Environmental sampling for trace analysis

Environmental Sampling for Trace Analysis Edited by Bernd Markert

0 VCH Verlagsgesellschaft mbH, D-69451 Weinheim (Federal Republic of Germany), 1994

Distribution:

VCH, P. 0. Box 10 1 1 61 D-69451 Weinheim, Federal Republic of Germany Switzerland: VCH, P. 0. Box, CH-4020 B a d , Switzerland United Kingdom and Ireland: VCH, 8 Wellington Court, Cambridge CB1 lHZ, United Kingdom USA and Canada: VCH, 220 East 23rd Street, New York, NY 10010-4606, USA

Japan: VCH, Eikow Building, 10-9 Hongo 1-chorne, Bunkyo-ku, Tokyo 113, Japan ISBN 3-527-30051-1 (VCH, Weinheim)

Environmental Sampling for Trace Analysis Edited by Bernd Markert

4b

VCH

Weinheim - New York Base1 - Cambridge - Tokyo

Editor: Prof. Dr. Bernd Markert Internationales Hochschulinstitut Zittau Markt 23 D-02763 Zittau

This book was carefully produced. Nevertheless, authors, editor and publisher do not warrant the information contained therein to be free of errors. Readers areadvised to keep in mind that statements, data, illustrations, procedural details or other items may inadvertently be inaccurate.

Published jointly by VCH Verlagsgesellschaft mbH, Weinheim (Federal Republic of Germany) VCH Publishers Inc., New York, NY (USA) Editorial Director: Dr. Hans-Joachim Krdus Production Manager: Dipl.-Wirt.-Ing. (FH) H.-J. Schmitt

Cover illustration: Heterogenous distribution of trace substances on the “xerothermic hills” near Wiirzburg, F. R.C..

Library of Congress Card No.: applied for

British Library Cataloguing-in-Publication Data: A catalogue record for this book is available from the British Library

Die Deutsche Bibliothek - CIP-Einheitsaufnahmc: Environmental sampling for trace analysis / ed. by Bernd Markert. -Weinheim ; New York ; Basel : Cambridge ; Tokyo : VCH, 1YY4 ISBN 3-527-30051-1 NE: Markcrt, Bcrnd [Hrsg.]

0 VCH Verlagsgesellschaft mbH, D-69451 Weinheim (Federal Republic of Germany), 1994

Printed on acid-free and chlorine-frec paper All rights reserved (including those of translation into other languages). N o part of this book may be reproduced in any form - by photoprinting, microfilm, or any other means - nor transmitted or translated into a machine language without written permission from the publishcrs. Registered names, trademarks, etc. used in this book, even when not specifically marked as such are not to be considcred unprotected by law. Composition, Printing. and Bookbinding: Druckhaus ,,Thomas Miintzer“ GmbH, D-99947 Bad LangcnSdZd

Printed in the Fcdcral Rcpublic of Germany.

Environmental analysis has entered a phase which is difficult to describe precisely because of its meteoric development. The advent of more and more modern systems of instrumental analysis now makes it possible to penetrate areas of ecology and ecotoxicology that would have been considered inaccessible a few years ago. Trace analysis and ultra-trace analysis of individual substances, in particular, have proved to be a useful dynamo for new scientific findings. One example is the development of possible concentration and action models for specific dioxin molecules which it would have been impossible to investigate without highly efficient analytical procedures. In addition to detecting lower and lower concentrations, such highly sophisticated measuring equipment has the important advantage of opening up totally new fields of work. Procedures such as multielement analysis, chemical fingerprinting and non-target screening are already familiar to many laboratories and will continue to pave the way for trendsetting research strategies. But orientation towards more and more efficient analytical methods also involves a risk that is all too easily overlooked. Too little attention is often given to analytical steps before and after actual instrumental measurement - sampling and the preparation of specimens, for example - because the measuring work itself is the focus of interest. Representative sampling, especially, has not kept pace with the development of increasingly sensitive analytical systems. Today it has to be said that the greatest error in the overall result of the analytical process is usually caused by improper sampling. This development becomes obvious when we compare the almost uncountable number of publications on the measurement of any arbitrary substance X in sample Y with the small number of scientific publications dealing with careful representative sampling suitable for trace and ultra-trace analysis. The ratio of publications is probably somewhere around 1000: 1. The reason for this development is clear. First of all, sampling is only a small step in the overall process of analysis and one that is unlikely to produce any “spectacular” data for a subsequent discussion of the overall result. This means that sampling is all too easily dismissed as uninteresting. Secondly, the equipment it requires is often less expensive than the apparatus needed for the instrumental measurement itself. In many cases a rigid PVC spade is all that is needed to acquire a soil specimen, so that manufacturers do not at once consider it lucrative to orient their product range towards sampling systems. Sampling as a field of work in its own right is therefore very quickly deemed unattractive from both the scientific and the commercial point of view. So it is not surprising that progress towards quality control in representative environmental sampling lags far behind that in actual instrumental mcasurcmcnt, where remarkable steps towards “true data” have

VI

Prc,fute

been made as a result of precise definition of the terms accuracy, reproducibility or concentration-dependence of the accuracy of the analytical results (“Horwitz Trumpet”), and the production of suitable reference materials. The results arc serious: the data from individual study groups cannot be compared with each other; harmonization of research in the various fields, so often demanded, is difficult to achieve even at the stage where the data are generated. A further problem is that the first principle of sampling, namely that “the sample taken from the system should have exactly the same chemical composition as the original material” cannot be followed even approximately in the field. It should never be forgotten that environmental processes are continuous in terms of space and time. Watercourses, for example, demonstrate most plainly that sampling has to be carried out in a time/space continuum if the findings are to be relevant. Many investigations, even recent ones, fail to take this into account sufficiently. In order to achieve meaningful and comparable results in respect of the current state of the environment it is necessary to develop standards for sampling similar to those that already exist in Germany for the water, sewage and sludge sector (German Standard Procedure for Water, Sewage and Sludge Analysis). By developing suitable guidelines at a national and, still more important, at an international level it must then be ensured that the src-ond and third principles of sampling can be implemented. These are: “The probability of being selected from a total population must be equal for each individual”, and “the amount of work required for sampling increases with the degree of dispersion of the individuals and the number of such individuals”. One must, however, bear in mind that practical difficulties during the actual sampling process make it virtually impossible to follow the first two principles. A sample taken out in the field can never have exactly the same chemical composition as the original material; at best it will be very similar. One of the reasons is that often only a tiny fraction of the original material is actually analyzed (e.g., 100mg of 100 kg of leaves from a forest ecosystem). The initial content of the material can be altered by contamination or volatilization of individual constituents while the sample is being taken and during transportation. Moreover, it is scarcely possible to give each individual the same likelihood of being selected when such individuals are diffusely distributed over the ecosystem. The objective must rather be to come as close as possible to the first two principles by means of a carefully prepared sampling strategy. Some practical, basic rules may be helpful here: Avoid contaminating the sample in any way with the equipment used, the containers, or by the person taking the sample. - Avoid any volatilization of chemical compounds as a result of microbial activity, absorption by the walls of the vessels in which the samples are kept or overheating of the samples during transportation and storage. - Take reasonably large samples, provided that there is enough material in the system and this is not subject to the nature conservation laws. - Take account of seasonal fluctuations in the composition of the original material and other parameters affecting its overall composition such as temperature, humidity, light etc.

-

Prefuce

VII

The aim of this book is to provide an overview of the techniques commonly used at present for taking many different kinds of environmental samples. Because of the host of different environmental samples a selection had to be made, which was painful in some cases but unavoidable in view of the limited printing capacity. The subject of human samples has been left out completely, for it is a domain of the medical profession and analysts with the relevant training. In water sampling emphasis has been put on fresh water; the sampling of “wet depositions” such as rain and snow has not been included, for excellent monographs already exist in this field. Most animals have also had to be excluded. Only ants are used as an example of how sampling might be conducted. The editor was not authorized by the Federal Office of the Environment to publish even an extract from the sampling guidelines of the “Environmental Specimen Programme” of the Federal Ministry for Research and Technology. This is a pity, especially with respect to long-term investigations, for the results are important and valuable. It remains to be hoped that these Standard Operating Procedures of the Federal Office of the Environment will soon be made accessible to a wide circle of readers in the form of a monograph. No attempt has been made to standardize the terminology used in sampling. The reason is that there is still a great need to establish unequivocal technical terms, especially in German and American English. In spite of the deficiencies of this book the editor and the authors agree that together they have been able to reflect the state of the art in a field which will have to undergo much more intensive development in the future. Again, this book would not have been come about if friends and associates had not created the scope necessary for its realization, from the initial idea to the finished product. The central figures were of course the authors, who once again succeeded in submitting, amending and improving the manuscripts within a year in spite of their numerous other commitments. The suggestions for improvements came from a multitude of associates in Germany and abroad and from my employees and associates at the GKSS Research Centre, Geesthacht. I was motivated not least by Mrs Vera Weckert of the study group on systems research at the University of Osnabriick, who missed no opportunity of urging me on and encouraging me in difficult phases. I also wish to thank VCH Publishers, especially Dr. Kraus, and Wirt.-Ing. Hans-Jochen Schmitt, for their generous and friendly assistance. The editor and the authors hope that the book will be disseminated widely and that the problems it discusses will be examined closely; they themselves will be pleased to listen to constructive criticism. Magdeburg, January 1994

Bernd Markert

Outline

List of Contributors XXVll

Part 1:

Historical Aspects

1

1 History of Sampling Demonstrated on the Ore Mining Industry Empirical and Theoretical Approaches 3 C. Kruff

Part 11: General Aspects 9 2 Gencral Aspects of Environmental Sampling 11 P . Hoffhiunn 3 Trace Elements Need Trace Analysis 73 I . Puis 4 Error Estimation in Environmental Sampling and Analysis M . H . Rumsey 5 Estimation of Varying Detection Limits 109 W. G. Wurren

Part 111: Examples for Sampling A. Air 123

123

6 Particle and Gas Measurements on Filters J . G. Wutson, C. C l i o ~ ~ 7 Organic Gas Sampling 163 B . Zielinsku, E. Fzrjita

B. Water

93

125

185 8 Sampling of Freshwaters for Estimation of all Detectable Elements 187 U. M . Cowgill 9 Guidelines for Sampling Freshwater for Eutrophication Management Programs 203 H . Klupper, W. Rust, D . Ulilmunn 10 The Sampling Strategy in the River Elbe - Experiences 223 H . Gtllir, E. Weher

X

Outline

1 1 Sampling Treated Wastewaters and Receiving Streams 249 J . E. Norris I2 Water and Wastewater Sampling for Environmental Analysis 255 E. M . Dick 13 Sampling of Groundwater for General Quality Monitoring 279 V. Schenk 14 Groundwater Sampling for Metals 287 R. Puls

C. Soils and Sediments 303 15 Representative Soil Sampling 305 0. Franzle 16 Problems and Results in the Development of International Standards for Sampling and Petreatment of Soils 321 A . Paeiz, G. Criflmann 17 Fixed and H ypothcsis-Guided Soil Sampling Methods - Principles, Strategies, and Examples 335 R. W. Scholz, N. Nothbaum, T. W. Muy 18 Sampling for Trace Analysis of Lake Sediments 347 U. M . Contgill 19 Sampling Design for Studying the Relationships between Heavy Metals in Soils, Sediments, and Discharged Wastewaters 365 Zueng-Sung Chen

D. Plants and Animals 379 20 Sampling of Plants for Environmental Trace Analysis in Terrestrial, Semiterrestrial and Aquatic Environments 38 1 W. H . 0. Ernst 21 On the Sampling of Vascular Plants for Monitoring of Heavy Metal Pollution 395 R. Djingova, I. Kuk# 22 Sampling of Terricolous Lichen and Moss Species for Trace Element Analysis with Special Reference to Bioindication of Air Pollution 41 5 Z . Tubu, Z . Csintalun, Z.Nugy, K. Szente, Z.Takacs 23 Comparative Investigation of the Distribution of Chemical Elements in an Aceri Tatarico-Quercetum Plant Community and in Stands of Cultivated Plants 435 M . Kozxics, K. Penksza, G . Turcsunyi, L. Kaszuh, S. Tbth, P. Szbke

Outline

XI

24 Sampling of Tropical Terrestrial Plants with Particular Reference to the Determination of Trace Elements 443 R. Jayasekeru 25 Sampling in the Stemflow and Throughfall Areas of Forests 449 G. Turcsanyi, K. Penksza, I . Siller, E. Fuhrer, S . Thth, M . Kovuc.~, S. Biittner 26 Sampling of Different Social Categories of Red Wood Ants (Formica s. str.) for Biomonitoring 465 V. Mauvara, A.-J. Martin, A . Oja, P. Nuorteva

Part IV: Literature Survey 491 27 Overview of References for Sampling and Related Topics 493 S. Hannappel

Index

511

Contents

List of Contributors XXVII Part I:

Historical Aspects 1

1

History of Sampling Demonstrated on the Ore Mining Industry - Empirical and Theoretical Approaches 3 G. Kruft

Part 11: General Aspects 9 2

2.1 2.2 2.3 2.3.1 2.3.2 2.3.3 2.3.4 2.3.5 2.3.6 2.3.7 2.3.8 2.4 2.4.1 2.4.2 2.5 2.6 3 3.1 3.1.1 3.1.2 3.1.3 3.2 3.2.1 3.2.2

General Aspects of Environmental Sampling 1 1 P. Hoffkunn Introduction 1 1 Terms and Definitions 12 Aspects of Sampling 13 Location, Place, and Position of Sample Removal Size, Quantity, and Volume of the Sample 14 Number of Samples to be Taken 14 Time, Duration, and Frequency of Sampling 15 Homogeneity of the Sample 16 Contamination of the Sample 17 Losses in the Sample 18 Sample Storage and Conservation 19 Guidelines and Norms 19 Quality Assurance 20 Environmental Protection 2 1 References 22 Appendix 23 Trace Elements Need Trace Analysis 73 I . Pui.5 Problems of Trace Element Analysis 73 Introduction 73 Problems of Biological Analysis 74 Indicator Organs in Biological Evaluation 76 The Importance of Trace Elements 77 Introduction 77 Criteria of Essentiality and Beneficiality 80

13

3.2.3 3.2.4 3.2.5 3.2.6 3.2.7 3.3 4 4. I 4.2 4.3 4.4 4.4.1 4.4.2 4.4.3 4.4.4 4.4.5 4.5 4.6 4.7 5

5.1 5.2 5.2.1 5.2.2 5.3 5.4 5.5 5.6

Critcria of Toxicity 81 Changes in Element Concentrations 83 The Importance of Trace Elements in the Environment Interactions between Differcnt Elements 86 The Importancc of Interdisciplinary Trace Element Research 88 References 89

Error Estimation in Environmental Sampling and Analysis 93 M . H . RLirnsey Introduction 93 Basic Concepts and Terminology 93 Sampling Error in Contcxt 95 Methods for Estimating Quality of Measurements 96 Measuring Analytical Precision 96 Measuring Sampling Precision 99 Targets for Acceptable Levels of Prccision in Sampling and Analysis 102 Measuring Analytical Bias 103 Estimating Sampling Bias 105 Targets for Acceptable Levels of Bias in Sampling and Analysis 106 Conclusions 107 References 107 Estimation with Varying Detection Limits W. G. Warren Introduction 109 Methodology 110 The One-Dimensional Case 1 10 The Bivariate Case 112 Examples 113 Discussion 1 18 References 1 18 Appendix 118

Part 111 : Examples for Sampling A. Air 123 6 6.1 6.2

84

109

123

Particle and Gas Measurements on Filters J. G . Watson, J . C. Chow Introduction 125 Filter Analysis Methods 126

125

Outline

6.2.1 6.2.2 6.2.3 6.2.4 6.3 6.4 6.4.1 6.4.2 6.4.3 6.4.4 6.4.5 6.5 6.6 6.7

Mass 126 Elements 129 Water Soluble Ions 130 Organic and Elemental Carbon 132 Filter Media 134 Aerosol Sampling Systems 140 Size-Selective Inlets 142 Sampling Surfaces 146 Filter Holders 146 Pumps and Flow Controllers 147 Sampler Configurations 148 Sampling and Analysis Procedures 150 Summary 152 References 153

