samplers do not represent a sample from the same statistical population in a strict sense, we calculated the relative range (RR) as dispersion measure for the estimated water concentrations: RRð%Þ ¼ ðmaxðXÞ minðXÞÞ
X
(4)
where X are the observations for the respective compound at a certain site and Xis the mean of X. The RR is a more conservative estimate of the sample dispersion compared to the relative standard deviation. All statistical computations and graphics (except Fig. 1 created with Quantum Gis, www.qgis. org) were created with the open-source software package R (www.r-project.org) using version 2.10.0 (R Development Core Team, 2009).
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 9 0 e4 6 0 0
3.
Results
3.1. Release of performance reference compounds from the passive samplers and sampling rates The PRCs spiked into the PDMS passive samplers before exposure exhibited different release rates for the compounds and in the sites. Acenaphthene-d10 dissipated almost completely from the receiving phase in all sites relating to an exchange rate coefficient ke of 0.15 d1, whereas perylene-d10 exhibited negligible dissipation into the water phase resulting in a very low ke (Fig. 2). Phenanthrene-d10 showed the largest variation between sites in terms of release rate from the passive samplers ranging from complete loss to high retainment in the receiving phase. In all sites, PRCs exhibited an increasing release from the receiving phase with a decrease of the log10 Ksw, which translates to a higher ke with a lower log Ksw for a PRC (Fig. 2). With a few exceptions, the sampling rates of compounds were relatively similar in a sampling site. The sampling rates between sites exhibited higher variation with Sites 13 and 15 having the highest sampling rates (Table 2).
3.2. Polycyclic aromatic hydrocarbons in the passive samplers and calculated water concentrations All target analytes were quantified above the LOD in the PDMS passive samplers and except for Site 22 positive detections in thin samplers corresponded to a detection in thick samplers.
4595
Compounds with lower molecular mass and lower log Ksw values were more frequently detected (Tables 1 and 3). Phenanthrene, benzo (a) pyrene, perylene and the sum of chrysene and benzo (a) anthracene were found in all sites and phenanthrene, pyrene and fluoranthene accounted for 91% of total PAH water concentrations overall sites (for 89% of mass accumulated in samplers) with several orders of magnitude higher concentrations compared to other PAHs determined in this study (Table 3). Site 15 and 22 exhibited 5e60-fold lower concentrations in PAHs compared to the other sites. Most compounds with a log Ksw < 5, which relates to an approximate log10 Kow < 6, reached equilibrium between the passive samplers’ receiving phase and the water phase within exposure time. By contrast, for samplers deployed at Sites 5 and 14 (the closest sites to the fires) all of the compounds in the samplers remained in the integrative uptake regime indicated by a mass ratio of thin to thick samplers of approximately 1 (Fig. 3). These sites differed from the other sites especially with regard to the compounds fluoranthene, pyrene and the sum of benzo (a) anthracene and chrysene relating to PAH species with four or less aromatic rings (Tables 1 and 3). Site 5 exhibited the highest estimated PAH water concentrations of all sites and compounds with up to 20-fold higher concentrations for fluoranthene and pyrene and 5- to 50-fold higher concentration in the total PAHs than the other sites (Table 3). The variation in the calculated water concentrations between the thin and thick passive samplers was relatively high ranging from 21% to 56% mean relative range for the different PAH species and reaching over 100% relative range for a few observations (Table 3).
3.3. Change of physicochemical parameters and biotic indices associated with the wildfires Most physicochemical parameters exhibited only a slight change (<20%) from the beginning of the wildfires to the period after the fires. Dissolved oxygen exhibited the greatest change with up to 50% decrease at Sites 3 and 5 (Supplementary material, Table S1). The continuous DO loggers showed a temporary decrease in Sites 5 and 14 that had a burnt catchment whereas Site 15 exhibited no response. The decrease was most pronounced in Site 5 where the daily minimum in %DO decreased from approximately 70e45% following the two rain events in march (Supplementary material, Fig. S1). No decrease in the SPEARorganic was observed from pre- to post-rainfall period indicating no change in the macroinvertebrate community with respect to the proportion of sensitive species (Fig. 4). The biotic index SIGNAL scores exhibited a decrease in only two of the sampling sites (5 and 13) when compared to the period at the beginning of the wildfires (Fig. 4). Fig. 2 e Relationship between the exchange rate constant ke and the sampler-to-water partitioning coefficient log KSW for the four performance reference compounds chrysene-d12 (Chr-d12), perylene-d12 (Per-d12), acenaphthene-d10 (Ace-d10) and phenanthrene-d10 (Phed10) in the passive samplers in the sampling sites. A four parametric log-logistic regression using maximum and minimum ke values as upper and lower limit, respectively, was used to model the relationship.
4.
Discussion
4.1. Performance of the PDMS passive samplers when used to monitor organic toxicants We used performance reference compounds (PRCs) as in-situ calibration to account for differences in the environmental
4596
Table 3 e Time-weighted average (italics) and equilibrium water concentrations (pg/L) of PAHs with variation (±Relative range) and selected PAH ratios determined with passive samplers of two thicknesses in the sampling sites (numbers rounded to first two digits). Sites sorted according to distance to fires in increasing order. Compounds sorted according to molecular mass in increasing order. P Site Phe Flu Pyr B(a)A Per B(e)P B(b)F B(a)P I(c,d)P B(g,h,i)P Total Flu/ I(c,d)P/(I(c,d)P Sumd þ Chr þ B(k)F (Flu þ Pyr) þ B(g,h,i)P) 14
5 11 13 10 15a 22b 23 MDLc
240 (20%) 180 (56%) 5300 (36%) 57 (54%) 250 (71%) 55 (16%) 30 <1
320 (7.7%) 130 (18%) 3300 (16%) 89 (83%) 260 (99%) 80 (63%) 27 <0.8
78 (14%) 8 (26%) 308 (7.7%) 310 (28%) 14 (66%) 130 (34%) 1.6 0.6
9.1 (82%) 24 (105%) 25 (26%) 21 (16%) 9.5 (86%) 6.9 (29%) 2.2 <0.08
7.2 (29%) 9.1 (25%) 12 (6.9%) 18 (55%) 8.5 (24%) 9.7 (0.0%) 1.5 <0.08
11 (57%) 38 (6.8%) 16 (8.7%) 28 (75%) 11 (11%) 22 (7.7%) < 0.07 0.08
12 (59%) 48 (4.2%) 27 (30%) 48 (0.0%) 32 (47%) 67 (69%) 8.3 0.09
12 (68%) 13 (36%) 14 (16%) 22 (83%) 13 (12%) 17 (15%) <0.04 0.06
3.9 (170%) 9.3 (34%) 8.7 (0.0%) 15 (86%) 13 (7.8%) 13 (15%) <0.02 0.04
126 (12%) 1
120 (17%) 0.8
4.8 (54%) 0.2
6.4 (47%) 0.08
5.3 (5.5%) 0.08
8.2 (6.9%) 0.07
4.9 (170%) 0.06
7.7 (10%) 0.04
6.1 (6.9%) 0.02
1200
0.43
0.75
1.18
1100
0.59
0.59
1.17
9300
0.62
0.61
1.22
890
0.39
0.58
0.97
1900
0.48
0.5
0.99
850
0.41
0.57
0.98
130 160 1100
0.53 e
e
ee 0.57
0.51
0.56
ee ee 1.07
a No variation available since the thin sampler was lost. b No variation given since estimated water concentrations were close to the LOD, resulting in observations
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 9 0 e4 6 0 0
3
550 (14%) 630 (43%) 260 (4.6%) 280 (31%) 1200 (130%) 460 (55%) 63 160 (27%) 780 (27%) 3
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 9 0 e4 6 0 0
Fig. 3 e Relationship between the mass ratio of thin to thick samplers for the PAH analytes and the sampler-towater partitioning coefficient log Ksw in the different sites. Site 15 is not displayed as the thin sampler was lost. Only ratios relating to concentrations above LOD in both samplers were included. Distance to burnt upstream sections was <1 km for Site 14, 5 km for Sites 3 and 5, 10 km for Sites 10 and 11, whereas all other sites had no burnt upstream catchment.
Fig. 4 e Ratio of two biotic indices (SPEARorganic and SIGNAL scores) from the post-fire to the beginning of the fire period in the different sampling sites. Sites 22 and 23 not displayed as flow ceased and therefore no biotic samples were available after the fires. Distance to burnt upstream sections was <1 km for Site 14, 5 km for Sites 3 and 5, 10 km for Sites 10 and 11, whereas all other sites had no burnt upstream catchment.