7

Organic Gas Sampling 163 B. Ziefinsku, E. Fujitu Introduction 163 Whole-Air Sampling 163 Sampling Media Selection and Preparation 164 Sampling 165 Storage and Transport 167 Preconcentration Methods 168 Preconcentration on Nonselective Solid Adsorbents 168 Sampling Media Selection and Preparation 168 Sampling 172 Storage and Transport 172 Selective Methods of Compound Preconcentration 172 Sampling Media Selection and Preparation 173 Sampling 175 Storage and Transport 175 Semi-Volatile Organic Compounds 175 Passive Sampling Techniques 177 Summary 178 References 18 1

7.1 7.2 7.2.1 7.2.2 7.2.3 7.3 7.3.1 7.3.1.1 7.3.1.2 7.3.1.3 7.3.2 7.3.2.1 7.3.2.2 7.3.2.3 7.4 7.5 7.6 7.7

B. Water 8

8. I 8.2

XV

185 Sampling of Freshwaters for Estimation of all Detectable Elements 187 U. M . Cowgill Introduction 187 Problems Associated with Sampling 187

8.2.1 8.2.2 x.2.3 8.2.4 8.2.5 8.3 8.3.1 8.3.2 8.4 8.4. I 8.4.2 8.5 8.6 9

9.1

9.2 9.3 9.3. I 9.3.2 9.3.3 9.4 9.5 9.6

Contamination from Sampling Devices and Laboratory Equipment 188 Sorption and Leaching of Pollutants by Sampling Tool Materials 189 Replication I90 Frequency of Sampling 192 Equipment, Field, and Sampling Blanks 193 Synple Fractions and Sample Preservation 194 Preservation of Samples 194 Pretreatment, Storage, and General Precautions 194 Sampling of Lakes. Rivers, and Groundwater 196 Stratified Bodies of Water 196 Unstratitied Bodies of Water 196 Sampling on Ice, Snow, Rain, Dew, and Fog 198 References 20 1 Guidelines for Sampling Freshwater for Eutrophication Management Programs 203 H . Klupper, W. Rast, D. Uhln2cinn Introduction 203 What to Sample 203 Necessary Temporal and Spatial Resolution for Data 206 Where to Sample 206 When to Sample 209 Sampling Strategies in Waterbodies with Longitudinal Water Quality Gradients 210 Calculating the Costs of Sample Collection 213 Compilation and Presentation of Data 21 8 References 220

The Sampling Strategy in the River Elbe - Experiences 2 2 3 H . Guhr, E. W(1ber 10.1 Characteristics of the River Elbe 223 10.1.1 Types of Use 223 10.1.2 Pollution Loads 223 10.1.2.1 Municipal Sewage Discharges 226 10.1.2.2 Discharges from Agriculture 227 10.1.3 Characteristics of Water Quality 227 10.2 Development of a Monitoring Strategy for the Rivcr Elbe Objectives 228 10.2.1 Selection of Sampling Points 229 10.2.2 Range of Measured Variables 230 10.2.3 Measuring Frequency 23 I 10

Outline

10.2.4 10.2.5 10.2.6 10.3

XVII

Use of Automatic Monitoring Stations 231 Data Flow, Data Processing, and Evaluation 232 Special Investigations 233 Experience Made in Implementing the Sampling Strategy Representativeness of Sampling Points 234 10.3.1 Frequency of Measurements 235 10.3.2 Data Collection 237 10.3.2.1 Sampling 237 10.3.2.2 Preparation and Preservation of Samples 240 10.3.2.3 Chemical Analysis 240 10.3.2.4 Measuring Errors 241 10.3.3 Consideration of Sewage Discharges 243 10.3.4 Data Processing and Evaluation 245 Conclusions 247 10.4 References 247 10.5

11 11.1 11.2 11.3 11.4 11.5 11.6

12

12.1 12.2 12.3 12.3.1 12.3.2 12.3.3 12.3.4 12.3.5 12.3.6 12.3.7 12.3.8 12.3.9 12.4 12.5 12.6

~

Sampling Treated Wastewaters and Receiving Streams 249 J . E. Norris Introduction 249 Sediment Sampling 249 Fish Sampling 250 Sampling of Industrial Wastewater Discharges 25 1 Sampling of Surface Waters: Receiving Streams 253 References 254 Water and Wastewater Sampling for Environmental Analysis 255 E. M . Dirk Introduction 255 Why Sample Water? 255 Elements of the Sampling Plan 256 Sampling Objective 256 Sampling Location 257 Sample Types and Collection Techniques 257 Sampling Equipment 257 Sample Containers and Sample Preservation 258 Sample Labeling and Shipping 258 Types of Analyses 258 Chain-of-Custody Documentation 258 Quality Assurance and Quality Control 259 Types of Samples 259 Sampling Programs 26 1 Sampling Equipment 262

XVIII

Outline

12.6.1 12.6.2 12.6.3 12.6.4 12.6.5 12.6.6 12.7 12.7.1 12.7.2 12.7.3 12.8 12.8.1 12.8.2 12.8.3 12.9 12.9.1 12.9.2 12.9.3 12.9.4 12.9.5 12.9.6 12.10

Power Source 264 Electronic Controller 264 Sample Intake 265 Sample Transport Line 265 Sample Storage 266 Sample Delivery System 268 Pumps and Representative Samples 269 Peristaltic Pumps 269 Vacuum Pumps 270 Bladder Pumps 271 Advancements in Sampling 272 Volatile Organic Sampling 272 Sample Volume Accuracy 272 Refrigeration 273 Preserving Sample Integrity 274 Sampling Equipment 274 Sampling Containers 275 Sample Handling 275 Sample Preservation 275 Sample Holding Time 275 On-Site Analysis 276 Conclusion 278

Sampling of Groundwater for General Quality Monitoring 279 V . Schmk Introduction 279 13.1 Sampling of Groundwater 279 13.2 13.2.1 Requirements for Sampling Sites 279 13.2.2 Sampling Equipment for Groundwater 28 1 Activities at the Sampling Site 282 13.3 13.3.1 Determinations and Preservations 282 13.3.2 Transport and Storage 283 Sampling Programs and Contents of Analyses 283 13.4 Interpretation 285 13.5 Conclusions 285 13.6

13

14 14.1 14.2 14.3 14.4

Groundwater Sampling for Metals 287 R. W. P U / S Introduction 287 Sampling Objectives 287 Sampling Point Design 288 Monitoring Well Development 289

Outline

14.5 14.6 14.7 14.8 14.8.1 14.8.2 14.8.3 14.8.4 14.9 14.10 14.1 1

XIX

Colloidal Transport 290 Well Turbidity 291 Sampling Preparation 293 Purging and Sampling 294 Low Flow Purging 294 Isolation of the Sampling Zone 295 Water Quality Indicator Parameters 296 Sampling Materials 298 Filtration and Analysis 298 Summary 299 References 300

C. Soils and Sediments 303 15

Representative Soil Sampling 305 0.Franzle 15.1 Selection of Representative Soil Samples 305 15.1.1 Methodology of Statistical Design 305 15.1.2 Small-Scale Variability of European Soils in the Light of Frequency Statistics 306 15.1.2.1 Determination of Regionally Representative Soils in Germany by Means of Crosstabulation and Neighborhood Analysis 306 15.1.2.2 Selection of Representative European Soils for SorptionTesting Purposes 309 15.1.3 Large-Scale Soil Variability in the Light of Variogram Analysis 310 15.1.3.1 Variogram Analysis 3 1 1 15.1.3.2 Model Applications of Variogram Analysis 3 15 15.2 Conclusions 3 19 15.3 Summary 319 15.4 References 320 16

16.1 16.2 16.3 16.4 16.5 16.6 16.7 16.8

Problems and Results in the Development of International Standards for Sampling and Pretreatment of Soils 321 A . Paetz, B. CroJmann Introduction 321 What is Soil? 322 Soil Sampling Objectives 323 Requirements on Sampling 323 Preliminary Investigation 325 Selection of Sampling Patterns 325 Sampling Depth 327 Sample Quantity 328

xx

0u t h e

16.9 16.10 16.11 16.12 16.13 16.14 16.15 16.16

Single Sample or Composite Samples 328 Preservation of Soil Samples 329 Use of Appropriate Sampling Tools and Containers 329 Sampling Report 330 Quality Control 330 Pretreatment of Soil Samples 33 1 Summary 333 References 333

17

Fixed and Hypothesis-Guided Soil Sampling Methods Principles, Strategies, and Examples 335 R . W . Scholz, N . Nothbuum, T. W. May Introduction 335 “Fixed” Sampling Plans/Grid Plans 338 Soil Sampling Plans Guided by Hypotheses 342 The IHEARU Schema 343 Conclusions 344 References 345

17.1 17.2 17.3 17.4 17.5 17.6 18 18.1 18.2 18.2.1 18.2.2 18.2.3 18.2.4 18.2.5 18.3 18.4 18.5 18.6

Sampling for Trace Analysis of Lake Sediments 347 U . M . Cowgill Introduction 347 Sampling Devices 348 Grab Samplers 349 Corers 352 Sampling Devices for the Collection of Suspended Sediments 355 Mud-Water Interface Samplers 355 Sediment Pore Water Sampling 356 Subsampling of Sediment Grabs and Cores 358 Quality Control 359 Statistical Considerations 36 1 References 362

Sampling Design for Studying the Relationships between Heavy Metals in Soils, Sediments, and Discharged Wastewaters 365 Zucng- Sung Clien Introduction 365 19.1 Case I : Studies of Rice-Growing Soils near Chemical 19.2 Plants 365 19.2.1 Principles of Sampling Design 365 19.2.2 Sampling Methods 367 19.2.2.1 Soil Sampling 367 19.2.2.2 Sampling Design for Water and Sediments 368

19

Outline

XXI

19.2.3 Analytical Methods 368 19.2.4 The Distribution of Heavy Metals in Soils, Discharged Water, and Sediments 368 19.2.5 The Relationship between Heavy Metals in Soils, Discharged Water, and Sediments 371 Case 11: Studies of Rice-Growing Soils near an Industrial 19.3 Park 372 19.3.1 Principles of Sampling Design 372 19.3.2 Sampling Methods 373 19.3.2.1 Soil Sampling 373 19.3.2.2 Sampling Design of Water and Sediments 373 19.3.3 Analytical Methods 373 19.3.4 The Distribution of Heavy Metals in Soils, Discharged Water, and Sediments 374 19.3.5 The Relationship between Heavy Metals in Soils, Discharged Water, and Sediments 376 19.4 Conclusions 377 19.5 References 377

D. Plants and Animals 379 20

20.1 20.2 20.3 20.4 20.4.1 20.4.2 20.4.3 20.5 20.6 20.7 21

21.1 21.2 21.3

Sampling of Plants for Environmental Trace Analysis in Terrestrial, Semiterrestrial and Aquatic Environments 38 1 W. H. 0. Ernst Introduction 38 1 What Do we Want to Know? 381 Are Trace Elements Relevant Objectives? 382 The Exploration of Environmental Compartments 384 The Hygro- and Hydrophytes 384 The Epiphytes 385 Soil Exploring Plants 386 Plant Parts and Life History 388 Conclusion 390 References 39 1 On the Sampling of Vascular Plants for Monitoring of Heavy Metal Pollution 395 R. Djingova, I. K u k f Introduction 395 Concentration Differences at Biological Levels 396 Seasonal Variations in the Heavy Metal Content of Plants 401

XXII

Outline

21.4 21.5 21.6

Sample Pretreatment Discussion 410 References 4 12

403

Sampling of Terricolous Lichen and Moss Species for Trace Element Analysis, with Special Reference to Bioindication of Air Pollution 415 Z . Tubu, Z . Csintulan, Z . Nagy, K . Szcnte, Z . Tak6cs Introduction 4 15 22.1 Materials and Methods 417 22.2 22.2.1 The Species of Investigation 417 22.2.2 The Original Habitat 417 22.2.3 The Cryptogam Transplantation Technique 417 22.2.4 Exposition 41 8 22.2.5 Sampling 418 22.2.6 Preparation of the Samples for Measuring Trace Element Contents 419 22.2.7 Element Analysis 419 22.2.8 Data Analysis 4 19 22.3 Results and Discussion 419 22.3.1 Vertical Distribution of Trace Elements within the Various Lichen and Moss Parts 419 22.3.2 Horizontal Distribution of Trace Elements within the Thalli of C. furcutu and the Cushions of T. rurulis 423 22.3.3 Distribution of Trace Element Contents in Parts of the Lichen Thalli and Moss Cushions Compared to Whole Thalli and Cushions, Respectively 424 22.3.4 Variability of the Trace Element Contents in the Lichen Thalli and Moss Cushions 426 22.3.5 Trace Element Contents in the Soil Substrate of the Lichen Thalli and Moss Cushions 426 22.3.6 The Influence of Sample Washing on Trace Element Contents 429 22.4 Summary 431 22.5 References 433 22

23

23.1 23.2

Comparative Investigation of the Distribution of Chemical Elements in an Aceri tatarico-Quercetum Plant Community and in Stands of Cultivated Plants 435 M . Kovucs, K . Penkszu, G . Turcsanyi, L. Kuszab, S . Toth, P . Szbke Introduction 435 Material and Methods 435

23.3 23.4 23.5

Results and Discussion 436 Summary 442 References 442

24

Sampling of Tropical Terrestrial Plants with Particular Reference to the Determination of Trace Elements 443 R. Juyasekera Introduction 443 Statistical Aspects 444 Plant Sampling 444 Concluding Remarks 447 References 447

24.1 24.2 24.3 24.4 24.5 25

25.1 25.2 25.2.1 25.2.2 25.2.3 25.2.4 25.2.5 25.2.5. I 25.2.5.2 25.2.5.3 25.2.5.4 25.2.6 25.2.7 25.2.8 25.3 25.4 25.5 25.6 26

26.1 26.2 26.3

Sampling in the Stemflow and Throughfall Areas of Forests 449 G. Turcshnyi, K. Penksza, I. Siller, E. Fuhrer, S . Toth, M . Kovucs, S. Buttner Introduction 449 Literature Data 449 The Amount of Stemflow and Throughfall in Stands of Different Tree Species 449 Chemical Composition of Stemflow and Throughfall 45 1 Physical Changes in Soils Due to Stemflow 452 Chemical Changes in Soils Due to Stemflow 452 Living Organisms Influenced by Stemflow 453 Microorganisms 453 Fungi 453 Mosses and Lichens 454 Other Plants 454 Impact of Stemflow on the Roots of Trees 454 Impact of Stemflow on Animals 454 Some Contradictions 455 Material and Methods 455 Results and Discussion 456 Conclusions 459 References 460 Sampling of Different Social Categories of Red Wood Ants (Formica s. str.) for Biomonitoring 465 V . Muavaru, A.-J. Martin, A . Oja, P. Nuorteva The Role of Ants in Nature 465 Pollutant Accumulation in Ants 466 Definition of the Red Wood Ants 468

XXIV

Outline

26.4 Structure of the Nests and Territories 469 26.4.1 The Nest Mound 469 26.4.2 The Base of the Nest 470 26.4.3 The Underground Part of the Nest 470 26.4.4 Thc Foraging Territory 471 26.4.5 Super- and Substructures 471 26.5 Social Structure of Ant Colonies 472 26.5.1 Sexuals 472 26.5.2 Workers 473 26.5.2.1 Reserve Workers 473 26.5.2.2 Nurses 474 26.5.2.3 Inside Workers 474 26.5.2.4 Outside Workers 47.5 26.6 Sampling 475 26.6.1 What Kind of Nest Mounds is Suitable for Sampling? 475 26.6.2 Sample Taking 476 26.6.3 Sampling Objects 477 26.6.3.1 Foragers Leaving the Nest 477 26.6.3.2 Foragers Traveling to the Nest 477 26.6.3.3 Surface Workers from the Top of the Nest 477 26.6.3.4 Workers and Brood from the Brood Chamber 478 26.6.3.5 Reserve Ants 478 26.6.3.6 Nest Material 478 26.6.4 Collection of Background Data and the Order of Sampling 479 26.6.4.1 Traffic Density of Foragers on Ant Roads 479 26.6.4.2 Ant Activity on the Nest Surface 479 26.6.4.3 Temperature of the Nest 479 26.6.4.4 Sample Taking from Ant Roads 479 26.6.4.5 Measuring the Size of thc Nest Mound 480 26.6.4.6 Sampling from the Nest 480 26.7 Feeding Experiments 480 26.7.1 Honey as a Feeding Substrate 480 26.7.2 Fish as a Feeding Substrate 481 Description of some Pilot Studies already Performed 48 1 26.8 26.8.1 Natural Cd Levels in Different Castes and Worker Groups 481 26.8.2 Cd Transfer to Ant Colonics under Pollution Stress 483 26.8.3 Artificial Cadmium Feeding Experiments 483 26.8.4 Effects of Metdl Pollution on the Enzyme Balance 486 References 486 26.9

Outline

XXV

Part IV: Literature Survey 491 27 27.1 27.2 27.2.1 27.2.2 27.2.3 27.2.4 27.2.5 27.2.6 27.2.7 27.3 27.3.1 27.3.2 27.3.3 27.3.4

Index

511

Overview of References for Sampling and Related Topics 493 S. Hannappet Introduction 493 Literature 493 General Aspects 493 Statistical Methodology 495 Air 498 Water 499 Soils, Sediments, Sludges, Rocks and Mining 503 Biota 504 Waste, Sewage, Sludge 507 Norms by the International Organization of Standardization in Geneva 507 Air 507 Water 508 Soil and Mining 508 Biota 509

List of Contributors

Dr. Sarolta Biittner Department of Botany and Plant Physiology Agricultural University PBter Karoly u. 1 H-2103 G6dii116 Hungary

Prof. Dr. W. H. 0. Ernst Department of Ecology and Ecotoxicology Faculty of Biology Vrije Universiteit De Boelelaan 1087 NL- 1081 HV Amsterdam The Netherlands

Prof. Dr. Judith C. Chow Desert Research Institutc University of Nevada System Energy and Environmental Engineering Center 5625 Fox Avenue P.O. Box 60220 Reno, NV 89506 U.S.A.

Prof. Dr. Otto Frlnzle Geographisches Institut der Christian-Albrechts-UniversitHt zu Kiel Ludewig-Mayn-Str. 14 D-24118 Kiel F.R.G.

Dr. Ursula M. Cowgill Department of Environmental, Population and Organismic Biology University of Colorado at Boulder P.O. Box 1327 Carbonale, CO 81623 U.S.A. Dr. Gerd CroRmann Landwirtschaftliche Untersuchungs- und Forschungsanstalt Postfach 5480 Nevinghoff 40 D-48 147 Miinster F.R.G. Dr. Zsolt Csintalan Plant Physiology Section Department of Botany and Plant Physiology Agricultural University H-2 103 God6116 Hungary Dr. Elie M. Dick ISCO Environmental Division 53 I Westgate Boulevard Lincoln NE 68528-1586 U.S.A. Dr. Rumiana Djingova Faculty of Chemistry University of Sofia I , J. Bouchier Blvd. BG-1126 Sofia Bulgaria

Dr. Ern6 Fuhrer Forest Research Institute Papret 17 H-9400 Sopron Hungary Dr. Eric Fujita University of Nevada System Energv and Environmental Engineering Center 5625-6ox Avenue P.O. Box 60220 Reno, NV 89506 U.S.A. Dr. Helmut Guhr Institute lor Inland Water Rescarch GKSS Research Centre Am Biederitzer Busch 12 D-39114 Magdeburg F.R.G. Mrs. Susanne Hannappel Department of Analytical Chemistry lnstitute for Inland Water Research GKSS Research Centre Gouvernementsberg 1 D-39104 Magdeburg F.R.G. Dr. Peter Hoflmann Fachgebiet Chemische Analytik Fachbereich Materialwissenschaft Technische Hochschule Darmstadt Hilpertstr. 31 D-64295 Darmstadt F.R.G.

XXVIll

Llrt of' Contribiciois

Dr. Ranjith Jayasekera Department of Botany University of Kelaniya Kelaniya Sri Lanka

Dr. Ants Martin Institute of Plant Protcction Estonian Agricultural University Riia 12 EE-2400 Tartu Estonia

Dr. Liszl6 Kaszab Department of Botany and Plant Physiology Agricultural University Pater Karoly u. 1 H-2103 Godollo Hungary

Dr. Thcodor W. May Gesellschaft I'ur Organisation und Ent. scheidung ApfelstraUe 119 D-33613 Bielefeld F.R.G.

Prof. Dr. Helmut Klapper Institute for Inland Water Research GKSS Research Centre Am Biederitzer Busch 12 D-39114 Magdeburg F.R.G.

Mr. Zoltan Nagy Plant Physiology Section Department of Botany and Plant Physiology Agricultural University of Godollo H-2103 Godollo Hungary

Prof. I h . Margit Kovacs Department of Botany and Plant Physiology Agricultural University Piter Karoly u. 1 H-2103 Godollo Hungary

Dr. Jamcs E. Norris BCM Engineers Inc. P.O. Box 1784 Mobile, AL 36633-1784 U.S.A.

Prof, Dr. Giinther Kraft Hans-Thoma-Str. 6 0-6'1476 Kron berg/TS I F.R.G. Dr. lvelin Kulcff Faculty of Chemistry University of Sofia 1, J. Bouchicr Blvd. BG- I 126 Sofia Bulgaria Dr. Vambola Maavara Institute of Plant Protection Estonian Agricultural University Riia 12 EE-2400 Tartu Estonia

Prof. Dr. Bernd Markert Lehrstuhl fur Umweltverfahrenstechnik Internationales Hochschulinstitut Zittau Markt 23 0-02763 Zittau F.R.G.

Dr. Norbert Nothbaum lnstitut fur Didaktik der Mathematik Universitit Bielefcld Universititsstrak D-33615 Biclefeld F.R.G. Prof. Dr. Pekka Nuortcva Department of Environmental Protection university of Helsinki Caloniuksenkatu 6 C 64 SF-00100 Helsinki Finland Mr. Ahto Oja Department of Environmental Protection University of Helsinki PB 21 SF-00014 Helsinki Finland Mr. Andrcas Pactz Deutsches Institut fur Normung Burggrafenstr. 6 D-10772 Berlin F.R.G.

List of’ Contributors

XXIX

Prof. Dr. Istvan Pais Department of Chemistry and Biochemistry University of Horticulture and Food Science Villanyi ut 29 - 3 1 H-1502 Budapest Hungary

Mr. Kalman Szente Plant Physiology Section Department of Botany and Plant Physiology Agricultural University H-2103 G6doll6 Hungary

Dr. Karoly Penksza Department of Botany and Plant Physiology Agricultural University Pater Karoly u. 1 H-2103 Godollo Hungary

Dr. Pal Sz6ke Department of Botany and Plant Physiology Agricultural University Piter Karoly u. 1 H-2 103 G6doll6 Hungary

Dr. Robert W. Puls Robert S. Kerr Environmental Research Laboratory U S . Environmental Protection Agency P.O. Box 1198 Ada, OK 74820 U.S.A.

Mr. Zoltan Takacs Plant Physiology Section Department of Botany and Plant Physiology Agricultural University H-2103 GSdijllo Hungary

Dr. Michael H. Ramsey Environmental Geology Research Department of Geology Imperial College London SW7 2AZ U.K. Prof. Dr. Walter Rast Water Resources Division US Geological Survey Austin, TX 18753 U.S.A. Dr. Volker Schenk Erftverband Pfaffendorfer Weg 42 D-50126 Bergheim F.R.G.

Dr. Sindor Toth Department of Botany and Plant Physiology Agricultural University Piter Karoly u. 1 H-2 103 God6116 Hungary Dr. Zoltin Tuba Plant Physiology Section Department of Botany and Plant Physiology Agricultural University H-2 103 God6116 Hungary Dr. Gabor Turcsanyi Department of Botany and Plant Physiology Agricultural University Piter Karoly u. I H-2103 Gijd611B Hungary

Dr. Roland W. Scholz ETH, Swiss Federal Institute of Technology Chair for Environmental Science Natural and Social Science Interface RamistraBe 101 CH-8092 Zurich Switzerland

Prof. Dr. Dieter Uhlmann Sektion Wasserwesen Technische Universitat Dresden Mommsenstr. 13 D-0 I069 Dresden F.R.G.

Dr. I r k Siller Department of Botany Veterinary University Rottenbiller u. 50 H- 1077 Budapest Hungary

Dr. William G. Warren Science Branch/CODE Department of Fisheries and Oceans P.O. Box 5667 St. Johns, N F AIC 5x1 Canada

XXX

List qj’ Contributors

Prof. Dr. John G. Watson Desert Research Institute University of Nevada System Energy and Environmental Engineering Center 5625 Fox Avenue P.O. Box 60220 Keno. NV 89506 U.S.A. Dr. Erich Wcbcr Institute for Inland Water Research GKSS Research Centre Heydeckstr. 9 D-39104 Magdchurg F.R.G.

Prof. Dr. Barbara Zielinska University o f Nevada System Energy and Environmental Engineering Ccnter 5625 Fox Avenue P.O. Box 60220 Reno, Nevada 89506 U.S.A. Prof. Dr. Zueng-Sang Chen Department of Agriculturitl Chemistry National Taiwan University Taipei, Taiwan 106 Republic of China

Part I Historical Aspects

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH VerlagsgesellschaftmbH, 1994

1 History of Sampling Demonstrated on the Ore Mining Industry Empirical and Theoretical Approaches * Gunther Kraft

The beginnings of sampling go back to the ore mining and metallurgy of the Middle Ages, i.e. the production of non ferrous metals, particularly precious metals and these beginnings were - inevitably - purely empirical. Georgius Agricola writes in Book VII of his famous work “De Re Metallica” in the year 1556: “Sampling of ores which serves to determine the metal content only can be distinguished from melting of the ores by the smaller amount of used material. By melting of smaller amounts we learn whether melting of bigger amounts will bring profit or not. If the metallurgist did not use these methods of investigation carefully, as said before, the melting of ores to metals would sometimes only cause loss or at least not be of benefit. We sample minted alloys, which we call coins, in the following way: smaller silver coins which are taken from the top, from below and from the sides of a heap are well cleaned firstly, then melted down in a crucible and the melt either granulated or poured to flat plates from which flakes are produced. Bigger coins weighing 1 drachm, 114, 1 /2 or even 1 ounce are hammered flat. One takes 1/2 pound of the granules, the same amount of the flakes and in the same way another 1/2 pound. Each amount is filled separately in small paper bags. The material obtained in this way is cupellated with lead.” The following quotation proves that even then double determinations were made : “Then you take the silver grains out of the cupel and free them from slags. If one of the scale pan in which the grains were put, is not pressed down more than the other one, but the weight of both grains is equal, then our sample has no mistake. But if one scale pan hangs down more than the other, then the sample is faulty and has to be repeated.” “Alloys of copper and silver are sampled in the following way: The sampler cuts pieces out of an ingot of copper, small pieces from small ones, medium-sized pieces from medium-sized ones and big pieces from big ones; the small pieces of the size of half a hazelnut, the big pieces not bigger than a chestnut, the medium-sized ones of a size in between. He takes these samples from the middle of the bottom of each ingot, puts them in a new, clean, three-cornered crucible and adds a piece of paper with the weight of each ingot of copper and how many ingots there have been.” This way Agricola. Then it still took a very long time, approximately to the end of the last century, until these purely empirical beginnings slowly became a kind of theoretical basis

* This article has been previously published in: Kraft, G. (ed.), 1993: Sampling in the Non-Ferrous Metals Industry, Trans Tech Publication, Clausthal-Zellerfeld.

G. Kraft

4

of sampling. It is virtually impossible to try to understand the development in detail. However, some stages can be well determined. First to be mentioned are the works of Reed (about 1880 to 1885) who established a connection between the size of a sample and the particle size of the material to be sampled in such a way that the amount of material to be taken as sample has to be proportional to the cube of the diameter of the biggest particle in the substance. As an example: if a sample amount of 3 g was right for a biggest particle of 0.1 mm, so a sample of about 1 t would be necessary for a particle size of 10 mm di amctcr . 3 Among others Richards (1903) did not agree to this v-rule, He took the view that just the composition of the biggest particles represents more the average of the 2 total material than the smaller particles. Therefore he pleated for a v-rule and additionally distinguished between types of ores. As an example: a sample weight of 2.25 g (similar to Reed's example) would only be sufficient for very poor or very uniform ores when the biggest particle which passes the mesh sizes of the sieve used is 0.1 mm, whereas for very rich or very inhomogeneous ores as much as 4.5 kg would be required. The values resulting from the derivations of Richards were compiled as a nomogram (Fig. 1) by Taggart (1948),which almost until today has been considered as the guide for ore sampling. Seeking a more exact, mathematical control of sampling Brunton (1895) proceeded on the assumption that the error relating to the amount of sample needed mainly originates from the presence of particles of richest content and biggest volume. The essential parameters are the following three quantities:

.-

16'

16'

I

10

10' mm lo3

Fig. 1. Dependence of sample weight on particle s i x and character of ore (from Taggart 1948). (a) Ores with very small or very uniformly distributed metal contents; (b) ores with small or uniformly distributed metal contents; (c, d) ores with medium contents and normal metal distribution; ( e ) rich ores or ores with irregularly distributed metal contents; (0 very rich ores or ores with very irregularly distributed metal contents.

History of Sampling

5

weight of the desired sample; obviously the finer the material is crushed, the smaller the sample should be the ratio of the contents of the richest pieces to the average content the specific gravity of the richest pieces; the higher the specific gravity the bigger its influence on the result. 3

The result of his examination is again a I/-correlation of the crushing degree with the sample weight.

D

=

2.1544

f . s . n(k

-

C)

D = mesh size of the sieve, cm (corresponds to the crushing degree) W = sample weight, kg k = percentage of the component of interest in the richest mineral c = average percentage s = specific gravity of the richest mineral n = number of biggest pieces of the richest ore f = ratio of the experimentally found weight of the biggest particle of richest material passing the sieve used and the weight of the biggest cube of richest material which could pass the same sieve (i.e. a kind of factor for the shape) p = allowed error, YO Of these quantities, W and p can be defined at will; k , s and c are either known for a given ore or can determined easily, f can only be determined empirically. In extensive mesurements the author found values between 1 and 6 for different ores of a milling degree of 0.05-0.1 mm. The reason for this big discrepancy in firstly the variability of the ores themselves and secondly the fact that many big pieces can be longish and thus pass through a sieve of a nominally smaller mesh size. Also n, which more or less is a measure for the quality of mixing, can only be determined experimentally; the values found are around 3. If the quantities W , p, s, f and n for a given material are summarized to the new constant R, the equation is simplified:

Gy (1955) obtains a very similar result from his derivation, which contains as a new factor the variance of the sampling error. His equation is: P.O2

~-

d3

-

const (c)

sample weight variance of the relative error which is made (or can be made) during sampling related to the contents of the component to be evaluated d = size of the biggest pieces ( = mesh size of the sieve which holds back 5 - 10% of the material) P

=

0’ =

6

G Kruft

c = a constant which characterizes the nature of the ore to be sampled. It includes: - a value a/a, the portion of usable ore, which is contained in thc lot to be

sampled the specific gravity 6, of the usable ore (if u / a < 70%) or the specific gravity 6, of gangue (if U/LY > 70%) - a parameter 1 which characterizes the crushing degree of the material but cannot be estimated easily a = content of sought element in the lot TX = content of the same element in the richest ore. -

Gy found the following correlations: I approaches one the purer the material is and/or the finer it is millcd. On the other hand this error becomes smaller the coarser the particles are compared to crushing degree. For example, if the biggest particles are 100 time the size of the medium crushing degree, I becomes approximately 0.05 only. But if all the material is milled to the crushing degree, I becomes 0.8. Of course, there was no shortage of efforts to solve this problem of sampling purely mathematically. Only the names Mika (1928) and Baule/Benedetti-Pilcher (1928) shall be mentioned here. In each case probability calculus was the basis; but finally only for two-component-systems could solutions be developed. Therefore these approaches shall not be further pursued here, although work still continues on their sophistication (i.e. Wilson, 1964). The most modern approach and perhaps even the solution to the sampling problem is the mathematical-statistical formulation (e.g. Kraft, 1978). It says vcry simply and pragmatically that

In words: take so many increments N , with any constant weight, of the material to be sampled, e.g. shovels of a weight of 1 kg each, so that this number is equal to the square of the quotient of the error of the complete sampling operation, including all the following dividing operations and the analytical determination s (s = standard deviation, t = Student factor, e.g. 2 for a statistical security of 95%) and the allowed uncertainty F U of the obtained result. The total weight G of the sample taken then is: G=N.g = weight of the increment). So we see: no data are needed on form factor, milling degrees, biggest particles, specific gravity or even values that can only determined empirically. Instead the standard deviation s, which indirectly includes all these factors, gains central importance. It must be repeatedly determined for each material to be sampled. This can be done, for example, by multiple samplings of at least 10, preferably 20 samples, which have to be assessed and analyzed separately. At a glance this seems to be excessive. But it is always the cheaper and above all the safer method whcn similar materials, for example concentrates from the same mine, have to be sampled over a longer period of time.