4597
parameters between sites and to estimate water concentrations for compounds in the PDMS passive samplers that remained in the integrative uptake regime during deployment. PRCs are usually structurally related analogues of the target compounds that are spiked into the receiving phase before deployment and provide information on the dissipation rate. The release rates can be used to estimate uptake rates under the assumption of isotropic exchange (Huckins et al., 2002). Ideally, the selected PRCs would neither completely dissipate nor be fully retained in the receiving phase to allow for a precise estimate of the exchange rate coefficient and of the differences in the environmental conditions between sites. However, three of the four PRCs used in this study were almost completely released (acenapthene-d10) or retained (perylene-d12, chrysene-d12) in all of the sites (Fig. 2). Only phenanthrene-d10 exhibited larger variation in the dissipation rate between sites resulting in a major influence on the modelling of the relationship between ke and log Ksw (Fig. 2). The low dissipation rate resulting in high retainment of perylene-d12 and chrysened12 is in accordance with a study on the elimination of PRCs from various passive samplers that also reported almost complete retainment in the receiving phase for compounds with a log Ksw > 4.5 over a 28-day field trial (Allan et al., 2009). By contrast, the predictions of high retainment in the receiving phase for compounds with a log Ksw as low as 4 in half of the sites by the model for the relationship between ke and log Ksw (Fig. 2) do not match with the study of Allan et al. (2009) where considerable dissipation was observed for such compounds. Furthermore, under the assumption of isotropic exchange low dissipation rates correspond to low uptake rates, which would result in remaining in the integrative uptake regime for compounds with a log Ksw between 4 and 5. This prediction of the model (Fig. 2) contrasts with the results obtained for the determination of the kinetic regime using passive samplers of two different thicknesses (Bartkow et al., 2004). Here, most compounds with a log Ksw < 4.5 reached equilibrium with the water phase during exposure (Fig. 3). Possible explanations for the discrepancy between the dissipation of PRCs and the assessment of the kinetic regime using passive samplers of two different thicknesses include (1) that the uptake was slightly higher than the release from the receiving phase as already observed in another study (Mu¨ller et al., 2001) and (2) that the modelled relationship underestimated the dissipation rates. From both explanations follows that ke and consequently the sampling rates were underestimated (see Equation (1)). Nevertheless, the sampling rates derived using the predictions of ke values ranged from 0.5 to 50 L/day (Table 2) and are in good agreement with PAH sampling rates of PDMS strips in a calibration study (Bauer, 2008). In this study, Rs was below 1 L/day under stagnant conditions, between 1 and 10 L/day under a flow of 5e17 cm/s and between 5 and 63 L/day under a flow of 32 cm/s. This matches generally with our observations as the Rs were between 0.5 and 18 L/day for streams with a flow 20 cm/s and the Rs values in the fastest flowing (approximately 25 cm/s) streams 13 and 15 were higher than the other streams reaching up to 50 L/day. However, the sampling rates determined in this study should be regarded as approximate estimate as they were derived from the modelled relationship
4598
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 9 0 e4 6 0 0
between ke and log Ksw and the model was highly influenced by a single PRC. For relatively polar compounds with a log Ksw < 4 the PRCbased model predicted a high exchange rate constant relating to equilibrium regime (Fig. 2) in the Sites 5 and 14, whereas the mass ratio of these compounds in thick and thin passive samplers indicated integrative uptake regime (Fig. 3). We suggest to explain this discrepancy with differences in the exposure profile. PRCs dissipate continuously from the receiving phase, whereas the uptake is discontinuous for exposure resulting from episodic events or for exposure commencing at an unknown point in time such as precipitation-driven input of PAHs associated with wildfires (Olivella et al., 2006; Vila-Escale et al., 2007). In this situation, PRC dissipation can correspond to equilibrium regime while the uptake of compounds with a similar log Ksw can remain integrative. In fact, the two sites were closest to the burnt catchment area and PAH exposure may have resulted from the input of ash slurry associated with rain events later during deployment. This explanation is supported by two further observations. Firstly, the DO loggers in the sites indicated at least a small drop in dissolved oxygen concentrations that is characteristic for the input of ash slurry after fires (Lyon and O’Connor, 2008) and was not observed in Site 15 without burnt catchment. Secondly, the concentrations of PAHs were elevated in both sites compared to sites with a similar surrounding (Supplementary material, Table S1). The variation in the water concentrations determined with the thick and thin samplers was relatively high (Table 3) and exceeded the variation observed using the same approach in air passive sampling (Bartkow et al., 2004). In addition, the mass ratios for the compounds in the thick and thin samplers were relatively scattered between the two ratios relating to equilibrium (0.29) and integrative uptake (1) (Fig. 3). We attribute the variation to four sources: (1) general variation in field trials of aquatic passive samplers with low replication (Scha¨fer et al., 2008b), (2) variation resulting from matrix interference in chemical analysis, (3) concentrations close to the LOD for several compounds and (4) some compounds may have been in the curvilinear regime between integrative uptake and equilibrium. Future studies should consider employing samplers with three or more different thicknesses and/or replicate samplers to decrease variation and increase the robustness of the results.
4.2.
PAH concentrations in the streams
The estimated water concentrations ranged from 0.1 to 9 ng/L in the streams of this study, with the highest total concentrations in Site 5 that had its catchment extensively burnt. Although the water concentrations for compounds in the integrative uptake regime may have been overestimated due to an underestimation of the sampling rates (see previous section), the concentrations found were in accordance with two other studies on PAH exposure from wildfires. A total of 5 ng/L of 18 PAHs were measured in grab water samples after the first rainfall event in a Spanish creek after wildfires (VilaEscale et al., 2007). Similarly, in 4 Spanish streams in riverine remote areas 2, 6, 45 and 160 ng/L of 12 PAHs were detected one month after extensive wildfires in the catchment
(Olivella et al., 2006). The concentrations found in our and the latter two studies were a factor of 4e200 lower than those reported for 9 European rivers reviewed in Olivella et al. (2006). Studies using passive samplers in freshwater systems found up to 500 ng/L in individual PAHs in areas with industrial agriculture in the US (Alvarez et al., 2008), 16e30 ng/L in a river draining industrialised agricultural and urban areas in the Netherlands (Allan et al., 2009) and 2.4e5.7 ng/L for 9 PAHs for an Australian urban stream (Mu¨ller et al., 1999). The distribution by ring size showed a predominance of three- and four-ringed PAHs and was highest in the Sites 5 and 14 that were closest to the fires (Tables 1 and 3). Similarly, the two studies on Spanish streams found mainly low molecular size PAHs (phenanthrene and fluoranthene) that decreased with time after the wildfires (Olivella et al., 2006; Vila-Escale et al., 2007). However, studies in non-wildfire areas also reported a high ratio of three- and four-ringed PAHs (Mu¨ller et al., 1999; Alvarez et al., 2008) so that this distribution pattern cannot be used to identify wildfire-borne PAH contamination. Other indicators to identify the source of PAHs include the ratio of certain pairs of PAH compounds such as fluoranthene/(fluoranthene þ pyrene) and methylsubstituted to non-substituted PAHs (Yunker et al., 2002) or the determination of C14. We calculated two ratios for pairs of PAHs (Flu/(Flu þ Pyr) and I(c,d)P/(I(c,d)P þ B(g,h,i)P), see Table 3), for which a ratio >0.50 has been attributed to the combustion of wood, grass and coal, and a ratio <0.5 to the combustion of liquid fossil fuels. Indeed, the sum of both ratios were >1 and significantly higher (Welch’s t-test, p < 0.01, n ¼ 7) for the three sites closest to the fires (14, 3 and 5) (Table 3). Nevertheless, care should be taken when using such ratios to infer the sources of PAHs as single ratios of these pairs were (1) >0.5 also for the sites most distant to the wildfires (22 and 23), where the PAHs exposure most likely originated from the combustion of liquid fossil fuels (Table 3, Supporting material Table S1) and (2) highly variable for Spanish streams exposed to wildfire emissions (Olivella et al., 2006; Vila-Escale et al., 2007) (own calculations, see Supporting material Table S3).
4.3.
Effects on the invertebrate community
The determination of effects on ecological communities is often based on biotic indices and multiple indices have been established for aquatic invertebrates (Bonada et al., 2006). However, most indices are not capable of differentiating between causes for community change (Bonada et al., 2006; Liess et al., 2008). Biotic indices relying on ecological and/or physiological traits were recently introduced and have demonstrated their capability in selectively identifying effects of specific stressors (Beketov and Liess, 2008; Doledec and Statzner, 2008; Liess et al., 2008). In this study we used the SPEARorganic index (Beketov and Liess, 2008) to identify effects of PAHs or other fire organics on the invertebrate communities and the SIGNAL index (Chessman, 1995) to detect general changes in the community that may have resulted from the fires. The SPEARorganic index showed no change from the beginning of the fires to the post-fire period (Fig. 4) suggesting that organic toxicants including PAH input had no effects on
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 9 0 e4 6 0 0
the communities. Our results are in accordance with a study on the acute toxicity of PAHs and pesticides monitored with passive samplers that found no effects of up to a factor of 100 higher than the PAH concentrations we estimated (Alvarez et al., 2008). Moreover, the concentrations detected in our study were approximately a factor of 1000 or more below the 48-h median lethal concentration (48-h LC50) of PAHs for Daphnia magna (Tables 1 and 3) and to our knowledge no field study has shown effects on invertebrate communities at this low toxicity (Scha¨fer et al., 2007; von der Ohe et al., 2009). Furthermore, widespread sublethal effects at such concentration levels are unlikely given that sublethal effects of aromatic hydrocarbons in aquatic communities were reported to occur a maximum of 24-times below the respective LC50 (Lange et al., 1998). Overall we suggest that the exposure to PAH and other organics related to the wildfire studied had no adverse short-term toxic effects on the macroinvertebrate community. Nevertheless, there may be long-term effects originating from PAHs and other fire organics adsorbed to sediments, which have not been the scope of this investigation (Maltby et al., 1995). In addition, in scenarios of more intensive rainfalls after fires the concentrations of PAHs and other fire organics may be higher and might reach levels of acute effects. The SIGNAL index showed a decrease in ecologically sensitive species in two sampling sites (5 and 13) from the beginning of the fires to the post-fire period (Fig. 4). Site 5 was presumably subject to an input of ash slurry as it received discharge from several smaller streams from the burnt region (5 km downstream of burnt catchment) and the DO saturation dropped in association with rain events. This input may have affected the invertebrate community via the decrease in DO or a rise in sediments, though this remains open to speculation until a thorough investigation is conducted in these single sites.
5.
Conclusions
Silicone-based passive samplers are suitable to monitor organic compounds of a wide-range of polarity and using passive samplers of two different thicknesses is superior to the PRC approach in determining the kinetic regime of a compound after field deployment. The estimated PAH water concentrations in streams in the vicinity of wildfires are of a similar order of magnitude or lower than those in streams in urban areas or in areas with industrialised agriculture. Acute toxicity from PAHs associated with wildfires is presumably not a key culprit for observed changes in aquatic communities following wildfires.
Acknowledgements The authors are grateful to Vincent Pettigrove for deployment of passive samplers at Site 15. We thank two anonymous reviewers for their helpful comments. BJK was funded by an Australian Research Council (ARC) fellowship and LH was funded by an ARC Linkage project (LP0669113). RBS received
4599
financial support through a Deutsche Forschungsgemeinschaft fellowship (SCHA1580/1-1). EnTox is a partnership between Queensland Health and The University of Queensland.
Appendix. Supplementary data Supplementary data associated with this article can be found in online version at doi:10.1016/j.watres.2010.05.044.
references
Allan, I.J., Booij, K., Paschke, A., Vrana, B., Mills, G.A., Greenwood, R., 2009. Field performance of seven passive sampling devices for monitoring of hydrophobic substances. Environmental Science & Technology 43, 5383e5390. Alvarez, D.A., Cranor, W.L., Perkins, S.D., Clark, R.C., Smith, S.B., 2008. Chemical and toxicologic assessment of organic contaminants in surface water using passive samplers. Journal of Environmental Quality 37, 1024e1033. Bartkow, M.E., Hawker, D.W., Kennedy, K.E., Muller, J.F., 2004. Characterizing uptake kinetics of PAHs from the air using polyethylene-based passive air samplers of multiple surface area-to-volume ratios. Environmental Science & Technology 38, 2701e2706. Bauer, U., 2008. Evaluation of Silicone-based Passive Samplers for Monitoring Organic Aquatic Pollutants. PhD thesis, Griffith School of Environment, Griffith University, Brisbane, Australia, p. 210. Beketov, M.A., Liess, M., 2008. An indicator for effects of organic toxicants on lotic invertebrate communities: independence of confounding environmental factors over an extensive river continuum. Environmental Pollution 156, 980e987. BOM, 2009a. Monthly Weather Review: Victoria, February 2009. Bureau of Metereology, Melbourne. BOM, 2009b. Monthly Weather Review: Victoria, January 2009. Bureau of Metereology, Melbourne. BOM, 2009c. Monthly Weather Review: Victoria, March 2009. Bureau of Metereology, Melbourne. Bonada, N., Prat, N., Resh, V.H., Statzner, B., 2006. Developments in aquatic insect biomonitoring: a comparative analysis of recent approaches. Annual Review of Entomology 51, 495e523. Booij, K., Smedes, F., van Weerlee, E.M., 2002. Spiking of performance reference compounds in low density polyethylene and silicone passive water samplers. Chemosphere 46, 1157e1161. Bradstock, R.A., 2008. Effects of large fires on biodiversity in south-eastern Australia: disaster or template for diversity? International Journal of Wildland Fire 17, 809e822. Chessman, B.C., 1995. Rapid assessment of rivers using macroinvertebrates e a procedure based on habitat-specific sampling, family level identification and a biotic index. Australian Journal of Ecology 20, 122e129. Doledec, S., Statzner, B., 2008. Invertebrate traits for the biomonitoring of large European rivers: an assessment of specific types of human impact. Freshwater Biology 53, 617e634. Earl, S.R., Blinn, D.W., 2003. Effects of wildfire ash on water chemistry and biota in South-Western USA streams. Freshwater Biology 48, 1015e1030. EPA, 2003. Rapid bioassessment methodology for rivers and streams. Guideline for environmental management, Victoria, p. 87. Hall, S.J., Lombardozzi, D., 2008. Short-term effects of wildfire on Montane stream ecosystems in the Southern Rocky
4600
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 9 0 e4 6 0 0
mountains: one and two years post-burn. Western North American Naturalist 68, 453e462. Hays, M.D., Fine, P.M., Geron, C.D., Kleeman, M.J., Gullett, B.K., 2005. Open burning of agricultural biomass: physical and chemical properties of particle-phase emissions. Atmospheric Environment 39, 6747e6764. Huckins, J.N., Petty, J.D., Lebo, J.A., Almeida, F.V., Booij, K., Alvarez, D.A., Clark, R.C., Mogensen, B.B., 2002. Development of the permeability/performance reference compound approach for in situ calibration of semipermeable membrane devices. Environmental Science and Technology 36, 85e91. ter Laak, T.L., Busser, F.J.M., Hermens, J.L.M., 2008. Poly (dimethylsiloxane) as passive sampler material for hydrophobic chemicals: effect of chemical properties and sampler characteristics on partitioning and equilibration times. Analytical Chemistry 80, 3859e3866. Lampi, M.A., Gurska, J., McDonald, K.I.C., Xie, F.L., Huang, X.D., Dixon, D.G., Greenberg, B.M., 2006. Photoinduced toxicity of polycyclic aromatic hydrocarbons to Daphnia magna: ultraviolet-mediated effects and the toxicity of polycyclic aromatic hydrocarbon photoproducts. Environmental Toxicology and Chemistry 25, 1079e1087. Lange, R., Hutchinson, T.H., Scholz, N., Solbe, J., 1998. Analysis of the ECETOC aquatic toxicity (EAT) database e II e comparison of acute to chronic ratios for various aquatic organisms and chemical substances. Chemosphere 36, 115e127. Liess, M., Scha¨fer, R.B., Schriever, C.A., 2008. The footprint of pesticide stress in communities-species traits reveal community effects of toxicants. Science of the Total Environment 406, 484e490. Lyon, J.P., O’Connor, J.P., 2008. Smoke on the water: can riverine fish populations recover following a catastrophic fire-related sediment slug? Austral Ecology 33, 794e806. Maltby, L., Boxall, A.B.A., Forrow, D.M., Calow, P., Betton, C.I., 1995. The effects of motorway runoff on freshwater ecosystems: 2. Identifying major toxicants. Environmental Toxicology and Chemistry 14, 1093e1101. Minshall, G.W., 2003. Responses of stream benthic macroinvertebrates to fire. Forest Ecology and Management 178, 155e161. Mu¨ller, J.F., Shaw, G.D., Mortimer, M., Connell, D.W., Sadler, R., 1999. Task Bioaccumulation Toxicants (BT) Final Report. Report No.: S2R2 No 24 South East Queensland Water Quality Strategy. Brisbane City Council, Brisbane. Mu¨ller, J.F., Manomanii, K., Mortimer, M.R., McLachlan, M.S., 2001. Partitioning of polycyclic aromatic hydrocarbons in the polyethylene/water system. Fresenius Journal of Analytical Chemistry 371, 816e822. von der Ohe, P.C., de Deckere, E., Pru¨ß, A., Munoz, I., Wolfram, G., Villagrasa, M., Ginebreda, A., Hein, M., Brack, W., 2009. Toward an integrated assessment of the ecological and chemical status of European river basins. Integrated Environmental Assessment and Management 5, 50e61. von der Ohe, P., Liess, M., 2004. Relative sensitivity distribution (RSD) of aquatic invertebrates to organic and metal compounds. Environmental Toxicology and Chemistry 23, 150e156. Olivella, M.A., Ribalta, T.G., de Febrer, A.R., Mollet, J.M., de las Heras, F.X.C., 2006. Distribution of Polycyclic aromatic hydrocarbons in riverine waters after Mediterranean forest fires. Science of the Total Environment 355, 156e166. Ouyang, G.F., Zhao, W.N., Bragg, L., Qin, Z.P., Alaee, M., Pawliszyn, J., 2007. Time-weighted average water sampling in