(g

References Baule, B., Benedetti-Pilcher, A . (1928). Z . Anal. Cketn. 74, 442. Brunton (1895), Trans. Am. Inst. Min. Eng. 25, 826. Gy. P. (1959, Firnc~tull8,B 199. Kraft, G. (1978). Erzmrtull31, 53. Kraft, G. (ed.) (199.3). Sumpling in the Non-t.i.rrons Metuls Industry. Trans Tech Publication, Clausthal-Zellerfeld. Mika, J. (1928). Z. Anal. Clrrm. 73, 257. Reed ( 1 881/82), School of' M i n ~ sQuurtedv 3, 253. Reed (1 884/85), Sclzool oj' Mines Qurrrierly 6 , 35 1. Richards, R. H. (1903), Ore Drcming ~ i n dConcentration, Vol. 2. London, New York, p. 843. Taggart, A. ( I 948), H(intlhook of Miricrrtl Drc,ssing, Ores ond Industrial Mineral,y. John Wiley, New York, p. 161.

Part I1 General Aspects

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

2 Genera Aspects of Environmental Sampling *

2.1 Introduction The first and in many instances most important step in every determination in analytical chemistry, namely the sampling, unfortunately rather seldom receives due attention. This is, as already formulated by many authors, because mistakes made during sampling can no longer be corrected. With today’s possibilities of trace and ultra-trace analysis, sampling has gained more significance. Through the development and application of instrumental measurement methods analytical chemistry has developed capabilities that had to be considered Utopia several years ago. Mass-spectrometry with a11 inductively coupled plasma for its ion source (ICP-MS) serves as an example here: within about ten years this method of measurement has become a workhorse of the trace analyst, since with its help detection limits in the range of 10 to SOpg/g can be achieved, in ideal instances, for around 60 elements (pure standard solutions). Nevertheless, the chemical analysis of a material not only consists of the measurement of the material in one of the available analysis apparatus, but also of a multitude of single steps, which must be coordinated: definition of the analytical problem, sampling, storage of samples, preparation of samples, measurement, evaluation, samplc comparisons (standard samples, reference materials) and assessment of the results. While in many instances the measurement process is founded on physical principles (for example, absorption and emission of electromagnetic radiation, separation of masses in electrical and/or magnetic fields), a well-grounded knowledge of chemical and often of biological reactions is necessary for the other steps [I]. In the framcwork of this introduction, only the processes which play a role in sampling will be discussed. Here the term “sampling” is not to be equated with the term “sampling” as used in Anglo-Saxon scientific literature, since there the term includes all processes which must be carried out before the measurement, hence also storage and preparation of samples (solution, disintegration, separation, concentration, etc.). In the course of this discussion on sampling, the procedures of sample storage and conservation are treated, since correct and skillful sampling includes these matters. Nevertheless, it must be noted that each sampling procedure is strongly connected with storage, preparation, analytical method and assessment of the results. The connection between sampling and sample storage and conservation is illustrated with an example from water analysis: in ground water samples available under “anaerobic” conditions, iron is found predominantly in the oxidation state + 2. With incorrect sample taking, storage and conservation, oxygen infiltrates into the

*

This article has been prcviously published in Nudir. Chen7. Tr.r/i.Lob. ( I 992) 40, M 2 (in German).

12

I]. Hojynxinn

sample, the redox potential of the solution changes, and the iron is carried over into the oxidation state 3. Since a pH value in the range of 5 to 8 exists in natural water, iron precipitates as Fe(II1)hydroxide (oxidehydrate). As a result, not only too little iron is found in dissolved state, but also the concentration of other elements decreases in the solution through coprecipitation, or through adsorption to the precipitate. In such cases sampling can only be considered correct if the problems of sample storage and conservation are solved. As one can recognize from this, in sample taking, storage and conservation, grave mistakes are possible which cannot be compensated by any measurement, no matter how correct and reproducible [2, 3,4]. In this chapter, further aspects will be discussed which must be taken into consideration in sampling.

+

2.2 Terms and Definitions In the literature there exists a multitude of special expressions concerning the subject “sample”, which frequently are not explained, or receive specific interpretations by the various authors. In accordance with selected publications [5, 6, 71, a compilation of several important terms follows, with short explanations provided: -

-

-

-

-

Single or spot sample: Material taken from the bulk quantity in a single sampling procedure. Random sample: One or more samples taken successively in short time intervals. Short-time mixed sample: Composed of a mixture of at least live random samples gathered in intervals of not less than two minutes. Raw, mixed, or composite sample: Different designations of the sum of all single/spot samples processed collectively. Intermediate sample: Created through reduction of the collective sample, and ready for processing. Sample set: Parts of the sample that accumulate during sample preparation, and need to be kept separate (for example, metallic - oxidic, fine - coarse, gaseous - liquid - solid). Final sample: The sample amount obtained from the sum of all single/spot samples (not necessarily identical to raw, mixed, or collective sample, since every single sample may have been separately processed), from which the analysis sample is taken. Analysis sample: The end product of the sampling and sample preparation, which has been prepared for the intended investigations.

The technical process of sampling mostly is evident from the notation itself (drill sample, filter sample, impactor sample, saw sample, strike sample, scoop sample, prick sample, etc.) and does not need to be elaborated on here. On the other hand, the sampling types that arc above all relevant to quality control must be presented with appropriate commentary: -

Time-dependent sampling: Single samples of identical volume are taken in previously established time intervals (in case of continuous material flux this corresponds to mass- or volume-proportional sampling).

General Aspects of Sumpling

13

Volume-dependent sampling : Single samples of identical volume taken in variable time intervals which are corresponding to the flux (flow rate). - Flow-dependent sampling: Single samples taken in identical time intervals with volumes corresponding to variable flow rate. - Continuous sampling at fixed flow rates: Continuous sampling of a constant volume. - Continuous sampling a t variable flow rates: Sampling of variable volumes which correspond to the flow rate. - Surface samples: Sum of all spot samples taken at the surface of, for example, stagnant waters. - Depth-integrated samples: Sum of all spot samples taken a t the same place at different depths of, for example, stagnant waters. - “Isokinetic” sampling: Used in heterogeneous systems (chimneys, tubes, rivers) where the sampling is done through a suction tube and at a rate that the velocities in the tube and in the main stream are equal. -

2.3 Aspects of Sampling Upon examination of pertinent literature and detailed discussions with colleagues and coworkers, a considerable number of viewpoints arose in response to the question: “Which influences are to be considered in sampling?” It certainly depends upon the analyst’s line of work as to which influence he considers most important. In any event, it can be stated objectively that consideration of sampling conditions becomes more important as more demands are put on the quality of the analytical results. The lower the concentration of the element that is to be determined in the sample, the higher the desired precision of the result, and, the better the desired temporal or topochemical resolution of the analysis, the more thought has to be given to the sampling procedure. When purchasing a sampling system, it is wise to test it for a few weeks before actually buying and paying for it.

2.3.1 Location, Place, and Position of Sample Removal This parameter has two completely different aspects. When a massive material must be tested to determine its quality, it is necessary to carry out the analysis according to statistical viewpoints. The metal and ore manufacturing industry has investigated this area, and it is dealt with at length in the literature. G. Kraft has proposed sample taking schemes for a number of metals and materials, and for various problem statements [ 5 ] : - Arrangement of bore holes on pig iron slabs, or sections of slabs; - saw pattern for lead blocks (also for large numbers of blocks); - arrangement of bore holes on lead blocks and soft lead blocks;

P Hoffmcinn

14

taking sample cuttings from two diagonal surfaces of a casting format; drilling model for copper plates; - sector sample from a steel club; - drilling diagram and arrangemcnt of the saw cuts for a zinc alloy bar. -

-

Corresponding and expanded proposals are found in F. Pottkamp [8]. A completely different aspect results from the search for the most appropriate sampling location for problcms i n the area of cnvironmcntal protection, or when invcstigating causes of contamination. In such cases one must not proceed statistically, but sampling must be carried out systematically near the source (for example, in the direction of the wind, or downstream). In the analysis of different water samples the sampling location also requires manifold description. This can be seen in examples found in forms used as a sampling protocol, which are listed in Section A1 1 -A20 of the German Standardized Procedure for Water, Waste Water, and Mud Analysis [9]. ~

~

-

Waste water: samplc taking location; stagnant water: sample taking location, water level; drinking water: sample taking location, armature, hydrant; flowing water: bank (right, left), river center, sample taking depth, coordinates.

2.3.2 Size, Quantity, and Volume of the Sample The size of the single sample is dependent upon the expected concentration of the element to bc detcrmincd in the material, the analysis procedure that is to be used, the precision of the result that is to be evaluated, and the state of distribution of the material. Thc grcatcr thc volumc ofa liquid sample from which the element to be determined can be concentrated, the more prccisc the determination can be. Experience shows, however, that liter samples represent an optimal measure, while 5-liter samples are very difficult to manipulate, e.g., in waste water laboratories. In addition, with solid matter, the size of the sample is determined by the grain size of thc matcrial and the homogcncity of distribution of the element that is to be analyzed. A nomogram by Taggart, in which the minimum weight of an average sample can be determined as a function of the grain size of the largest grain, and of the content and distribution of the element to be determined, was developed for ores, but has also been applied to earth samples, rocks, salts, and grain [5, lo].

2.3.3 Number of Samples to be Taken The number of samples that need to be taken in most cases depends upon the problem, or the way the problem is posed. The necessary number N of single samples is determined according to the following equation: N

=

(ts/L)’

General Aspects of Sampling t =

15

Student-Factor ( = 1.98 for a statistical certainty of about 95%);

s = estimate of the standard deviation of the arithmetic mean value of all single

values (preliminary examination of at least 20 single samples using absolutely identical procedures: sample taking apparatus, sample quantities, component to be determined, measurement procedure); L = tolerable uncertainty (stated in YO)of the result. This equation, as well as its application, has been explained in detail by F. Holstein, using practical examples [I 11.

2.3.4 Time, Duration, and Frequency of Sampling These aspects have a significant influence when the material that is to be analyzed is exposed to a chronologically changeable influence, i.e., if a streaming system is under consideration, or if the way in which reactions proceed needs to be observed. The choice of these parameters is dependent upon - from an analytical standpoint - the speed of the observed changes only. Investigations by the Department of Water and Waste Management in Hagen, Germany (121 have shown that an official supervision of waste water is necessary “around the clock”, in order to register compliance with legal limits. In reality, however, the times of sample taking are determined by the working hours of the personnel in many instances. The duration of sampling in water and waste water supervision is officially regulated according to decisions regarding water laws. One distinguishes between the random sample, the short-time mixed sample, the two-hour mixed sample, and the 24-hour mixed sample, which corresponding to the problem at hand are established and defined in the form of norms by the water authorities on the basis of the relevant waste water management ordinances according to { 7a WHG [13]. The time requirement for taking a sample for gas analysis from liquid metals plays an important role in the metal manufacturing industry, whereby the diffusion coefficient for the gas in question (for example, hydrogen) at different temperatures is decisive. Diverse, and in part very costly, sampling apparatus are applied here: probes, suction guns, vacuum pipettes, suction pipettes, vacuum suction molds, and others. The frequency of sampling concerning waste water is officially regulated and depends on the following criteria: danger of harmful substances in waste water, special risk factors, effect on the pre-flooding system, utilization demand on the pre-flooding system. Four frequency levels of sampling result from this: twice per week, twice per month, four times per year, and six times per year. In many instances a continuous measurement of analytical quantities is desired. In one of the further steps, the necessity of automatic sampling and measuring may arise. Only a few analytical procedures, due to their character, are suitable for that purpose. X-ray fluorescence analysis has often been used for such purposes, especially its energy dispersive version, with radionuclides for excitation sources [ 15, 161. The efforts to develop Flow-Injection-Analysis (FIA) procedures and hyphenated methods such as combinations of chromatographic separation procedures with sensitive measuring systems (GC/MS, IC/ICP-AES, HPLC/AAS, IC/TXRF) have

16

P. Hofjtnunn

to be interpreted with regard to the desire to create continuously registering methods without any concrete sampling. Reports about automatic sampling are scarcely found in literature [17].

2.3.5 Homogeneity of the Sample In heterogeneous multi-phase systems such as aerosols, fogs, suspensions, emulsions, sediments, sludges, etc., the difficulties in sampling increase exponentially, and one has to accept the fact that for some of these mixtures no useful sampling devices are available. The homogeneity of the sample always plays a role when the material under investigation is not mixed thoroughly, when solid substances are suspended in the solutions, or when solid and liquid substances are distributed in gases. A thorough treatment of this complex of problems can be found in a paper by Ortner [18], where the determination of heterogeneously and homogeneously distributed trace impurities in refractory metals is discussed. Using several examples, it is shown in this chapter, that homogeneity is not ensured in all cases where thorough mixture of all components can be assumed. Inorganic as well as organic components exist in varied chemical forms and distributions, depending on their origin and history. For example, iron in a bivalent or trivalent state can exist in atmospheric water-based samples (clouds, fog, rain water). Thereby the distribution of the oxidation states depends on the pH-value and on the redox potential. These parameters are also responsible for the existence of Fe(J1 I) in dissolved, colloidal, or fine to coarsely dispersed form. So it is not irrelevant sampling is done without filter, with filter, or with ultrafilter [19]. In this example, one recognizes that the sampling type is directly related to the analytical problem statement. The sampling procedure is fundamentally different if the total iron concentration, the dissolvcd portion, or the portion of a definite valence is to be determined. This is similarly valid for mineral oil samples, in which trace elements can be present as solutions of organo-metallic compounds or as suspended and finely dispersed particles [20]. Sampling from natural gas can lead to an incorrect result if it is not observed that the removed sample and the area that is to be studied are geologically separate and therefore not in balance. In solid materials unexpected inhomogeneities can appear. Also glasses and alloys produced from smelting show that a homogeneous distribution is not given for all components. The production technique of antique glasses has not progressed so far that homogeneity can be presumed. In the study of such pieces, taking small samples for analysis can lead to results that are not representative of the whole object. The following observation made in modern materials rests on a different foundation. From a production charge of 200 t of steel, a random sample of 100 g is taken, 100 mg of which is examined by X-ray fluorescence analysis. The outcome of the measurement is representative for the complete amount, although only about 10-9th part was analyzed [21]. On the other hand, small eroded particles (pg-scale)

General Aspects of’ Sumpling

17

from high-grade steel tools (10 g-scale) do not show the composition of the original material, since here the single sample comes from an inhomogeneous structural region [22]. As is shown by this example no generally valid level of reduction for a sample can be given, and the decision must be made in each case. One of the most complicated sampling tasks to perform is the investigation of scrap material, waste dumps, and mining rubble, since in these instances the inhomogeneity can be especially striking. Among experts on the subject, the scrap material waste dump on which a container with 1 kg of mercury (80 mL) is found, is discussed as the archetypical example. In this area the widest spectrum of sampling methods is applied, from manual probes to heavy excavating equipment. Especially here, the expert and experienced “human” sampler is especially valued. Another example about sampling from inhomogeneous solid body accumulations shows a different human influence. With the help of a mobile X-ray fluorescence analysis apparatus, the composition of silver mining rubble in the Black Forest, Germany, had to be tested for its content of precious metals. Samples were taken from all over the dump area, then crushed, ground, mixed, separated and measured [23]. All results agreed within a margin of error with those that were obtained by other groups years earlier. An exception was the element uranium, for which distinctly lower concentrations were found in the newer measurement. The reason was a collection of radioactive samples that in the meantime had been taken by hobby mineralogists, who had scoured the dump with suitable detectors. The suspicion that the number of chunks containing uranium was depleted only on the surface, was confirmed by a new set of samples, where deeper layers and rocks that could not have been moved by human force were also included in the considerations. The values measured earlier were also confirmed for uranium by this second set of measurements. This example shows that for correct sampling, the history of the materials in question must also be known [l].

2.3.6 Contamination of the Sample This section, as well as the following concerning the topic of “losses in the sample” can, in spite of their enormous significance, be kept rather brief, since a survey article exists that deals with all these problems as extensively as necessary, with comprehensive tabulated material and a bibliography of no less than 577 quotations [24]. Since this article unfortunately is not yet generally accessible, and since it does not appear to make sense to limit oneself to singular quotations, the contents of the following paragraphs are tightly patterned after that text [24]. Contaminations can occur, for example, through the following influences: -

-

air: industrial area, street vicinity, forest area; premises: walls, ceilings, floors; furniture: tiles, wood, synthetics, history of the laboratory; human: clothing, cosmetics, medicines, smoke of cigarettes/cigars; reagents: acids, water, complex forming agents;

18

-

P.Hofrmunn

receptacle materials: containers, pipettes, glass, quartz, carbon fiber, platinum, and synthetics such as Polyethylene (PE), Polytetrafluorethylene (PTFE), and Polypropylene (PP).