Lake Ontario with solid-phase microextraction passive samplers. Environmental Science & Technology 41, 4026e4031. R Development Core Team, 2009. R: a Language and Environment for Statistical Computing, Reference Index Version 2.10.0. R Foundation for Statistical Computing, Vienna, Austria. Scha¨fer, R.B., Caquet, T., Siimes, K., Mueller, R., Lagadic, L., Liess, M., 2007. Effects of pesticides on community structure and ecosystem functions in agricultural streams of three biogeographical regions in Europe. Science of the Total Environment 382, 272e285. Scha¨fer, R.B., Paschke, A., Liess, M., 2008a. Aquatic passive sampling of a short-term thiacloprid pulse with the Chemcatcher: impact of biofouling and use of a diffusionlimiting membrane on the sampling rate. Journal of Chromatography A 1203, 1e6. Scha¨fer, R.B., Paschke, A., Vrana, B., Mueller, R., Liess, M., 2008b. Performance of the Chemcatcher passive sampler when used to monitor 10 polar and semi-polar pesticides in 16 Central European streams, and comparison with two other sampling methods. Water Research 42, 2707e2717. Schletterer, M., Fu¨reder, L., Kuzovlev, V.V., Beketov, M.A., 2010. Testing the coherence of several macroinvertebrate indices and environmental factors in a large lowland river system (Volga River, Russia). Ecological Indicators 10, 1083e1092. Shaw, M., Mueller, J.F., 2009. Time integrative passive sampling: how well do chemcatchers integrate fluctuating pollutant concentrations? Environmental Science & Technology 43, 1443e1448. Smedes, F., 2007. Monitoring of chlorinated biphenyls and polycyclic aromatic hydrocarbons by passive sampling in concert with deployed mussels. In: Greenwood, R., Mills, G.A., Vrana, B. (Eds.), Comprehensive Analytical Chemistry 48: Passive Sampling Techniques in Environmental Monitoring. Elsevier, Amsterdam, pp. 407e449. Vieira, N.K.M., Clements, W.H., Guevara, L.S., Jacobs, B.F., 2004. Resistance and resilience of stream insect communities to repeated hydrologic disturbances after a wildfire. Freshwater Biology 49, 1243e1259. Vila-Escale, M., Vegas-Vilarrubia, T., Prat, N., 2007. Release of polycyclic aromatic compounds into a Mediterranean creek (Catalonia, NE Spain) after a forest fire. Water Research 41, 2171e2179. Vrana, B., Popp, P., Paschke, A., Schuurmann, G., 2001. Membrane-enclosed sorptive coating. An integrative passive sampler for monitoring organic contaminants in water. Analytical Chemistry 73, 5191e5200. Vrana, B., Allan, I.J., Greenwood, R., Mills, G.A., Dominiak, E., Svensson, K., Knutsson, J., Morrison, G., 2005. Passive sampling techniques for monitoring pollutants in water. TrAC-Trends in Analytical Chemistry 24, 845e868. Wennrich, L., Vrana, B., Popp, P., Lorenz, W., 2003. Development of an integrative passive sampler for the monitoring of organic water pollutants. Journal of Environmental Monitoring 5, 813e822. Yuan, H.S., Tao, S., Li, B.G., Lang, C., Cao, J., Coveney, R.M., 2008. Emission and outflow of polycyclic aromatic hydrocarbons from wildfires in China. Atmospheric Environment 42, 6828e6835. Yunker, M.B., Macdonald, R.W., Vingarzan, R., Mitchell, R.H., Goyette, D., Sylvestre, S., 2002. PAHs in the Fraser River basin: a critical appraisal of PAH ratios as indicators of PAH source and composition. Organic Geochemistry 33, 489e515.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 5 0 e4 5 5 8
Available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/watres
Various physico-chemical stress factors cause prophage induction in Nitrosospira multiformis 25196- an ammonia oxidizing bacteria Jeongdong Choi, Shireen M. Kotay, Ramesh Goel* Department of Civil and Environmental Engineering, University of Utah, Salt Lake City, USA
article info
abstract
Article history:
Bacteriophages are virsues that infect bacteria and contribute significant changes in the
Received 7 December 2009
overall bacterial community. Prophages are formed when temperate bacteriophages
Received in revised form
integrate their DNA into the bacterial chromosome during the lysogenic cycle of the phage
5 April 2010
infection to bacteria. The prophage (phage DNA integrated into bacterial genome) on the
Accepted 27 April 2010
bacterial genome remains dormant, but can cause cell lysis under certain environmental
Available online 7 May 2010
conditions. This research examined the effect of various environmental stress factors on the ammonia oxidation and prophage induction in a model ammonia oxidizing bacteria
Keywords:
Nitrosospira multiformis ATCC 25196. The factors included in the study were pH, tempera-
N. multiformis
ture, organic carbon (COD), the presence of heavy metal in the form of chromium (VI) and
Bacteriophage
the toxicity as potassium cyanide (KCN). The selected environmental factors are
Mitomycin-C
commonly encountered in wastewater treatment processes, where ammonia oxidizing
Induction
bacteria play a pivotal role of converting ammonia into nitrite. All the factors could induce
Ammonia oxidation
prophage from N. multiformis demonstrating that cell lysis due to prophage induction
Chromium
could be an important mechanism contributing to the frequent upset in ammonia
Cyanide
oxidation efficiency in full scale treatment plants. Among the stress factors considered, pH
Temperature
in the acidic range was the most detrimental to the nitrification efficiency by N. multi-
COD
formis. The number of virus like particles (VLPs) increased by 2.3Eþ10 at pH 5 in 5 h under acidic pH conditions. The corresponding increases in VLPs at pH values of 7 and 8 were 9.67Eþ9 and 1.57Eþ10 in 5 h respectively. Cell lysis due to stress resulting in phage induction seemed the primary reason for deteriorated ammonia oxidation by N. multiformis at lower concentrations of Cr (VI) and potassium cyanide. However, direct killing of N. multiformis due to the binding of Cr (VI) and potassium cyanide with cell protein as demonstrated in the literature at higher concentrations of these toxic compounds was the primary mechanism of cell lysis of N. multiformis. Organics represented by the chemical oxygen demand (COD) did not have any effect on the phage induction in N. multiformis. This AOB remained dormant at low temperature (4 C) without any phage induction. Significant decrease in the number of live N. multiformis cells with a corresponding increase in the number of VLPs was recorded when the temperature was increased to 35 C. Death of N. multiformis at 45 C was attributed to the destruction of cell wall rather than to the phage induction. Published by Elsevier Ltd.
* Corresponding author. Fax: þ1 801 585 5477. E-mail address:
[email protected] (R. Goel). 0043-1354/$ e see front matter Published by Elsevier Ltd. doi:10.1016/j.watres.2010.04.040
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 5 0 e4 5 5 8
1.
Introduction
Bacteriophages are virsues that infect bacteria. Bacteriophages are known to be very important components of freshwater and marine bacterial communities (Suttle, 2006; Wommack and Colwell, 2000). World’s oceans are estimated to contain greater than 1 1029 phages in total (Breitbart and Rohwer, 2005). On the other hand, activated sludge systems have been shown to contain 108e109 phages per ml (Ewert and Paynter, 1980; Otawa et al., 2007), a number comparable to or greater than the number of phages found in most of the aquatic systems. Even though the role of bacteriophages in changing bacterial community in marine environment is well understood, the role they play in maintaining any bacterial population in activated sludge systems is not properly understood. Several research attempts (Farahbakhsh and Smith, 2004; Khan et al., 2002a,b; Havelaar et al., 1991; Ewert and Paynter, 1980; Hantula et al., 1991; Hertwig et al., 1999; Thomas et al., 2002) have been made on the diversity and physiological identity of phages in activated sludge processes, but these studies limited their scope to hetrotrophic population. Two categories of bacteriophages are common; virulent and temperate. Virulent phages infect and kill the host and replicate often 100’s of progeny (daughter phages) in a short time called “lytic infection”. On the other hand, temperate phages can recombine with host cell DNA during “lysogenic infection” forming a dormant prophage (phage DNA integrated into bacterial genome) where viral genes that are detrimental to bacterial cells are not expressed (Casjens, 2003) and are incorporated into bacterial genome. Temperate phages can provide necessary virulence and fitness factors affecting cell metabolism, bacterial adhesion, colonization, immunity, destruction of competing bacteria, antibiotic resistance and serum resistance (Wagner and Waldor, 2002). It is speculated that more than 80% of bacterial strains contain prophages and phage DNA can contribute to as much as 10e20% of bacterium’s genome (Canchaya et al., 2003). Ammonia nitrogen (NH3eN) removal in wastewater treatment plants is accomplished through biological oxidation of ammonia by ammonia oxidizing bacteria (AOBs). AOBs are chemolithoautotrophic organisms which catalyze the biochemical oxidation of NH3eN to nitriteenitrogen (NO2eN). AOBs are extremely slow growers and their growth is sensitive to many environmental factors such as pH, dissolved oxygen, heavy metals and toxic chemicals (Shammas, 1986; Dangcong et al., 2000, Blum and Speece, 1991). Effective ammonia oxidation requires a high solid retention time (SRT) in order to ensure a healthy population of AOBs in the system and to prevent biomass washout (Rittmann and McCarthy, 2001). Incomplete ammonia oxidation has been attributed to low specific activities of nitrifying bacteria as a result of toxic inhibition by chemicals or ammonia concentration in the influent wastewater (Burgess et al., 2002). The failure of ammonia oxidation has also been attributed to a low population or the absence of AOBs (Wanner et al., 2004). One of the possibilities often overlooked about the poor ammonia oxidation in treatment plants is the phage infection to AOBs, either by lysogenic or lytic viral attack. Lysogeny has been extensively studied in heterotrophic bacteria but not for ammonia oxidizing community. Recently,
4551
prophageebacterial interactions have received tremendous attention due to the significant role that they play in the microbial ecology of various bio-systems. Researchers have also tried to study the effect of environmental factors on phage induction in natural bacterial populations (McDaniel and Paul, 2005). Environmental conditions affect the switch between the lytic and lysogenic life styles of the well-studied temperate Escherichia coli phage (Echols, 1972; Herskowitz and Hagen, 1980). It has been shown for bacterial communities that environmental pollutants can be more efficient inducing agents than universal inducer such as mitomycin-C (Cochran et al., 1998). Due to rapid industrialization, municipal wastewater treatment plants frequently receive heavy metals and other toxic substances. Negative effects of pH shift from the suggested optimum values, presence of heavy metals and substances causing toxicity on the performance of ammonia oxidizing community are well established (You et al., 2009; Park and Ely, 2008). Lysogeny and induction of prophages into the lytic cycle in lysogenic bacteria are common in the environment (Ackerman and Dubow, 1987). Lysogeny due to the induction of prophages can be caused by several environmental factors. Lysogeny can adversely affect the bacterial population and in turn the process efficiency. However, the significance of lysogeny in ammonia oxidizing bacteria due to pH shift, presence of heavy metals and toxicity has not been studied. There has been very little research to date on lysogeny in autotrophic AOBs. In this research, we studied the effect of selected physical and chemical factors on the lysogeny of AOBs with Nitrosospira multiformis (ATCC # 25196) as the model AOB. Several studies have shown that genus Nitrosospira is the dominant terrestrial population among the three AOB genera: Nitrosomonas, Nitrosospira, and Nitrosococcus (Avrahami and Bohannan, 2007). The stain N. multiformis ATCC 25196 have been extensively studied and its genomes have been completely sequenced (Norton et al., 2008). From the genomic data, the strain was established to contain prophage. Taking into consideration the importance of the strain and preestablished presence of prophage in N. multiformis genome, this AOB was suitable for studying phage induction and its effect on ammonia oxidation under the applied environmental stresses. The performance of N. multiformis for ammonia oxidation and phage induction (viral abundance) under the effect of stresses caused due to changes in pH, the presence of heavy metals and toxic substances was investigated in this research are reported here. To the best of our knowledge, no such study has been conducted in the past which investigated the AOB performance affected by phage induction due to environmental stresses. The present study was an endeavor in this direction to understand the influence of various stress factors on phage induction in a model AOB, N. multiformis ATCC 25196.
2.
Materials and methods
2.1. Bacterial strain, culture conditions and experimental setup N. multiformis ATCC 25196T was grown and maintained lithoautotrophically in ATCC medium 929 at 28 C in the dark as described previously (Norton et al., 2002). ATCC medium 929
4552
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 5 0 e4 5 5 8
contained NH3eN. Hence the batch reactors were not supplemented with NH3eN. The NH3eN concentration in the bulk liquid during all batch tests was 351 19 mg/L. Cultures grown for 48 h (4eþ10 cells/mL) were used for all experiments. Experimental setup used for stress experiments is depicted in Fig. 1. Batch reactors containing 200 mL N. multiformis culture and supplied with regulated amount of filter sterilized air were agitated using magnetic stirrer. pH was maintained using sterile 1 N HCl and 1 N NaOH solutions. Samples were aseptically collected at 0 and 5 h for monitoring the performance of batch reactor in terms of ammonia oxidation and prophage induction. Collected samples were immediately processed for subsequent analysis for NH3eN, NO2eN, viral and bacterial enumeration. AOBs are shown to be very sensitive to pH changes, the presence of heavy metals and toxic substances (You et al., 2009; Stasinakis et al., 2003). Hence, three stress factors viz., pH, Cr (VI) as heavy metal and potassium cyanide (KCN) as toxic substance were considered in the study. Furthermore, temperature and the presence of organics (represented by COD) can also affect the performance of AOBs, these factors were also considered in this study. The pH values of 5, 7 and 8 were considered to evaluate the effect of pH on phage induction. Since Cr and KCN were toxic, sub values of corresponding minimum inhibitory concentration (MIC) reported in the literature were considered. To study the effect of fluctuations in chemical oxygen demand (COD) on prophage induction in the model AOB, glucose
concentrations corresponding to COD values of 100e400 mg/L were supplemented to the N. multiformis liquid cultures. Similarly, experimental temperature range of 4 Ce45 C was considered to study the effect of temperature fluctuations on bacteriophages induction in N. multiformis.
2.2.
To induce prophage from N. multiformis, 15 mL of 1 mg/mL mitomycin was added to 15 mL of fresh culture of N. multiformis. The mixture was incubated overnight on a gyratory shaker. Afterwards, the mixture was processed for phage extraction and enumeration. The sample was filtered through 0.22 mm pore size filter paper (Millipore Co. Bedford, MA) to remove the bacterial debris. 900 mL of this filtered sample was transferred to a fresh and pre-autoclaved glass tube containing 100 mL of the reaction buffer and 2 mL of RNase free DNase I (Invitrogen, CA). After DNase treatment, samples were incubated at room temperature for 20 min and 35 mL of 0.5 M EDTA solution was added thereafter to stop the DNase activity.
2.3. Epifluorescent microscopy (EFM) and viral enumeration Aliquots of DNase treated sample (100 mL) were suspended in 900 mL of sterile deionized water and were vacuum filtered through a stack of 25 mm filters consisting of 0.02 mm Anodisc (Whatman Int’l Ltd., Maidstone, England), a 0.22 mm Durapore membrane filters (Millipore, Ireland), and a glass fiber prefilter (Millipore, Ireland). Anodisc containing captured virus like particles (VLPs) were stained by adding 10 SYBR Gold dye (Invitrogen Co.). Anodisc was incubated for 20 min in the dark and analyzed by EFM with BX 51 microscopy (Olympus, Japan) using a Cy3 filter. Pictures were digitally captured at a magnification of 1000X with a DPI-71 camera. VLPs were enumerated manually from the micrographs using glass slide with grid.
2.4.
Fig. 1 e Schematic of Batch Reactor Used for Stress Tests and Ammonia Oxidation by N. multiformis. In the schematic, (1) Aquarium pump to supply air, (2) pH probe, (3) cable to the pH controller, (4) pressure release tube, (5) 0.22 micron filter, (6) acid or base tube, (7) glass bottle, (8) growth media, (9) magnetic stir bar, (10), magnetic stir plate, (11) feed tube.
Mitomycin-C induction and phage extraction
Transmission electron microscopy
Mitomycin-C induced 500 mL culture of N. multiformis was centrifuged at 4500 g for 20 min to pallet down the cell debris. The supernatant containing the induced prophages was further centrifuged at 8300 g overnight at 4 C. The phage particles were purified by isopycnic centrifugation at 115,000 g for 3 h through cesium chloride gradient. 5 mL of the purified phage extract was loaded on a 400 grid formvar coated copper grids (Fisher Scientific) and allowed to settle for 1 min. Excess liquid was soaked by holding bibulous paper (Fisherbrand) at 90 to the grid. The grids were stained for 1 min using 2 mL freshly prepared and filtered 1% uranyl acetate and excess stain was soaked by holding bibulous paper at 90 to the grid. Dried grids were subsequently examined under Tecnai T12 Transition Electron Microscope (FEI, Japan). The accelerating voltage used for imaging was 80 kV and images of negatively stained phage particles were recorded.
2.5.
Bacterial enumeration using live/dead assay
The samples (1 mL) for bacterial enumeration were filtered through 0.22 mm (PCTE balck, GE Water & Process
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 5 0 e4 5 5 8
Technologies) using vacuum manifold. The bacteria captured on the membrane filters were stained with Baclightä bacterial viability kit (Molecular Probes Inc.), incubated in dark for 20 min and analyzed under BX 51 microscope (Olympus, Japan) using Cy3 and a FITC filters to capture live and dead cells respectively. At each parameter tested, multiple pictures were captured and each picture was divided into four compartments using a grid system. Number of live and dead cells in each compartment was manually counted and was averaged. The percentage increase or decrease in the number of cells was calculated based on the average values.