The following set of problems puts the significance of contamination into the right light: reagent-grade water (repeatedly distilled, deionized, filtered, purified by sub-boiling method) is used in inorganic, as well as in organic, trace analysis. For inorganic trace analysis, containers made of PE or PTFE are preferred, while for organic analysis glass or quartz bottles are used. Nevertheless, it has been observed that organic substances, dissolved out of synthetic materials, lead to systematic errors in inorganic trace analysis through reactions (formation of complexes, oxidation) with the trace elements that need to be determined. contamination problems turn out to be the analyst's most serious problems at concentrations below pg/g. It is strongly advised to wear gloves while working with such samples, and if possible, to work in a clean environment (clean-room conditions, clean-benches, closed evaporation or steam vents). It is also important to mention that such work can only be performed by trained and qualified personnel. On the cleaning of receptacles, various suggestions are made corresponding to the various problems [18,24,251.

2.3.7 Losses in the Sample It is also necessary here to refer to the abovementioned survey article [24]. Losses can occur because of adsorption of elements or their compounds on rcceptable walls or on particular components (precipitation, suspended materials), as well as through evaporation. With sorption, the most significant effects are observed at the sides of glasses and in hydroxide precipitates, since there exchange reactions are decisive. The losses of trace elements are most important in solutions of low dielectric constants (e.g., organic solvents). Synthetics and quartz show noticeably lower adsorbing power and for this reason are most suitable for storage of aqueous solutions. It must be further mentioned that surfaces of receptacle materials change through time and contact with various chemicals, whereby the adsorbing power usually rises drastically. This observation can be made especially clearly when bases come into contact with glass walls, or if hydrofluoric acid comes into contact with quartz. Evaporation is observed with mercury in elementary form if conditions for reduction exist in the solution, while other elements evaporate as oxides (e.g., As, Sb, Re), halides (e.g., elements of groups IV, V, and VI of the periodic system), or hydrides ( e g , As, Sb, Se, Te), or they may diffuse through the sides of receptacles (synthetics). In the analysis of organic solvents (e.g., hydrocarbons, halogenized hydrocarbons), the effect of evaporation must be especially taken into account. Standard solutions stored in plastic bottles alter their concentrations by about I YOper year, since the solvent, water, escapes through the receptacle walls.

General Aspects of Sunipling

19

2.3.8 Sample Storage and Conservation Samples that cannot be processed immediately after sampling must be long-term protected against contamination, losses or other changes. So it makes sense to design the sampling in such a way that the sample can be stored and conserved without too much trouble. The sampling receptacles must be selected in order to reduce contamination of losses to a minimum also in long-term storage. When storing samples, biological activities, hydrolysis, and evaporation of parts of the sample must be suppressed, too. This can be accomplished with the following chemical and physical operations: acidifyingto pH 1.5; minimization of adsorption; prevention of metabolic processes of microorganisms; avoidance of hydrolysis and precipitation. Cooling and freezing: Reduction of bacterial activity, even this simple procedure has to be carried out very carefully. A paper was published describing an accelerated oxidation of nitrite to nitrate and of sulfide to sulfate by freezing an aqueous sample to about -20 "C [27]. - Addition of complex forming agents: Production of anionic complexes reduces the danger of losses through adsorption or evaporation. - Filtration: Prevention of reactions of particles with dissolved components. - UV-radiation: Destruction of biological and organic components, in order to avoid formation of larger complexes, frequently in combination with H 2 0 2 -additions. -

It must not remain unmentioned that samples conserved in a such manner can also be stored for a limited amount of time only before changes must be suspected. In general, a storage time of up to 28 days is recommended, but in some cases not more than six hours [6]. Further suggestions about sample storage and conservation can be taken from previously mentioned review articles [24]. Samples in which the chemical form ofelements or their state of distribution are to be investigated [26], can just not be stored and conserved at all, since in the above mentioned processes conditions change so that the original state can no longer be recognized.

2.4 Guidelines and Norms Analytical results have become the basis of planning, decisions, and legal proceedings [28]. Quality control of raw materials and products, the development and production of chemicals, pharmaceuticals, foodstuffs, pesticides, etc. as well as the monitoring of our environment, are founded on highly developed, integrated analytical procedures. These procedures must be painstakingly adhered to, for reasons of reproducibility and to guard against legal problems. This is why industry and authorities were forced to pass binding regulations, guidelines, norms and laws, according to which objective decisions can be achieved. These regulations include the complete analytical procedure and thereby also detailed specifications with regard to sampling. The following thoughts are the basis

20

P. Hojfmann

of all publications: no step is permitted to be left to chance, and the development of an analytical outcome must be completely retraceable. For this reason all instructions and information exist in writing. Subsequently, the structure and the contents of work instructions, standard operation procedures (SOPS),standard test procedures, test regulations, work notes and protocols are demonstrated in the form of some selected examples, and these may serve as models. It would be beyond the scope of this survey to attempt mentioning all publications of this type.

2.4.1 Quality Assurance A work guide on sampling for quality control must first of all define the objective and the scope. In an introduction it is established which analyses and tests on which materials are carried out. The description of the manner in which to proceed must be composed of thc following segments: -

-

-

-

Stipulation of the localities where the sampling has to take place. Containers whose contents have to be tested must be marked clearly, set up separately, visually examined and cleaned. Furthermore, waybills must be at hand. Sampling apparatus must be stipulated: type, area of use, cleaning. Preparation for sampling: description of clothing and personal cleaning and security precautions. Handling for sampling apparatus, preparation of labels/test report forms. Sampling: comparison of all statements on waybills and containers. In case of inconsistencies, an office to report to must be established. Contamination-free sampling (sometimes very specific instructions with dangerous, toxic or otherwise problematic materials). Prescribed number of samples, location of sampling (surface, under the surface, at the bottom of the receptacle). Homogenizing of material (melting, shaking, stirring). Observation of the article (smell, color, grain size, foreign substance). Closing and marking the container. Filling out the sampling record. Release of article, or rejection (documented by stickers). Enclosure: sample stickers, record forms, instructions.

The analysis of ores, cement, concrete, road construction materials, ceramic materials, bituminous coal, coke, liquid and gaseous fuels, lubricants, emissions, imissions, drinking water and waste water, raw iron, steel, metals alloys, painting materials, and filter paper is of such a large and general significance that numerous German (DIN) and international (ISO) norms were formulated for the analysis and therefore also for the sampling of these materials. A summary of these joint publications and guidelines is found in the “Handbuch fur das Eisenhuttenlaboratorium” [29]. The I S 0 norms for sampling of iron ores are used here as a model and cxamplc [30, 311.

General Aspects of Sumpling

21

2.4.2 Environmental Protection In order to obtain a justifiable or at all meaningful analytical result with regard to strains on the environment, sampling norms must likewise be observed which have already been cited in preceding sections on water, waste water and mud [9, 131. Therein the following matrices are included: waste water (A ll), stagnant water (A 12),ground water (A 13), raw water, drinking water (A 14), running water (A 15), ocean (A 16), falling wet precipitation in the liquid state (A 17), water from mineral and medicinal springs (A 18),swimming and bathing water (A 19), tidal water (A 20), and cooling water for industrial use (A 22). In spite of this, these norms must also be critically handled and constantly improved [32]. The norm in the subject of sample conservation is still being prepared [33]. Corresponding international norms that gain more and more significance with the integration of Europe are to be observed as well [34, 35, 361. For sampling in the atmosphere, the VDI guidelines above all are valid [37,38] in which the analysis - and therefore also the sampling - of components like sulfur dioxide, carbon monoxide, nitrogen dioxide, ozone, dust, or hydrocarbon is determined. Monographs treating this subject range can only be cited selectively [39,40,41,42]. The various collection techniques, like filter collection, multi-stage gas sampling tubes, adsorption on surface-active substances (Amberlit, Chromosorb, Silicagel, Tenax, etc.), are especially considered in [43]. In this context it has to be mentioned that at present bioindicators are also used officially in the investigation of air pollutants [44,45,46], whereby sampling takes on a new aspect. As already mentioned, sampling in solid materials of the environment (soil, waste, garbage, compost, sewage, cinders, ashes, flora, fauna), which are therefore especially afflicted by inhomogeneity, represents a particularly serious problem. Especially here the experienced human sampler, who can skillfully react to a given situation, is irreplaceable. He is aided by guidelines which have yet been cast in the form of norms or laws [46,47,48,49]. In summary, it must be stated that complaints are brought up frequently about the lack of proposals on sampling in various areas. Intensive studies of this matter, however, have shown, that many references can be found in the original literature and in monographs. Certainly not all problems have been solved, and many regulations or proposals require improvement. But one must also point out that sampling, in practice, must choose a middle course between scientifically unobjectionable procedure and economically justifiable expense. The importance of this object is shown by the fact that a “Sampling Club” is planned in the United Kingdom, which will assist its members with problems, give information, and encourage the interaction with other colleagues [50]. Acknowiedgement. In this sense I would like to thank all those who made it possible for me to deal with the complex field of sampling from a very diverse set of viewpoints - 1 would especially like to thank J. Bartl, J. Dahmen, G. Grubert, H. Gudernatsch, K. Hanewald, T. Hofmann, W. Kreisel, K. H. Lieser, B. Markert, K. Ohls, H. M. Ortner, G. Rasenberger, R. Voigt, H. Wunsch. The translation was carried out by M. and G. Kluckner.

22

P. Hojfmann

2.5 References [I] Baiulescu, G. E., Dumitrcscu, P., Zugravescu, P. G. (1991) Sampling. New York: Ellis Horwood, p. 12ff. [2] Kratochvil, B., Wallace, D., Taylor, J. K. (1984) Sampling for Chemical Analysis, Anal. Chem. 56. 113R. [3] Kratochvil, B., Taylor, J. K. (I98 1) Sampling for Chemical Analysis, A n d . Chem. 53,924A. [4]Klockow, D. (1987) Zum gegenwdrtigen Stand dcr Probcnahmc von Spurcnstoffen in der freien Atmosphiirc, Fresenius 2. Anal. Chem. 326, 5. [5] Kraft, G. (1980) Probenahme an festen Stoffen. Analytiker-Ta.schenhueh, Bd. I . Berlin, Heidelberg, New York: Springer-Verlag. [6] Gudernatsch. H. (1983) Probenahme und Probcaufarbcilung von W Taselzenhucti, Bd. 3. Berlin, Heidelberg, New York : Springer-Vcrlag. [7] Gassen, M.. Woffen, B. (1978) Zur Hiiufigkeit der Probenahme und der Bcurtcilung dcr Leistungsfiihigkeit von Klaranlagen, pf-Wasser/Ahwusser 119, 455. [XI Pottkamp, F. (1980) Probenahme von Blei und Zink, Prohenuhnze - Thwrie und Praxis 36, 155K der Schriftenreihe der Gesellschaft Deutscher Metallhutten- und Bergleute. Weinheim, Deerfield Beach, Basel: Verlag Chemie. [9] Deutschos Einheit.si)c,rfahrc.n zur Wasser-, Ahwassrr- und Sr.hlamrnunter.suchung, Bd. 1. Weinheim: Verlag Chemie. [ 101 Kraft, G. (1980) Theoretische Grundlagen der Probenahme, Prohenahme - Theorie wid Pruxis 36, I ff der Schriftenreihe der Gesellschaft Deutscher Metallhiitten- und Bergleute. Weinheim, Deerfield Beach, Basel: Verlag Chemie. [ I I] Holstein, F. (1980) Probenahme von NE-Konzentraten (Beispiel: Kupferkonzentrate), Probenalzme - Theorie und Praxis 36, 93ff der Schriftenreihe der Gesellschaft Deutscher Metallhutten- und Berglcute. Wcinhcim, Deerfield Bcach, Basel: Vcrlag Chcmic. [ I21 Selent, K. (1988) Die amtliche Abwasserprobcnahmc, Prohenahnesc/iulung Wasser und Abwasser. Landesamt fur Wasser und Abfall Nordrhein-Westfalei, Diisseldorf, 1988. [ 131 DIN 38402, Teil 1 I - Probenahme von Abwasser. [ 141 Feichtinger, H.-K. (1980) Probenahme aus fliissigen Metallen zum Zwecke der Gasanalyse. Probenahme - Tlzeorie und Praxis 36. 237 der Schriftenreihe der Gesellschaft Deutschcr Metallhiitten- und Bcrglcute. Weinhcim, Dccrfield Bcach, Basel: Verlag Chemie. [ I51 Autorenkollektiv (Federfuhrung: H. Ehrhardt) ( 1988) Ront~en/luorrs~enzona/yse - An wendung in Betriehslaboratorien, 2. Auflage. Leipzig: VEB Deutscher Verlag fur Grundstoffindustrie. [16] Pilz, N., Hoffmann, P., Lieser, K. H. (1988) In-line Determination of Heavy Elements by Gamma Ray-Induced Energy-Dispersive K-Line XRF, J. Rndioand. Nucl. Chem. 130, I4 1. [I71 Stock, H . D., Grubcrt, G., tc Hecscn, D., Sclcnt, K . (1990) Dcr selbstentleerende Ruckstellprobennehmer - ncuc Chanccn der Abwasscriiberwachung, Gewiisserschurz - Wusser - Abwusser 118, 285. [IK] Ortner. H. (1980) Uber die Problematik dcr Probcnahmc bci Sondcrmclallen, Prohenahme Tlzeorie und Pruxis 36. 18 1 ff der Schriftcnreihe der Gesellschaft Deutscher Metallhiitten- und Berglcutc. Wcinhcim, Decrficld Bcach, Basel: Vcrlag Chcmic. [IY] Hofmann, H., Hoffmann, P. Licser, K. H. (1991) Transition Metals in Atmospheric Aqueous Samples, Analytical Dctcrmination and Speciation, Fresenius J . Anal. Chem. 340, 59 I . [20] Karchmer, K. H., Gum, E. L. (1952) A n d Chem. 24,1733, in: Koch, 0.G., Koch-Dcdic, G. A., Handbircli deer Spurenunulyse, 2. Aufl. Bcrlin, Hcidclbcrg, Ncw York: Springer-Vcrlag, 1974. [21] Ohls, K., Hoesch Stahl AG, Dortmund, personal communication. [22] Hoffmann, P., Paller, G., Thybusch, B., Stingl, U . (1991) Dctcrmination of Stainlcss Steel Constituents in Plastics, Fresenius J. Anal. Chem 339, 230. [23] Hoffmann, P., Lieser, K. H., Hofmann, T., Sommer, R. (1983) A Mobile Installation for Energy-Dispersive Multielement X-Ray Fluorescence Analysis for Application in the Field, X-Kuy Sprctrom. 12 (4). 175. [24] Licht, K., Jank, B., Birkhahn, J, Scharf, H., Spieles, M., Koehler, P., Winnefeld, C., Kulick, I.. Konranzinntion und Verluste - Ursachen ,fir systematische Fehler in der anorganischrri Spurenrmulyse. Carl Zciss, Jcna.