2.6.
Other analytical methods
Ammonia (NH3) concentration was analyzed spectrophotometerically using HACH reagent kit (DR 5000, HACH). 10020 (Chromotropic Acid method).
Mitomycin-C has been employed in previous studies as a positive control for prophage induction (Cochran et al., 1998)). Mitomycin-C was added to a batch containing freshly grown N. multiformis culture. The effect of mitomycin-C on N. multiformis with time is shown in Fig. 2. The bright big dots shown by yellow arrows in all micrographs are multiformis cells and the tiny small dots shown by red arrows are VLPs. An increase in viral population and a decrease in N. multiformis population were observed over the time (Fig. 2aed). Lysis is clearly evident comparing the micrographs 2a (at 0 time) and 2d (after 24 h) in Fig. 2. Over 90% of the N. multiformis population was dead after 24 h following the induction with mitomycin-C. Transmission electron micrography (TEM) was performed on the induced phages and the TEM micrographs are shown in Fig. 3. The induced phages belong to the family podoviridae. Brightly stained capsids of diameter 70e85 nm without any tails were observed enclosing darkly stained DNA within them.
3.2.
3.
4553
Effect of pH
Results and discussions
A blast search at GenBank database using terminase gene revealed that N. multiformis contained prophage with greater than 25 kb sequence length. Terminase gene synthesizes terminase protein which helps bacteriophages in DNA packaging (Casjens, 2008). Because of the presence of prophage element on the genome of N. multiformis, this AOB was considered for stress experiments as a model AOB. The performance of the bacteria for ammonia oxidation and phage induction under pH, temperature and COD changes, due to the presence of Cr (VI) and toxic potassium cyanide were investigated. It is highlighted here that the purpose of this research work was not to evaluate the inhibitory effects of these selected stress factors on AOB kinetics. The effects of toxic substances, heavy metals and other operational factors have been thoroughly researched in the past (You et al., 2009; Neufeld et al., 1986). This manuscript details one of the mechanisms (phage induction) that may be responsible for inhibitory effects of the studied stress factors. In other words, in this manuscript, we have tried to answer the question like “why AOB activity is inhibited in the presence of certain stress factors which have been shown to be important by the past researchers”
3.1. Prophage induction of N. multiformis using mitomycin-C Recent studies have confirmed that N. multiformis (ATCC 25196T) has two prophages integrated within its chromosome (Norton et al., 2008). It is also an established fact that under the influence of environmental stress factors, prophages can be induced (coming out of the chromosome) causing cell lysis of the host organism. The process of induction releases phages in the surrounding environment. Mitomycin-C is an universal prophage inducing agent used widely to induce prophages. Mitomycin-C was used in the present study to confirm the induction of prophage of N. multiformis to ensure that the prophage of N. multiformis is not defective or is in a state of mutational decay. Mitomycin-C is not naturally found in the wastewater treatment systems and therefore was suitable surrogate for environmental stress factor on N. multiformis.
The optimum pH condition for the ammonia oxidation by N. multiformis ranges from 6.0 to 7.3 with an optimum value of 7 (Norton et al., 2002). Performance of N. multiformis for the ammonia oxidation and phage induction was tested at lower and higher pH values with respect to the neutral pH. N. multiformis cell viability, number of VLPs, ammonia disappearance and nitrite production were monitored and the data is shown in Fig. 4. The number of VLPs increased at pH 5 and 8, indicating prophage induction and the induction was more prominent at pH 5. The number of VLPs increased by 2.3Eþ10 in 5 h as compared to the VLP number at 0 time when the pH was 5. Similarly, the VLPs increased by 9.67Eþ9 and 1.57Eþ10 in 5 h at pH 7 and 8 respectively. Although the increase in VLPs at pH 5 and 8 was almost 2 and 1.5 fold more respectively than the number of VLPs at pH 7 (control), it was surprising to see phage induction at the pH 7. VLPs were monitored at the beginning and at the end without any intermediate monitoring. It is also noticeable from Fig. 4b that dead cells were present in all batch reactors at different pH value at the beginning. Although the increase in the number of dead cells at pH 7 from time zero to time 5-h was 37% and considerably less than the corresponding numbers at pH values of 5 (150%) and 8 (121%), this small decrease must have contributed to the increase in VLPs in the batch at pH 7. Furthermore, a net increase of almost 29% in the number of live cells was recorded at pH 7 because of the growth of N. multiformis. On the other hand, 64% and 36% decrease in the number of live cells were recorded at pH values of 5 and 8 respectively. Assuming first order degradation, the rate of NH3eN oxidation rate was 3.62 mg N L1h1 at pH 7. Considerable decrease in ammonia oxidation rates was observed at pH values of 5 and 8. The NH3eN oxidation rates were 1.9 mg N L1h1 and 1.5 mg N L1h1 at pH values of 5 and 8 respectively. All batch experiments at different pHs were originally started from the same source. Hence, the number of live cells was nearly same in all batches. The increase in VLPs and the decrease in viable N. multiformis cells at pH 5 and 8 are concomitant with the corresponding decrease in ammonia oxidation rates at these pH values with respect to N. multiformis performance at the pH 7. The process of ammonia oxidation leads to a net acidification of the environment. Where ammonia deposition and
4554
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 5 0 e4 5 5 8
Fig. 2 e Epiflorescent VLPs micrographs on N. multiformis induced by mitomycin-C. The micrographs show N. multiformis (yellow arrows) and VLPs (red arrows) in the mitomycin induced sample after (a) at the beginning, (b) after 3 h, (c) after 12 h and, (d) after 24 h. The 10 red circles represent VLPs released after cell lysis due to prophage induction with mitomycin-C. The scale bar represents 50 mm.
nitrification rates are high, this may contribute to the lowering of the environmental pH (Beiderbeck et al., 1996; Kowalchuk and Stephen, 2001). Chemolithoautotrophic nitrifying bacteria like N. multiformis, which catalyzes the first step of ammonia to nitrite are known to be sensitive to low pH values. Bacterial growth of ammonia oxidizing bacteria (AOB) is usually limited at pH of 5.8 (Watson et al., 1989) and their activity cease typically below pH of 5.5 (Hayatsu and Kosuge, 1993) in liquid pure culture. Even though the failure of AOB to cope with acidic condition is thought to be the unavailability of the substrate with decreasing pH values, lysogenicelytic conversion of phage life cycle can be the reason for failure of AOB growth and ammonia oxidation as evident by results from this research. High pH values above 8 are also shown to be detrimental for AOBs (Painter, 1986). Adverse effects of either low or higher pH than the optimal pH were observed in this research where more VLPs, slow ammonia oxidation by N. multiformis and greater number of dead cells were observed at pH values of 5 and 8. These results suggest that fluctuations in pH may act as potential stress factor for AOBs resulting in prophage induction. The results also project that prophage induction is more imminent in acidic pH than in basic pH.
3.3.
Effect of Cr(VI)(heavy metal)
The influence of heavy metals on the performance of conventional activated sludge processes (You et al., 2009; Stasinakis et al., 2003) and nitrification (Park and Ely, 2008)
has been widely studied. In this study, chromium was used as a model heavy metal to evaluate its effect on the performance of N. multiformis as an AOB. Chromium is usually released from chromium electroplating and chemical industry in the oxidation state of III and VI in the environment, which eventually ends up in wastewater systems unless removed. Cr (VI), which is more powerful as a toxic agent than Cr (III) and, can be a more potent inhibitor. Once inside the cell, Cr (VI) could attack DNA, proteins, and membrane lipids, thereby disrupting cellular integrity and functions. The DNA damage triggers a SOS response which eventually may cause induction of prophage (Meyn et al., 1977). In the current experimental scenario, five different concentrations of Cr (VI) ranging from 0.002 mM to 1 mM were used to evaluate their effect on N. multiformis performance and its phage induction. The results for number of VLPs and cell viability are shown in Fig. 5. At Cr (VI) concentrations of 0.002 mM, 0.01 mM and 0.1 mM, the number of VLPs increased by 5.00Eþ09, 6.33Eþ09 and 10.30Eþ09 respectively as depicted in Fig. 5a. The corresponding NH3eN oxidation rates were 2.7, 2.1 and 1.3 N L1h1 at Cr (VI) concentrations of 0.002, 0.01 and 0.1 mM respectively. It can be seen that NH3eN oxidation rates decreased with increasing concentrations of Cr (VI). These concentrations also caused significant decrease in live N. multiformis cells as shown in Fig. 5b. At Cr (VI) concentrations of 0.002 mM, 0.01 mM and 0.1 mM, decreases in the number of live cells were 11, 33 and 54% respectively. The corresponding increases in the number of dead cells were 26, 73 and 203% at Cr (VI) concentrations of 0.002 mM, 0.01 mM
4555
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 5 0 e4 5 5 8
Fig. 3 e Transmission electron micrographs showing morphology of bacteriophages obtained after induction with mitomycin-C. Bars represent 100 nm.