General Aspec f s

of

Sampling

23

12.51 Mart, L. (1979) Prevention of Contamination and Other Accuracy Risks in Voltammetric Trace Metal Analysis of Natural Waters, Fresenius Z . Anal. Chem. 296, 3.50. 1261 Takenaka, N., Maeda, Y. (1992) Acceleration of the rate of nitrite oxidation by freezing in aqueous solution, Nuture 358, 736. [27] Lieser, K. H . (1992) Spezialion: Eine Herausforderung fur die Analytische Chemie, GIT Fachzeitschrft , f i r das Lahoratorium 36, 293. 1281 Vogel, H. (1990) Cute Analytische Praxis, in Anulytikw Taschenhuch, Bd. 9. Berlin, Heidelberg, New York, London, Paris, Tokyo, Hong Kong: Springer-Verlag, p. 3ff. [29] Hand6uch , f i r das Ei.senhurr~nlahoruto~ium (1987) Chemikerausschufi des Vereins Deutscher Eisenhuttcnleute (ed.), Bd. 3A - Probenahme. Dusseldorf: Verlag Stahleisen mbH. [30] I S 0 3081 (1987) Iron ores - Increment sampling - Manuul method. [3 I] I S 0 3082 (1987) Iron ores - Increment.samplingand,sampleprepuration- Mechunicalmethod. [32] Gortz, W.. Grubert, G. (1986) Probleme bei der Entnahme von Mischproben fur die Untersuchung von Wasser und Abwasser, Gewusserschutz - Wasser- Ahwusser 86, 2 1. [33] DIN 38409, Teil 21, Konservieren von Proben. [34] I S 0 5667/1 (1980) Water quality - Sampling - Part I ; Guidance on the Design of Sampling Programmes. 1351 I S 0 5667/2 (19x2) Water qualit), - Sampling - Purt 2: Guidance on Sampling Techniques. [36] I S 0 5667/3 (1985) Wutrr quality - Sampling - Purt 3: Guidance on the Preservation and Handling of Samples. [37] VDI-Handhuch - Reinhultung dcr L u f f , Beuth-Vertrieb GmbH, Berlin - Koln. 1381 VDI-Kommission Reinhaltung der Luft (1987) Aktuelle Aufgaben iker MeJtechnik in der Luftrcinhaltung. Dusseldorf: VDI-Verlag. [39] Birkle. H. (1979) M
2.6 Appendix The following market survey (Tabs. 1 - 3 ) was mainly compiled from Germany and the Western European region.

electrical 220 v yes PC, printer. software. ultrasound bath for dispersing industry, quality control environmental protection, research, control laboratory solid grain size sediment-photometric

mechanical and electrical 220 v -

extensive accessories for air. water and ground samples

mechanical, electrical limited -

5-1OOOmLor 1Olg-2Og

solid and liquid as needed according to customer equipment and wishes according to requirements

Which phase are investigated Which components are investigated Analysis method (AAS, XRF, MS, GC, LC)

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces Cooling of sample

up to ambient temperature

-

in the sample container

Yes Yes -

water trap for liquid samples

provision exists

or given facts and standards

industry, quality control environmental protection, research. control laboratory solid, liquid and gaseous hydrocarbons, CFCs GC, GC/MS

industry, quality control

0.1 -0.5 g particle separation from the liquid phase no no no

15 kg

ca. 20 kg -

420 5601230 mm -

photosedimentometer LUMOSED Paar!Austria Retsch DM 28850.- not incl. tax

Peak Master EV CDSiFLOWCHEM FLOWCHEM from DM 2 1000. 36/48/61 cm 10 to 1000 ccm

Retsch GmbH

FLOWCHEM

Area of application

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump etc.)

Dimensions (width, height, depth) Sample space size (width, weight, depth) Weight (mass)

Debus Debus

Name of apparatus Model Manufacturer Distributor Price incl. tax -

F. Debus

Company

Table 1. Solid Phase

on-line or off-line Yes

clean 1- 2 times per week not required

cross-section or trend sampling

no

Maintenance procedure Maintenance accessories

Method of sample taking

Possibility of taking an aliquot sample

Yes rinsing, heating automatic or manual maintenance-free included with the apparatus

-

self-cleaning to a limited extent

user specific

drying properties of the material and its environment occurs, varying from material to material user specific

quartz teflon user specific

user specific

user specific

stainless steel, synthetic material stainless steel, synthetic material for the most part prevented

depending on given situation and needs seconds -minutes

yes (air, gas, liquid samples) continuous and discontinuos

yes continuous and discontinuous yes -

-

no

Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure

Materials in contact with sample Sample space sealing material Contamination risks during sampling Danger of partial sample loss

Length of sampling time

Sampling location

Possibility of sample storage in container Possibility of automatic sampling Sampling Redundancy of sampling procedure (Precaution against sample loss)

Yes

filling

-

rinsing the cuvette

YS . rinsing

no

irrelevant

none known

through gas components

-

glass/quartz

3-120min

laboratory

no discontinuous no

Yes

39

$

a

2

'4

.o

5r?

b

c,

P7

Strohlein Strohlein according to equipment

Strohlein Strohlein according to equipment

solid (gaseous possible) dust, substances contained in dust

Which phases are investigated Which components are investigated -

research, industry, environmental protection, control laboratory, emission measurements

Area of application

Analysis method (AAS, XRF, MS, GC, LC)

depending on equipment, accessories may not be necessary

Necessary accessories (e.g. pump. etc.)

GC

research, industry, environmental protection, control laboratory, emission measurements solid and gaseous hydrocarbons, PAHs, dioxins, PCB research, industry, environmental protection, control laboratory, emission measurements solid and gaseous heavy metals, hydrocarbons, PAHs, dioxins, furans GC, possibly also MS

-

electrical 220 v Yes

-

electrical 220 v yes, individual components mounted on a wheeled cabinet

-

-

control unit: 7 kg pressure pick-up: ca. 33 kg

electrical 220 v Yes

depends on probe -

depends on probe and sampling tube -

370./270/270mm pressure gauge: 650/330/360 mm

Operation of apparatus Power source (voltage) Transportable

Dimensions (width, height. depth) Sample space size (width, weight, depth) Weight (mass)

-

-

-

Strohlein Strohlein according to equipment

DPN 4

PAH 250

MRU 40

Strohlein GmbH

Name of apparatus Model Manufacturer Distributor Price incl. tax

Strohlein GmbH

Strohlein GmbH

Company

Tab. 1. (continued)

93

2 .* 3' -3

a

N 01

Method of sample taking Possibility of taking an aliquot sample

Apparatus is disposable reusable Cleaning procedure Maintenance procedure Maintenance accessories

Sample loss/time unit (e.g. via evaporation, diffusion)

Materials in contact with sample Sample space sealing material Contamination risks during sampling Danger of partial sample loss

Length of sampling time

Sampling location

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces Cooling of sample Possibility of sample storage in container Possibility of automatic sampling Sampling Redundancy of sampling procedure (Precaution against sample loss) -

automatic/isokinetic Yes

Yes

automatic/isokinetic

isokinetic Yes

-

-

-

cleaning t..; filter heal -

no Yes extraction

losses are avoided due to a cooled probe -

titanium, quartz glass head seal (titanium), silicone seal none known

depends on concentration

in chimney systems

no Yes cleaning the probe

none known (cooling to 50 "C EPA-norm) evaporation of highly volatile components which are irrelevant for the analysis

none known

-

titanium

depends on concentration

in chimney systems

Yes automatic/discon tinuous Yes

no discontinuous Yes

no no Yes Yes

no no yes YCS

depends on concentration ca. 4 m3/h included in the apparatus

varied 5 - 50 m3/h included in the apparatus

no Yes -

-

none known

stainless steel, titanium teflon or graphitized seals none known

in chimney systems (behind smoke filters, incinerators, etc.) depends on concentration of dust

-

yes automatic/discontinuous

no no no Yes

2-40 m3/h included in the apparatus

-

-

electrical 220 v no possibly a dilutoriperistaltic pump or other conveyor systems

research, industry, quality control, environmental protection, control laboratory liquid various

Dimensions (width. height, depth) Sample space size (width, weight, depth) Weight (mass)

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

Area of application

Analysis method (AAS, XRF, MS: GC, LC)

Which phases are investigated Which components are investigated

AAS, ICP, LC

M221 /M222 (sample dispenser without valve) M231/4.1 (autoinjectors with injection valve) Gilson Medical Electronics. France ABIMED Analysentechnik GmbH 12100 DM (M221); 13975 DM (M222) 18400 DM (231); 25 570 DM (M232 Bio-version)

Name of apparatus Model

Manufacturer Distributor Price incl. tax

ABIMED

Liquid Phase

Company

Tab. 2.

industry, quality control, environmental protection, control laboratory liquid nutritive substance parameters, substances contained in beer, wine. etc. XRF, AAS

electrical 220 v no possibly a pump to wash sample containers, possibly a control unit (if no control via analysis system)

17 kg

-

370/330/540 mm

Alliance Instruments S.A.. France Alliance Instruments GmbH 12312 DM

Alliance Autosampler 52/104

Alliance Instruments

research, industry. quality control, control laboratory liquid liquid

electrical 220 v no sample container

On Line Systeme ASA GmbH 20000 to 30000 DM

OLS-Mixer -

~~

ASA GmbH

a

a

N

Possibility of taking an aliquot sample

Method of sample taking

septums for sample containers none known

Sample space sealing material Contamination risks during sampling Danger of partial sample loss Sample loss/time unit ( e g via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure Maintenance accessories

Yes

-

no Yes rinsing not necessary -

now known depends on the sample

stainless steel, teflon, ceramic

Materials in contact with sample

Yes

-

-

suction through pump (peristaltic pump, for example) Yes

-

rinsing maintenance-free

none known closed system

none known

-

VA

material flow ca. 20 min

Yes continuous Yes

Yes no no Yes

up to 18 mL -

no Yes rinsing cleaning

needle (for very small sample volumes) evaporation depends on the room temperature

glass or PS sample containers, sample needle made of steel, PE, or glass

beside the analysis system depends on measurement parameter

Yes continuous no

yes continuous and discontinuous no

laboratory depends on sampling volume

Yes no no no

0.2 - 6 mL depends on peristaltic pump

no no Yes Yes

solid phase extraction

-

1 pL- 1000 mL

Sampling location Length of sampling time

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces Cooling of sample Possibility of sample storage in container Possibility of automatic sampling Sampling Redundancy of sampling procedure (Precaution against sample loss)

liquid -

Which phases are investigated Which components are investigated

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops. gas) Homogenization of sample no Reduction of sample into small pieces no no Cooling of sample

-

industry. quality control

Area of application

Analysis method (AAS, XRF, MS, GC, LC)

no sample container

-

-

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump. etc.)

Dimensions (width, height, depth) Sample space size (width, weight, depth) Weight (mass)

On Line System ASA GmbH 150000 to 300000 DM

industry, environmental protection, control laboratory liquid pH-value, conductivity, temperature, oxygen content, cloudiness

industry, environmental protection

no no no

-

no no Yes

-

20-350 mL -

200 L/h coarse screen filtering

GC, FID

-

electrical 220 v no conveyor pump

electrical 220 VACi200 VA no possibly a pump for liquids

liquid hydrocarbons. chlorohydrocarbons

ca. 170 kg

1360/1640/620 mm 570/1000/500 mm

30 kg

-

4601 1000i340 mm

10032 DM

-

Bayer Diagnostic -

Outflow measuring station PRF-MOS MS 12/T Biihler

-

Compur-Stripper

STONE 1

Name of apparatus Model Manufacturer Distributor Price incl. tax

-

Buhler GmbH

Bayer Diagnostic

ASA GmbH

Company

Tab. 2. (continued)

3

g

&

%

0

W

large quantities of dirt none known -

no

none known

none known no loss

no

Possibility of taking an aliquot sample

Method of sample taking

Maintenance accessories

none known not known

-

-

Yes

-

yes rinsing cleaning

PVC, glass, ABS, silicone rubber sealing on PE-frame none known

glass, PVC

VA

-

continuous sampling from drainage/cooling water, and expulsion of hydrocarbons, chloroh ydrocarbons for measurements in GC/FID for example

rinsing response test with pL-injection

Yes

-

water softening principle

calibration of the measuring systems none

wiping YeS

no

-

Materials in contact with sample Sample space sealing material Contamination risks during sampling Danger of partial sample loss Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

-

equal to duration of sample material flow

outflow purification

Yes discontinuous -

Length of sampling time

industrial area, such as drainage canal, cooling water canal, and also Ex-Zone 1

-

continuous

-

no

isokinetic in material flow

yes continuous yes

Yes

Sampling location

Possibility of sample storage in container Possibility of automatic sampling Sampling Redundancy of sampling procedure (Precaution against sample loss)

3c

%’

%

3

2

2

e

3

b

e

2

g

spot check 20 - 330 mL -

-

20- 350 mL -

-

Sample size (mass or volume) spot check 20 - 330 mL Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces Yes Cooling of sample

-

-

liquid -

Which phases are investigated Which components are investigated Analysis method (AAS, XRF, MS, GC, LC) -

research, industry, environmental protection, industrial laboratory

Area of application

electrical 220 v no distributor, sample bottles

liquid -

electrical 220 v Yes distributor, sample bottles

electrical 20 v no distributor. sample bottles

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

ca. 91 kg

research, industry, environmental protection, industrial laboratory liquid

ca. 91 kg

ca. 91 kg

680/1430/540 mm 580/850/420 rnm

-

-

Vacuum sampler PP-92 Biihler

Biihler GmbH

research, industry, environmental protection, industrial laboratory

6801540 mm 580/850/420 mm

-

-

680/1430/540 mm 580/850/420 mm

-

-

Dimensions (width, height, depth) Sample space size (width, weight, depth) Weight (mass)

Sampler PRF PRF-MOS Biihler

Vacuum sampler PP-MOS Biihler

Name of apparatus Model Manufacturer Distributor Price incl. tax

Biihler GmbH

Biihler GmbH

Company

Tab. 2. (continued)

a

3

$'

i .

?

N

W

Method of sample taking Possibility of taking an aliquot sample

Maintenance accessories

Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

Sample space sealing material Contamination risks during sampling Danger of partial sample loss

Materials in contact with sample

Length of sampling time

Sampling location

Possibility of sample storage in container Possibility of automatic sampling Sampling Redundancy of sampling procedure (Precaution against sample loss)

vacuum principle -

no Yes cleaning upon inspection maintenance-free, clean now and then, inspect moving parts

no Yes cleaning upon inspection maintenance-free, clean now and then, inspect moving parts -

vacuum principle -

-

no Yes cleaning upon inspection maintenance-free, clean now and then, inspect moving parts none

volatile material through stripping during suction of the sample not known, samples are taken daily as a rule

glass, silicone, PVC, ABS, on inquiry: PVDF, stainless steel rubber sealing on PE-frame none known

at open drains, wells, basins, max. 8 m suction height 50- 110 s

Yes discontinuous -

water softening principle

-

not known

none known

volatile material through stripping during suction of the sample

not known, sammples are taken daily as a rule

glass, silicone, PVC, ABS, on inquiry: PVDF, stainless steel rubber sealing on PE-frame none known

glass, silicone, PVC, ABS, on inquiry: PVDF, stainless steel rubber sealing on PE-frame none known

connection to pumps, pressure line, hydraulic pressure ca. 20 s

-

-

at open drains, wells, basins, max. 8 m suction height 50-110s

Yes discontinuous

yes discontinuous

Yes

150 kg

ca. 13 kg

ca. 22 kg

Sample size (mass or volume) 20 - 200 mL Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces no Cooling of sample

-

Yes

-

-

no

spot check 20-330 mL -

-

research, industry, environmental protection, control laboratory liquid -

electrical 220 v no distributor, sample bottles

20-200 mL

-

-

-

liquid

liquid

Which phases are investigated Which components are investigated Analysis method (AAS, XRF, MS, GC, LC)

research, industry, environmental protection, control laboratory

research, industry, environmental protection, control laboratory

-

-

Storage battery 24 Vi6.5 AA Yes

-

Storage battery 24 V/5.7 AA Yes charging device

Area of application

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

930,!1650/720 mm 830/1000/600 mm

300/5001200 mm separate, diameter 3181140 mm

450/720/385 mm 450/445/385 mm

-

Self-emptying vacuum sampler PP-MOS DRA Biihler -

Dimensions (width, height. depth) Sample space size (width, weight, depth) Weight (mass)

Vacuum sampler PBT Buhler

Vacuum sampler PB-MOS Biihler

Biihler GmbH

Name of apparatus Model Manufacturer Distributor Price incl. tax

Biihler GmbHc

Biihler GmbH

Company

Tab. 2. (continued)

2

3-

.3 *

?J

P

W

Method of sample taking Possibility of taking an aliquot sample

Maintenance accessories

vacuum principle -

vacuum principle -

-

no Yes wiping charging the storage battery, now and then inspect moving parts

no

Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

Yes wiping charging the storage battery, now and then inspect moving parts -

volatile substances through stripping during suction of the sample not known

volatile substances through stripping during suction of the sample not known

vacuum principle -

-

no Yes cleaning upon inspection maintenance-free, clean now and then, inspect moving parts

volatile substances through stripping during suction of the sample not known, samples are taken daily as a rule

-

none known

-

none known

Sample space sealing material Contamination risks during sampling Danger of partial sample loss

glass, silicone, PVC, ABS, on inquiry: PVDF, stainless steel rubber sealing on PE-frame none known

glass, silicone, PVC, ABS

glass, silicone, PVC, ABS

Materials in contact with sample

at open drains, wells, basins, max. 3 m suction height 50-110s

-

-

at open drains, wells, basins, max. 6 m suction height 50-110s

Yes discontinuous

Yes discontinuous

Length of sampling time

-

yes discontinuous

Yes

at open drains, wells, basins, max. 8 m suction height 50-110s

Sampling location

Possibility of sample storage in container Possibility of automatic sampling Sampling Redundancy of sampling procedure (Precaution against sample loss)