and 0.1 mM respectively. These results suggest that the number of live cells decreased and the number of dead cells increased with the corresponding increase in Cr(IV) concentrations. These results are in agreement with those obtained by Stasinakis et al. (2003). These researchers demonstrated that nitrification efficiency decreased with increasing concentration of Cr (VI) in laboratory scale reactors containing heterotrophs and ammonia oxidizers. Increase in the number
of induced prophages (in terms of VLPs) also suggests that heavy metals can trigger prophage induction in AOBs and in general all bacteria, which in turn may initiate the lysogenic cycle of hostephage resulting in killing of bacteria. At higher concentrations of Cr (VI) beyond 0.1 mM, the increase in the number of VLPs dropped significantly along with a significant decrease in the number of live cells. This
2e+10
2e+10
1e+10
1.0e+10 8.0e+9 6.0e+9 4.0e+9 2.0e+9 0.0 0.002 mM 0.01 mM
5e+9 pH 5
0.5 mM
Live (0 hr) Live (5 hrs) Dead (0 hr) Dead (5 hrs)
4e+10
1 mM
8e+10 Live (0 hr) Live (5 hrs) Li Dead ad (0 hr)) Dead (5 hrs) s)
8e+10
6e+10
6e+10
4e+10
2e+10
2e+10
0 pH 5
pH 7
pH 8
Fig. 4 e Changes in (a) number of virus like particles and (b) number of live and dead cells at different pH values.
m
M
M m 5 0.
m M 1 0.
M m 01
2 00 0.
0
m
m
M
M
0
0.
C ells/m L
b
pH 8
Cells/ml
b
pH 7
0.1 mM
1
3e+10
1.2e+10
Increase in VLPs (per mL)
a I n cr ea se in VL P s (p er m L )
a 1.4e+10 3e+10
Fig. 5 e Changes in (a) number of virus like particles and (b) number of live and dead cells at different Cr (VI) concentrations.
4556
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 5 0 e4 5 5 8
of prophages in N. multiformis. Concentration of 0.15 mM resulted in highest induction of prophages with nearly 2 fold increase when compared with results using 0.015 mM KCN (Fig. 6a). At 0.3 mM concentration, both live and dead bacterial populations dropped to zero after 5 h exposure and low number of VLPs were recorded (Fig. 6a). Consequently, the rates of ammonia oxidation were affected as KCN concentrations increased and were almost completely inhibited at 0.3 mM KCN concentration.
a 1.8e+10 Increase in VLPs (per mL)
1.6e+10 1.4e+10 1.2e+10 1.0e+10 8.0e+9 6.0e+9 4.0e+9
3.5.
Effect of COD and temperature
3.5.1.
Effect of COD
2.0e+9 0.0 0.015 uM
b
0.075 uM
0.15 uM
8e+10
Live (0 hr) Live (5 hr) Dead (0 hr) Dead (5 hr)
6e+10
Cells/ml
0.3 uM
4e+10
2e+10
0 0 uM
0.015 uM 0.075 uM 0.15 uM
0.3 uM
Fig. 6 e Changes in (a) number of virus like particles and (b) number of live and dead cells at different cyanide concentrations.
contradicts the fact that a decrease in live cells should cause a corresponding increase in the number of VLPs as was seen at other lower concentrations of Cr (IV). This could be explained in light of the mechanisms of cell lysis caused at higher concentrations of Cr (VI). There are two primary mechanisms for cell lysis; (1) lysis due to the direct toxic effect of Cr (VI) on the cell wall where this heavy metal binds onto functional groups of proteins and inactivating them, rapidly bringing metabolism to a standstill (Talaro, 2005) and, (2) lysis due to the induction of intracellular prophage because of the stress generated by the presence of heavy metal. At concentrations above 0.1 mM, the Cr(VI) was too toxic for the cells resulting in leakage of cell contents and cell lysis was dominated by the first mechanism. However, at lower concentrations of Cr (VI), cell lysis was primarily caused by induction of prophages and as a result, an increase in the number of VLPs was observed at lower Cr (VI) concentrations. At 1.0 mM complete inhibition of N. multiformis metabolism and consequently no ammonia oxidation was observed.
3.4.
Effect of KCN (toxicity effect)
The results of exposure of N. multiformis to various concentrations of potassium cyanide, a known toxic substance to the bacterial cell, are shown in Fig. 6. Four different concentrations ranging from 0.015 mM to 0.3 mM were used. Exposure to various concentrations of KCN also resulted in the induction
Under the influence of COD fluctuations, there was no increase in the number of VLPs observed suggesting no prophage induction. Furthermore, ammonia oxidation rates showed 2.4 mgNL1hr1 and 3.6 mgNL1hr1 at COD concetrations of 300 mg/L and 400 mg/L, respectively implying no negative effect of high COD concentrations on AOBs. However, in actual wastewater treatment plants, fluctuations in COD are known to effect AOB population and nitrification in general. It may therefore be speculated that higher biodegradable organic COD directly promotes growth of heterotrophic bacterial populations but, doesn’t directly influence AOBs. The subsequent increase in heterotrophic population may indirectly influence AOB population by competing for space, oxygen and other nutrients (Rittmann and McCarthy, 2001).
3.5.2.
Effect of temperature
The growth of N.multiformis was inhibited at 4 C. However, there was no significant changes in VLPs, suggesting that no prophage induction or lysogeny occurred at this low temperature. However, at 35 C significant decrease in live N. multiformis cells and significant increase in VLPs was observed. The total number of N. multiformis cells decreased by 2.30Eþ10 when the temperature was raised from 4 C to 35 C. The corresponding increase in number of VLPs was 6.50Eþ10. At much high temperature of 45 C, the growth of bacterial cells was inhibited by heat shock or heat inactivation of cellular enzymes and proteins. As a results, although decrease in the number of live N. multiformis cells was recorded at 45 C, increase in the number of VLPs was minimal. This suggests that the death of N. multiformis cells at 45 C was not due due to lysogeny as a result of prophage induction. Even though, there is evidence that high temperature (42 C) treatment of mutants lysogenic bacteria leads to prophage induction and release of bacteriophage (Schuster et al., 1972), it was more likely that there is no prophage induction at 45 C occurred in the present study as no significant increase in VLPs was detected. Ammonia oxidation did not occur at 4 C and 45 C as it was apparent from a no increase in nitriteenitrogen.
4.
Conclusions
The overall effect, as observed from the experiments was that prophage was induced under all tested environmental factors from N. multiformis indicating that lysogeny could be an important pathway for the occasional deterioration in ammonia oxidation efficiency in full scale treatment plants.
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 5 0 e4 5 5 8
N. multiformis cells responded to stress caused by pH changes, heavy metal and toxic substance through prophage induction. More VLPs were recorded under acidic pH conditions than under basic pH values when compared to the VLPs number at the neutral pH. During the process of ammonia oxidation to nitrite, alkalinity in form of carbonated or bicarbonates is consumed which causes the system pH to go down and the ammonia oxidation to slow down. Evidence that low pH induces prophage in ammonia oxidizers is novel. Although the effect of Cr (VI) and Cyanide (Vazquez et al., 2006; Kelly et al., 2004 and Neufeld et al., 1986) has been fully studied on the process of nitrification, their effect on phage induction has not been evaluated. Both Cr (IV) and KCN caused significant induction in N. multiformis even at concentrations recommended as minimum inhibitory values. These results suggest that these toxic compounds can not only inhibit the activity of AOBs through their toxic effect but also put some sort of environmental stress that ultimately leads to prophage (if present) induction in these organisms. Higher concentrations of Cr (VI) and KCN completely inhibited ammonia oxidation by N. multiformis. Kim et al. (2008) observed significant reduction in nitrification efficiency by free cyanide in an anoxiceaerobic activated sludge system. Kim et al. (2008) also demonstrated that ammonia nitrogen as high as 350 mg/L was not inhibitory for ammonia oxidizers. Hence, there was no possibility of phage induction due to inhibitory effect of ammonia. It was interesting to note that, among the factors tested pH has more stronger inducing ability than Cr(IV) and KCN. This can be further advocated by the fact that disturbance in performance of nitrification in wastewater treatment plants due to the fluctuations in pH is more common than attributed for presence of heavy metals like Cr(IV) or chemicals like KCN. Nevertheless, further studies are essential to establish the precise mechanism behind induction of prophages under the influence of stress factors. Also, it is necessary to investigate how the released phages can affect other microbial communities in wastewater systems. This study used a pure culture of an AOB. AOBs exist inside the flocs in activated sludge processes. Although, the research presented in this manuscript is a paradigm shift from the current understanding on the inhibitory effects of stress factors on AOBs, more experiments will be needed to evaluate the effect of these stress factors on AOBs in a complex environment such as the one that exists in activated sludge systems.
references
Ackerman, H.W., DuBow, M.S., 1987. Viruses of Prokaryotes: General Properties of Bacteriophages. CRC Press, Boca Raton, Florida. Avrahami, S., Bohannan, B.J.A., 2007. Response of Nitrosospira sp. strain AF-like ammonia oxidizers to changes in temperature, soil moisture content, and fertilizer concentration. Appl. Environ. Microbiol. 73, 1166e1173. Beiderbeck, V.O., Campbell, C.A., Ukrainetz, H., Curtin, D., Bouman, O.T., 1996. Soil microbial and biochemical properties after ten years of fertilization with anhydrous ammonia. Can. J. Soil Sci. 76, 7e14.