5ss$

$

5

%

4

g

ca. 37 kg

27 kg

ca. 45 kg

-

5-25 mL -

-

10- I 20 mL no no no

liquid -

13 mL

no no no

Which phases are investigated Which components are investigated Analysis method (AAS, XRF, MS, GC, LC)

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces Cooling of sample -

no no no

-

research, industry, environmental protection, control laboratory liquid

industry, environmental protection

Area of application

-

environmental protection, control laboratory liquid

mechanical compressed air Yes -

electrical 220 v no

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

electrical 220 v no sample cylinde~

500/790/300 mm diameter 320/400 mm

diameter 420/8 15 mm diameter 420/400 mm

no separate space

600,/800/300 mm

-

Dimensions (width, height, depth) Sample space size (width, weight, depth) Weight (mass)

Sludge sampler SPN-MOS Biihler

Canal sampler Pk-ex Biihler -

Sampler Toskana Biihler

Name of apparatus Model Manufacturer Distributor Price incl. tax

Biihler GmbH

Biihler GmbH

Biihler GmbH

Company

Tab. 2 (continued)

a

o\

w

PVC, glass, ceramic none known

Materials in contact with sample

Sample space sealing material Contamination risks during sampling Danger of partial sample loss

flat slide valve system

Method of sample taking -

none

Maintenance accessories

Possibility of taking an aliquot sample

yes. wiping maintenance-free

-

vacuum principle

no Yes wiping refill compressed air, operate regularly -

ball stop-cock between sample cylinder with separating flask and drain -

none

maintenance-free

-

no Yes

not known

no

not known

none known

volatile material through stripping during suction of the sample not known

none known

Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

none known

stainless steel, PTFE, ceramic, glass, PVC

10 s

none known

-

PVC, glass, silicone

10 s

Length of sampling time

on sludge pressure lines

-

-

in drains max. 3 m suction height 40 s

Yes discontinuous

Yes discontinuous

in pressure lines, at by-passes

yes discontinuous -

Yes

Sampling location

Possibility of sample storage in container Possibility of automatic sampling Sampling Redundancy of sampling procedure (Precaution against sample loss)

~~

up to 5 L,'min screening filters (optional)

no

-

24x1 L/8x2L/2xIOL 3500 mL,'min suction basket

no

-

24x1 L/4x4L/1 x9.5L 3500 mL/min suction basket

no

Analysis method (AAS, XRF, MS, GC, LC)

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample

Which phases are investigated Which components are investigated

research, industry, environmental protection liquid heavy metals, hydrocarbons

research, industry, environmental protection liquid heavy metals, hydrocarbons

-

liquid heavy metals, hydrocarbons. CFCs, and everything that can occur in ground water -

environmental protection

Yes system consists of: pump, lines, control unit. compressed air supply

-

Area of application

pneumatic

Colora MeDtechnik GmbH from 7000 DM

1 x 0

Ground water sampling apparatus Accu Well

Colora MeBtechnik GmbH

~

electrical 220 V, low voltagecurrent, battery Yes suction line, power supply

~~

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

-

IlOkg

16 kg electrical 220 V no

70/1300/700 mm 600/200/400 mm

diameter 505/H: 640 mm diameter 505/H : 200 mm

Dimensions (width, height, depth) Sample space size (width, weight, depth) Weight (mass)

Model Manufacturer Distributor Price incl. tax

Water and waste water sampling apparatus PNTS Colora MeBtechnik GmbH Colora MeBtechnik GmbH 10900 DM

Water and waste water sampling apparatus 3700 ISCO. Lincoln, NE-USA Colora MeDtechnik GmbH, Lorch 7524DM

Name of apparatus

~

Colora MeDtechnik GmbH

~

Colora MeBtechnik GmbH

(continued)

Company

Tab. 2.

3

$'

5 k

Q

s

30

w

polypropylene, vinyl, teflon, medical silicone tube -

Materials in contact with sample

Sample space sealing material Contamination risks during sampling Danger of partial sample loss Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure Maintenance accessories

Method of sample taking Possibility of taking an aliquot sample

depends on problem at hand

Length of sampling time

sampling from water and waste water Yes

from ground water Yes

-

-

sampling from water and waste water Yes

no Yes rinsing clean regularly

no Yes rinsing change pump-tube

no Yes rinsing change hose none

-

-

none known

stainless steel, teflon, acetalsynthetic, polypropylene depending on configuration and analysis silicone or teflon none known

ground water level from 2 inches depends on the depth

none known -

none known

polypropylene, vinyl, teflon, medical silicone tube

-

waste water canals

none known

none known

water-, waste water canals

no continuous Yes

yes discontinuous yes Yes discontinuous Yes

no no Yes

no possible Yes

Sampling location

Reduction of sample into small pieces Cooling of sample Possibility of sample storage in container Possibility of automatic sampling Sampling Redundancy of sampling procedure (Precaution against sample loss)

$

%

F

2

b

8 2

from 480 DM 8 different models with different sample volumes, made of pure teflon mechanical

Price incl. tax

Dimensions (width, height, depth) Sample space size (width, weight, depth) Weight (mass)

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

~~~

24 x 0.5 L/12 x 0.5 L / l x 3.8 L 2500 mL/min suction basket

-

from 300 mL screening filter (optional)

-

-

liquid heavy metals, hydrocarbons, CFC

Which phases are investigated Which components are investigated Analysis method (AAS, XRF, MS, GC, LC)

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas)

research, industry, environmental protection liquid heavy metals, hydrocarbons

environmental protection

20 to 500 mL/sample adjustable 14 L collection container, or sample bottles ( 4 x 10 L. or 1 2 x 2 L, or 24 x 0.5 L)

industry, environmental protection liquid substances in waste water -

electrical 220 V, battery Yes built-in

ca. 10 kg

10 kg electrical 220 V. low voltagecurrent, battery Yes suction line, power supply

360/270/180 mm -

+

~~~

LIQUI-BOX,ASP 9461 E + H Wetzer E + H MeDtechnik GmbH co. ca. 4000 - 6500 DM depending on configuration

Sample collector

Endress & Hauser

diameter 440/H: 590 mm diameter 440/H : I80 mm

5500 DM

Water and waste water sampling apparatus 2900 ISCO Colora MeDtechnik GmbH, Lorch

~

Colora MeBtechnik GmbH

Area of application

Yes retrieval cord

-

Test sample container for ground water Accu Well ISCO Colora MeDtechnik GmbH, Lorch

Name of apparatus

Model Manufacturer Distributor

Colora MeBtechnik GmbH

(continued)

Company

Tab. 2.

9

n

9

%

a

g

-

no yes rinsing changing the tube

none known -

no Yes rinsing, heating -

Maintenance accessories from ground water Yes

none known

teflon none known

Sample space sealing material Contamination risks during sampling Danger of partial sample loss Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

Method of sample taking Possibility of taking an aliquot sample

-

polypropylene, vinyl, teflon, medical silicone tube none known

teflon

Materials in contact with sample

sampling from water and waste water Yes

-

none known not known

water and waste water canals depending on problem at hand

ground water level from 1 inch < 10 min

Sampling location Length of sampling time

vacuum not relevant

rinsing depending on medium ca. once per month -

-

none known

-

PVC, PE, glass (as standard materials)

waste water canals, basins, etc. sampling cycle 1 - 2 min

Yes discontinuous Yes

no discontinuous yes Yes discontinuous no

no no possible Yes

no no no Yes

Homogenization of sample Reduction of sample into small pieces Cooling of sample Possibility of sample storage in container Possibility of automatic sampling Sampling Redundancy of sampling procedure (Precaution against sample loss)

39

5

.q $

s

9

h

5

P3

11 kg

ca. 100 kg

10pg-650mg

not necessary not necessary

liquid substances in waste water -

20 - 500 mL/sample; 10 - 30 L container, sample changer: 4 x 10 L to 1 L no

Which phases are investigated Which components are investigated Analysis method (AAS, XRF, MS, GC, LC)

Sample size (mass or volume)

Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample

research, industry, quality control, environmental protection, control laboratory liquid, solid hydrocarbons, CFC GC, MS, FTIR

industry, environmental protection

Area of application

-

included with the apparatus

-

electrical 220 v no built-in

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.) electrical 220 v

40/14/33 cm L: 25 mm x 1.9 or 6 mm diameter

600/1000/500 mm

Dimensions (width, height, depth) Sample space size (width, weight. depth) Weight (mass) -

Pyro sample 2000 CDSiFLOWCHEM FLOWCHEM ca. 30000 DM

Sample collector ASP-9465, ASP-Station A E + H Wetzer E + H MeBtechnik GmbH + Co. 5500 to 12000 DM depending on configuration

Name of apparatus Model Manufactur Distributor Price incl. tax

FLOWCHEM

Endress & Hauser

(continued)

Company

Tab. 2.

not necessary

not necessary

10 kg-650 mg

research, industry. quality control. environmental protection. control laboratory liquid, solid hydrocarbons, CFC GC, MS, FTIR

electrical 220 v included with the apparaturs

44/14/33 cm L: 25 mm x 1.9 or 6 mm diameter 11 kg

Pyro sample 1000 CDSiFLOWCHEM FLOWCH EM ca. 20000 DM

FLOWCHEM

sa3'

&

Q

."a

$3

Method of sample taking Possibility of taking an aliquot sample

Maintenance accessories

discontinuous -

discontinuous -

-

-

vacuum not relevant

-

rinsing depending on medium, ca. once a month

F

Yes

W P

2

2 3 3

9

5 -

-

Yes

E

7

2 b

Yes heating at 1400 "C maintenance-free

Yes heating at 1400 "C maintenance-free

-

-

-

-

-

none known not known

vitron, graphite, vespel -

vitron, graphite, vespel

quartz glass, quartz wool

-

quartz glass, quartz wool

user specific user specific

-

user specific user specific

not necessary -

-

not necessary

none known

PVC, PE, glass (as standard materials)

Materials in contact with sample

Sample space sealing material Contamination risks during sampling Danger of partial sample loss Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

waste water canals, basins, etc. sampling cycle at least 2 min

yes discontinuous no

no Yes Yes

Sampling location Length of sampling time

Reduction of sample into small pieces Cooling of sample Possibility of sample storage in container Possibility of automatic sampling Sampling Redundancy of sampling procedure (Precaution against sample loss)

GAT-BIOAS 3V

Name of apparatus

Which phases are investigated Which components are investigated Analysis method (AAS, XRF, MS,GC, LC)

biotechnology, research, industry, environmental protection, quality control, control laboratory liquid whatever, depends on analysis procedure XRF, LC

ca. I 0 0 kg (ca. 21 kg) (values in parentheses for portable apparatus)

12.8 kg

Area of application

620/400/650rnm (350/400/360mm)

-

Sample space size (width, weight, depth) Weight (mass)

-

industry, environmental protection, purification plants, fish-hatcheries, water treatment plants liquid water, sewage

electrical 220 v (220124 V) Yes optional rolling foundation, mixed sample container and fractional bottles, 5m tube, screening filter and suction pipe

720/1500/760mm (370/850/380mm)

250/165/420 mm

Dimensions (width, height, depth)

mechanical, electrical 220 v no

ca. 10250DM

33972 D 1

Price incl. tax

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

GIMAT GmbH

GAT

Distributor ca. 7 30 Dm

GIMAT GmbH

GAT

Sampling apparatus FSV (FSV-TB) stationary portable ’

GIMAT GmbH

Manufacturer

Model

Gamma Analysentechnik

(continued)

Company

Tab. 2.

liquid C , Si, S, Mn, P, Al, transition metals, Zr spark or laser erosion in combination with ICP-AES, XRF, F-OES, GD-OES

research, industry, quality control, control laboratory

mechanical Yes pump for suction probe

ca. S0g-2kg

diameter 40- 80/L: 100- 2000 mm diameter 35 x 100 mm

MINKON Sampler-Technik GmbH MINKON Sampler-Technik GmbH 1 to 6 DM/2.25 to 100 DM

1. casting ray sample, immersion and suction probes 2. probes for subtances

MINCO-probes

Minkon

no Yes self-cleaning through back-rinsing and blowing out cleaning sample container when the medium is dirty

no Yes rinsing

suction through use of vacuum Yes

Method of sample taking

Possibility of taking an aliquot sample

Maintenance accessories

Maintenance procedure

vacuum suction of sample with 1.5-2.5 bar Yes

-

none known not relevant

none known -

none known

destruction of sample taking device no

none known -

steel

synthetic container PU-foam, 45 mm V2 A-case type of protection: IP 65 none known

stainless steel, glass, PEEK, tenon

Materials in contact with sample Sample space sealing material

Contamination risks during sampling Danger of partial sample loss Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure

steel, quartz, glass steel

depending on sample volume, < 1 min per sample

depending on the program

Length of sampling time

-

blast furnace channel to chill mold SGA, from all melting furnaces, treatment pans and installations max. 10 s

outdoor assembly directly beside the medium

mainly, bioreactors, sample-taking location

Yes discontinuous no

Yes continuous and discontinuous no

ca. l 0 0 g ca. 100,s/sample, protection cap as slag protection no no Yes Yes

50-500 L adjustable: time, amount, event

Yes no no no

particle separation

-

0-1 mL

Sampling location

Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces Cooling of sample Possibility of sample storage in container Possibility of automatic sampling Sampling Redundancy of sampling procedure (Precaution against sample loss)

Sample size (mass or volume) Output capacity

vl P

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas)

Which phases are investigated Which components are investigated Analysis method (AAS, XRF, MS, GC, LC)

Area of application

sample-taking device allows horizontal sampling, layer syphon

-

0.5 L

environmental protection, control laboratory

Yes plumb-line

mechanical

-

none

-

5- 500 mLjsuction process

-

-

5 - 500 mL:suction process none

environmental protection, control laboratory, inlet monitoring purification plants liquid

environmental protection, control laboratory, inlet monitoring purification plants liquid

electrical battery Yes programmer, charging device, suction tube (with filtering basket variable) various sample bottle sizes

electrical battery Yes programmer, charging device, suction tube, sample bottles

2.8 kg

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

ca. 10 kg

ca. 10 kg

0.5 L

Sample space size (width, weight, depth) Weight (mass)

~

diameter 360 x 565 mm diameter 396 x 835 mm max. 12 L

~~

diameter 360 x 565 mm diameter 396 x 835 mm max. 1 2 L

40 x 865 mm

Dimensions (width, height, depth)

EPIC 1011 T Epic Products Ltd. Regler- u. Verfahrens GmbH 4700 to 7500 DM depending on configuration

EPIC 1011 Epic Products Ltd. Regler- u. Verfahrens GmbH 4700 to 7500 DM depending on configuration

Sampler V2A hysp 0.5 Nordmeyer KG Nordmeyer KG 1619 DM

Name of apparat us Model Manufacturer Distributor Price incl. tax

Regler- u. Verfahrens GmbH

Regler- u. Verfahrens GmbH

Nordmeyer K G

Company

Tab. 2. (continued)

i:

2

9

;r

a

8

Possibility of taking an aliquot sample

Method of sample taking

scooped sample with layer-specific sampling possibility Yes

vacuum sample taking device -

vacuum sample taking device -

no yes rinsingnone, possibly cleaning the measuring chamber none

no Yes rinsing none, possibly cleaning the measuring chamber none

no Yes rinsing, heating cleaning

Maintenance accessories

none known depends on conditions and user

none known depends on specific application and user

improper handling

-

Buna “N” none known

V2 A (PaBschliff) none known

teflon, glass, silicone, stainless steel Buna “N” none known

Sample space sealing material Contamination risks during sampling Danger of partial sample loss Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

silicone, PVC, PE, stainless steel

typically 1 min depending on conditions

variable

Yes continuous and discontinuous yes ( 3 attempts)

no no no Yes

V2 A

typically 1 min and more, depending on application and conditions

variable

Yes continuous and discontinuous yes (3 attempts when outlet tube is clogged)

no discontinuous no

bodies of water, ground water observation wells 1 - 15 min depending on depth of sample taking location

no no no Yes

no no no no

Materials in contact with sample

Length of sampling time

Sampling location

Homogenization of sample Reduction of sample into small pieces Cooling of sample Possibility of sample storage in container Possibility of automatic sampling Sampling Redundancy of sampling procedure (Precaution against sample loss)

e

2

B

i:

b

3

environmental protection, control laboratory, purification plants

liquid -

Area of application

Which phases are investigated Which components are investigated

5 - 500 mL/suction process none

-

electrical 220 v stationary (but also transportable) programmer, bottle basket with bottles

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas)

environmental protection, control laboratory, inlet monitoring, Ex-Zone 1 liquid

ca. 60 kg

-

ca. 10 kg

max. 24 L

Sample space size (width, weight, depth) Weight (mass)

Analysis method (AAS, XRF, MS, GC, LC)

diameter 360 x 565 mm diameter 396 x 835 mm max. 12 L

860/250/790 mm

Dimensions (width, height, depth)

-

none

-

0.01 -0.15 L/h

liquid accumulation of sorption agents in flowing waters GC

-

electrical 220 v Yes integrated

ca. 15 kg

-

510/430/390 mm

Strohlein Strohlein ca. 20000 DM

-

UPN 8

Strohlein GmbH & Co.