4557
Blum, D.J.W., Speece, R.E., 1991. A database of chemical toxicity to environmental bacteria and its use in interspecies comparisons and correlations. J. Water Pollut. Control Fed. 63, 198e207. Breitbart, M., Rohwer, F., 2005. Here a virus, there a virus, everywhere the same virus? Trends in Microbiol. 13, 278e284. Burgess, J.E., Stuetz, R.M., Stehpeoson, T., 2002. Dinitrogen oxide detection for process failure early warning systems. Water Sci. Technol. 45, 247e254. Canchaya, C., Proux, C., Fournous, G., Bruttin, A., Bru¨ssow, H., 2003. Prophage genomics. Microbiol Mol. Biol. Rev. 67, 238e276. Casjens, S.R., 2008. Diversity among the tailed-bacteriophages that infect the enterobacteriaceae. Research in Microbiol. 159 (5), 340e348. Casjens, S., 2003. Prophages and bacterial genomics: what have we learned so far? Mol. Microbiol. 49 (2), 277e300. Cochran, P.K., Kellogg, C.A., Paul, J.H., 1998. Prophage induction of indigenous marine lysogenic bacteria by environmental pollutants. Mar. Ecol. Prog. Ser. 164, 125e133. Dangcong, P., Bernet, N., Moletta, R., 2000. Effects of oxygen supply methods on the performance of a sequencing batch reactor for high ammonium nitrification. Water Environ. Res. 72, 195e200. Echols, H., 1972. Developmental pathways for the temperate phage: lysis vs lysogeny. Annu. Rev. Genet. 6, 157e190. Ewert, D.L., Paynter, M.J., 1980. Enumeration of bacteriophages and host bacteria in sewage and the activated sludge treatment process. Appl Environ Microbiol. 39, 576e583. Farahbakhsh, K., Smith, D.W., 2004. Removal of coliphages in secondary effluent by microfiltrationdmechanisms of removal and impact of operating parameters. Water Res. 38 (3), 585e592. Hantula, J., Kurki, A., Vuoriranta, P., Bamford, D., 1991. Ecology of bacteriophages infecting activated sludge bacteria. Appl Environ Microbiol. 57, 2147e2151. Havelaar, A.H., Butler, M., Farrah, S.R., Jofre, J., Marques, E., Ketratanakul, A., Martins, M.T., Ohgaki, S., Sobsey, M.D., Zaiss, U., 1991. Bacteriophages as model viruses in water quality control. Water Res. 25 (5), 529e545. Hayatsu, M., Kosuge, N., 1993. Autotrophic nitrification in acid tea soils. Soil Sci. Plant Nutr 39, 209e217. Herskowitz, I., Hagen, D., 1980. The lysis-lysogeny decision of phage lambda: explicit programming and responsiveness. Annu. Rev. Genet. 14, 399e455. Hertwig, S., Popp, A., Freytag, B., Lurz, R., Appel, B., 1999. Generalised transduction of small yersinia enterocolitica plasmids. Appl Environ Microbiol. 65, 3862e3866. Khan, M.A., Satoh, H., Katayama, H., Kurisu, F., Mino, T., 2002a. Bacteriophages isolated from activated sludge processes and their polyvalency. Water Res 36, 3364e3370. Khan, M.A., Satoh, H., Mino, T., Katayama, H., Kurisu, F., Matsuo, T., 2002b. Bacteriophageehost interaction in the enhanced biological phosphate removing activated sludge system. Water Sci Technol 46, 39e43. Kelly II, R.T., Henriques, I.D.S., Love, N.G., 2004. Chemical inhibition of nitrification in activated sludge. Biotechnol. Bioeng 85, 683e694. Kim, Y.M., Park, D., Lee, D.S., Park, J.M., 2008. Inhibitory effects of toxic compounds on nitrification process for cokes wastewater treatment. J. Hazard. Mater. 152, 915e921. Kowalchuk, G.A., Stephen, J.R., 2001. Ammonia oxidizing bacteria: a model for molecular microbial ecology. Annu. Rev. Microbiol. 55, 485e529. McDaniel, L., Paul, J.H., 2005. Effect of nutrient addition and environmental factors on prophage induction in natural populations of marine Synechococcus species. Appl Environ Microbiol. 71, 842e850.
4558
w a t e r r e s e a r c h 4 4 ( 2 0 1 0 ) 4 5 5 0 e4 5 5 8
Meyn, M.S., Rossman, T., Troll, W., 1977. A protease inhibitor blocks SOS functions in Escherichia coli: antipain prevents lambda repressor inactivation, ultraviolet mutagenesis, and filamentous growth. Proc Natl Acad Sci U S A. 74, 1152e1156. Neufeld, R., Greenfield, J., Rieder, B., 1986. Temperature, cyanide and phenolic nitrification inhibition. Water Res. 20, 633e642. Norton, J.M., Klotz, M.G., Stein, L.Y., Arp, D.J., Bottomley, P.J., Chain, P.S., Hauser, L.J., Land, M.L., Larimer, F.W., Shin, M.W., Starkenburg, S.R., 2008. Complete genome sequence of Nitrosospira multiformis, an ammonia-oxidizing bacterium from the soil environment. Appl. Environ. Microbiol. 74, 3559e3572. Norton, J.M., Alzerreca, J.J., Suwa, Y., Klotz, M.G., 2002. Diversity of ammonia monooxygenase operon in autotrophic ammoniaoxidizing bacteria. Arch Microbiol. 177, 133e149. Otawa, K., Lee, S.H., Yamazoe, A., Onuki, M., Satoh, H., Mino, T., 2007. Abundance, diversity, and dynamics of viruses on microorganisms in activated sludge processes. Microb Ecol 53, 143e152. Painter, H.A., 1986. Nitrification in the treatment of sewage and wastewaters. Spec. Publ. Soc. Gen. Microbiol. 20, 185e213. Park, S., Ely, R.L., 2008. Candidate stress genes of Nitrosomonas europaea for monitoring inhibition of nitrification by heavy metals. Appl. Environ. Microbiol. 74, 5475e5482. Rittmann, B.E., McCarthy, P.L., 2001. Environmental Biotechnology: Principles and Applications. McGraw-Hill, New York. Schuster, H., Beyersmann, D., Mikolajczyk, M., Schlicht, M., 1972. Prophage induction by high temperature in thermosensitive DNA mutants lysogenic for bacteriophage lambda. J. Virol. 11 (6), 879e885.
Shammas, N.K., 1986. Interactions temperature, ph, and biomass on the nitrification process. J. Water Pollut. Control Fed. 58, 52e58. Stasinakis, A.S., Thomaidis, N.S., Mamais, D., Papanikolaou, E.C., Tsakon, A., Lekkas, T.D., 2003. Effect of chromium (VI) addition on the activated sludge process. Water Res. 37, 2140e2148. Suttle, C.A., 2006. Viruses in the sea. Nature 437, 356e361. Talaro, K.P., 2005. Foundations in Microbiology, fifth ed.. McGraweHill, New York, NY. Thomas, J.A., Soddell, J.A., Kurtboke, D.I., 2002. Fighting foam with phages? Water Sci Technol 46, 511e518. Vazquez, I., Rodrıguez, J., Maranon, E., Castrillon, L., Fernandez, Y., 2006. Study of aerobic biodegradation of cokewastewater in a two and three-step activated sludge process. J. Hazard. Mater. 137, 1681e1688. Wagner, P.L., Waldor, M.K., 2002. Bacteriophage control of bacterial virulence. Infect. Immun. 70 (8), 3985e3993. Wanner, J., Ruzickova, I., Krhutoka, O., Probyl, M., 2004. Activated sludge population dynamics and wastewater treatment plant design and operation. Water Sci. Technol. 41, 217e225. Watson, S.W., Bock, E., Harms, H., Koops, H.-P., Hooper, A.B., 1989. Nitrifying bacteria. In: Staley, J.T., Bryant, M.P., Pfennig, N., Holt, J.G. (Eds.), Bergey’s Manual of Systematic Bacteriology, vol. 3. The Williams and Wilkins Co., Baltimore, Md, pp. 1808e1834. Wommack, K.E., Colwell, R.R., 2000. Virioplankton: viruses in aquatic ecosystems. Microbiol. Mol. Biol. Rev. 64, 69e114. You, S.J., Tsai, Y.P., Huang, R.-. Y., 2009. Effect of heavy metals on nitrification performance in different activated sludge process. J. Hazard. Mater. 165, 987e994.