5 - 500 mL/suction process

-

electrical battery yes programmer, charging device, suction tube, sample bottles, external battery

EPIC (explosion proof) 1511 Epic Products Ltd. Regler- u. Verfahrens GmbH 10450 to 13 115 DM depending on version

~

Regler- u. Verfahrens GmbH

EPIC 1021-v Epic Products Ltd. Regler- u. Verfahrens GmbH 11230 to 13500DM depending on version

~

Regler- u. Verfahrens GmbH

Name of apparatus Model Manufacturer Distributor Price incl. tax

Company

Tab. 2. (continued)

'a

&

vacuum pump -

vacuum pump -

no no Yes Yes rinsing rinsing possibly cleaning measuring chamber none

no Yes rinsing none

Method of sample taking Possibility of taking an aliquot sample

none known -

none known

-

PTFE, silicone, glass, stainless steel none known

glass, teflon, PE, silicone, steel Buna “N” none known

Yes

through hoses with piston pump

-

none known

none known

-

glass, teflon, nickel

-

Materials in contact with sample Sample space sealing material Contamination risks during sampling Danger of partial sample loss Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure Maintenance accessories

in flowing waters

variable typically 1 min

variable according to sampling area typically 1 min

Sampling location Length of sampling time

yes (3 attempts)

Yes continuous, 8 switchable channels Yes

Yes continuous and discontinuous

yes (3 attempts)

no no no Yes

no no no yes

Redundancy of sampling procedure (Precaution against sample loss)

Homogenization of sample Reduction of sample into small pieces Cooling of sample Possibility of sample storage in container Possibility of automatic sampling Yes continuous and discontinuous Sampling

W P

GQ.

2 3

‘p, 3

2

2

k

%

3

g

Tab. 3.

ca. 50cm3

135/300 mm (cylindrical)

22 kg mechanical 220 v Yes control unit

gas hydrocarbons, CFC, HCI, HF, gaseous emissions NDIR, FID, continuous measuring procedures

135/300 mm (cylindrical) ca. 50 cm3

21 kg

electrical 220 v Yes

industry

gas HCI, HF, gaseous emissions, hydrocarbons, CFC

GC, NDIR, FID, continuous measuring procedures

Dimensions (width. height, depth) Sample space size (width, weight, depth) Weight (mass)

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

Area of application

Which phases are investigated Which components are investigated

Analysis method (AAS, XRF, MS, GC, LC)

industry

Gas sampler GS 312 Article-Nr. 170300 Desaga GmbH Desaga GmbH 9063 DM

COMPUR dilution probe Type 21 1 Bayer Diagnostic Bayer Diagnostic 9901 DM

COMPUR probe Type 201 Bayer Diagnostic Bayer Diagnostic 7222 DM

Name of apparatus Model Manufacturer Distributor Price incl. tax

research, industry, quality control, environmental protection, control laboratory, incinerator, power plants purification plants, trade inspection offices, technical supervision agencies gas emission and irnmission measurements, MAK-value monitoring, investigation of process gases AAS, RFA, MS, GC, LC

electrical 220 V, battery Yes complete apparatus e.g. water bottles are required for sample taking

13 kg

410:220/330 mm -

J . Desaga GmbH

Bayer Diagnostic

Gaseous Phase

Bayer Diagnostic

~

Company

~

3

$'

3 ,a

.a

variable, measurements are possible in any surrounding, independent of electric networks, also for mobile units adjustable duran glass tenon negative negative negative

-

chimney process operation

continuous high-grade steel 14571, NDF silicon carbide, sapphire, fluorosint PVDF none known none known -

no Yes blow filter cell clean changing filter cell, testing the dilution ratio special key included with apparatus tube in exhaust canal -

chimney process operation

continuous high-grade steel 14571, PVDF silicon carbide PVDF none known none known

no Yes blow filter cell clean changing the filter cell special key included with apparatus tube in exhaust canal -

Sampling location

Length of sampling time

Materials in contact with sample

Sample space sealing material Contamination risks during sampling Danger of partial sample loss Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

Maintenance accessories

Method of sample taking Possibility of taking an aliquot sample

-

microprocess-driven Yes

Yes continuous yes

Yes

-

continuous -

no no

no

continuous -

1-9999 L 0.2 - 12 L/min -

continuous up to 2000 L/h none, is carried out automatically by the system no no no no

continuous 1000 L/h none, is carried out automatically by the system no

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces Cooling of sample Possibility of sample storage in container Possibility of automatic sampling Sampling Redundancy of sampling procedure (Precaution against sample loss)

vl

-

Model Manufacturer Distributor Price incl. tax

-

ca. 10 kg

6g

mechanical Yes gas tester pumps, Quantimeter 1000, Polymeter, Accura, Accura 1000

research, industry, environmental protection, control laboratory. occupational safety

gas hydrocarbons, CFC. solvents of every kind GC

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

Area of application

Which phases are investigated Which components are investigated

GC, LC, DC

gas hydrocarbon

research, industry

mechanical yes syringe with hypodermic needle

L: 75 mm

-

LABC-Labortechnik LABC-Labortechnik NS 29 19.30 DM/ apiece NS 18.8 18.70 DMiapiece NS 14.4 18.10 DMiapiece All with PE-perforated caps

small glass tube 8 mm diameter

Analysis method (AAS, XRF, MS, GC. LC)

~

LABC-beaded rim adaptor for NS 29/18,5/14.5

~

LABC Labortechnik

Dimensions (width, height, depth) Sample space size (width, weight, depth) Weight (mass)

Dragerwerk Dragerwerk 10 pieces 40 DM

Small sampling tube (tilled with activated charcoal. silica gel)

Name of apparatus

~

Dragerwerk AG

(continued)

Company

Tab. 3.

GC. HPLC

research, industry, environmental protection. control laboratory. immission measurements gas hydrocarbon, aldehydes

electrical 220 v Yes -

-

500/420/570 mm

S t ro hlein Strohlein ca. 20000 DM

IPN

Strohlein GmbH & Co.

a

w

wl

Yes discontinuous, via switchable channels Yes outside airs ca.3 h

no discontinuous

no Yles . nnsing

-

Yes no

Method of sample taking Possibility of taking an aliquot sample

Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure Maintenance accessories

liquid desorption Yes

-

-

with syringe Yes

yes . nnsing -

not known (probably very slight)

none known

-

no

none known

PE-caps none known

Sample space sealing material Contamination risks during sampling Danger of partial sample loss

outside air measurement Yes

chemical reaction or adsorption

selective for aldehydes

-

glass. teflon, nickel

V4 A hypothermic needle synthetic (PE) or glass PTFE. rubber none known

gas, activated charcoal, silica gel

Materials in contact with sample

~~

from reaction flasks seconds

no

-

no no no Yes

ca. 180 L 1 L/min

no no no

none

-

variable

anyplace 10 min-8 h

110

Yes continuous and discontinuous

no no no Yes

1-100L 0.2-1.5 L

Sampling location Length of sampling time

Redundancy of sampling procedure (Precaution against sample loss)

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction ofsample into small pieces Cooling of sample Possibility of sample storage in container Possibility of automatic sampling Sampling

Q

F

W

ul

x9'

e.

B

ti

rg

(continued)

~

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces Cooling of sample

Analysis method (AAS. XRF, MS, GC, LC)

Which phases are investigated Which components are investigated

Area of application

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

Dimensions (width, height, depth) Sample space size (width, weight, depth) Weight (mass)

Name of apparatus Model Manufacturer Distributor Price incl. tax

Company

Tab. 3.

-

GC, NPD 1.5-48 L 0.0 1 - 0.1 L/min -

no no no

research, industry, environmental protection, occupational safety gas formaldehyde NIOSH 2502 GC, FID I-15L 0.01 -0.05 l/min -

no no no

Yes Pump research, industry, environmental protection, occupational safety gas acrolein (NIOSH 2501)

ca. 15g

ca. l o g -

6/4/100 mm l20/160 mg

Small sampling tube Orbo-23 SUPELCO SUPELCO 25 pieces 180.12 DM

SUPELCO

6841100 mm 120/160 mg

Orbo-22 SUPELCO SUPELCO 25 pieces 182.40 DM

-

SUPELCO

a

2

c

c 3.

'a

s

excessive flow rate and/or sampling time

Possibility of taking an aliquot sample

Method of sample taking

Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure Maintenance accessories -

breaking open at sampling location Orbo-tube-cutter

-

breaking open at sampling location Orbo-tube-cutter

elution of both beds (separately) with toluene yes (after elution)

Yes no

Yes no

elution of both beds (separately) with isooctane yes (after elution)

-

-

phenol, acid fumes

-

excessive flow rate and/or collection time

Sample space sealing material Contamination risks during sampling Danger of partial sample loss

glass, silanized glass wool, Supelpak 20 N (20/40 mesh) impregnated with 2-(hydroxymethy1)piperidine other aldehydes/ketones

glass. glass wool, N-benzylethanolamine on SUPELPAK 20F (20/40 mesh)

Materials in contact with sample

anyplace 15 min-80 h

yes (depends on pump used) continuous and discontinuous yes (depends on pump used) Yes (2 collecting beds)

yes (depends on pump used) continuous and discontinuous yes (depends on pump used) yes ( 2 collecting beds)

anyplace

yes, 4 weeks at 25 "C

yes, 6 weeks at 25 'C

Sampling location Length of sampling time

Redundancy of sampling procedure (Precaution against sample loss)

Possibility of sample storage in container Possibility of automatic sampling Sampling

$ *

3

2

%

2

2

L

B

2

g

Yes Pump

-

no no

no no

no no

various various

1-12L 0.01 -0.05 L/min -

3-48 L 0.1 -0.2 L/min -

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces

Analysis method (AAS, XRF, MS, GC, LC)

gas acetaldehyde, NIOSH 2538 OSHA 68 GC, FID

gas formaldehyde, acrolein OSHA 52 GC, NPD

Which phases are investigated Which components are investigated

research, industry. environmental protection, occupational safety gas many types of organic compounds MS, GC. LC

Yes Pump

-

6/4/70 mm 100150 mg

research, industry, environmental protection, occupational safety

Yes Pump

research, industry, environmental protection, occupational safety

-

ca. 20 g

ca. SOg

-

ca. l o g

8/6/110 mm 450/225 mg

6/4/100 mm 1501175 mg

Area of application

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

Dimensions (width, height, depth) Sample space size (width, weight, depth) Weight (mass)

-

Orbo-32 (small) SUPELCO SUPELCO 50 pieces 155.04 DM

-

Orbo-25 SUPELCO SUPELCO 25 pieces 186.96 DM

-

Name of apparatus Model Manufacturer Distributor Price incl. tax

SUPELCO

Orbo-24 SUPELCO SUPELCO 25 pieces 180.12 DM

SUPELCO

SUPELCO

(continued)

Company

Tab. 3.

3

3'

2

&

a

m

u

glass, silanized glass wool, urethane foam, coconutactivated charcoal various excessive flow rate and/or collection time -

Yes no breaking open at sampling location Orbo-tube-cutter

glass, silanized glass wool, urethane foam, 2-(hydroxymethy1)piperidine on SUPELPAK 20N (20/40 mesh) none known excessive flow rate and/or sampling time -

Yes breaking open at sampling location

glass, silanized glass wool 2- (hydroxymethy1)piperidine on SUPELPAK 20N (20/40 mesh)

-

excessive flow rate and/or collection time

-

no yes

breaking open at sampling location

Orbo-tube-cutter

elution of both beds (separately) with toluene (contains D M F as intl. standard) yes (after elution)

Materials in contact with sample

Sample space sealing material Contamination risks during sampling Danger of partial sample loss

Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

Maintenance accessories

Method of sample taking

Possibility of taking an aliquot sample

anyplace various

anyplace 20 min - 20 h

anyplace 15 min-8 h

Sampling location Length of sampling time

yes (after elution)

elution of both beds (separately) with toluene

Orbo-tube-cutter

yes (after elution)

various

Yes 2 collecting beds

Redundancy of sampling procedure (Precaution against sample loss)

no

(depends on pump used) continuous and discontinuous

yes (depends on pump used) continuous and discontinuous (depends on pump used) Yes 2 collecting beds

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes 2 collecting beds

other aldehydes

no various

no 21 days 0 "C

no Yes

Cooling of sample Possibility of sample storage in container Possibility of automatic sampling Sampling

(continued)

no no no

no no no

no no no

-

-

-

0.5-5 L 0.01 -0.05 Lpnin

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles. drops, gas) Homogenization of sample Reduction of sample into small pieces Cooling of sample

gas ethyleneoxide GC, ECD

research. industry, environmental protection, occupational safety gas pesticide. USAF-method GC

Yes Pump

-

-

ca. l o g

6/4/70 mm 66/33 mg

Orbo-42 (small) SUPELCO SUPELCO 50 pieces 243.96 DM

-

SUPELCO

variable variable -

gas many kinds of organic compounds MS, GC, LC

Which phases are investigated Which components are investigated Analysis method (AAS, XRF, MS, GC, LC)

research. industry, environmental protection. occupational safety

Yes pump

Yes Pump

research, industry, environmental protection. occupational safety

-

-

ca. 20g

ca. I S g

-

8/6jl SO min 7001390 mg

8/6/100 mm 400/200 mg

-

Orbo-33 SUPELCO SUPELCO 50 pieces 264.48 D M

-

SUPELCO

Orbo-32 (large) SUPELCO SUPELCO 50 pieces 225.72 DM

SUPELCO

Area of application

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

Dimensions (width, height. depth) Sample space size (width, weight. depth) Weight (mass)

Name of apparatus M ode1 Manufacturer Distributor Price incl. tax

Company

Tab. 3.

-

2

:

.\

-0

?J

wl m

-

Yes no breaking open at sampling location Orbo-tube-cutter

excessive flow rate and/or sample collection time -

Yes no

varies yes (after elution)

Method of sample taking

Possibility of taking an aliquot sample

-

breaking open at sampling location Orbo-tube-cutter

Sample loss/time unit (e.g. via evaporation. diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure Maintenance accessories

elution of both beds (separately) with 99: 1 Benzol: CS2 yes (after elution)

excessive flow rate andior collection time

-

Sample space sealing material Contamination risks during sampling Danger of partial sample loss varies

glass, silanized glass wool, urethane foam, petroleum-activated charcoal none known

glass, silanized glass wool, urethane foam, activated charcoal

Materials in contact with sample

anyplace 10-500 min

anyplace varies

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collecting beds)

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collecting beds)

- 10 “C

4 weeks at

varies

Sampling location Length of sampling time

Redundancy of sampling procedure (Precaution against sample loss)

Possibility of sample storage in container Possibility of automatic sampling Sampling

yes (after elution)

-

Orbo-tube-cutter

excessive flow rate andlor collection time -

-

-

glass, glass wool, urethane foam, Supelpak 20E (20/40 mesh)

-

anyplace

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collecting beds)

-

h

h

ca. 1 5 g

ca. 15g

200- 1000 L 2 L/min no no no

GC, NPD 120 L 1 L/min

no no no

Analysis method (AAS, XRF, MS, GC, LC)

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces Cooling of sample -

GC, FID, HPLC, fluorescence

gas pesticides NIOSH S299

Which phases are investigated Which components are investigated

gas PAHs (NIOSH 5506, NIOSH 5525)

research, industry, environmental protection, occupational safety

Area of application

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.) research. industry, environmental protection, occupational safety

8/6/100 mm I OOj50 mg

10/8/100 mm lOOj50 mg

10 g

no

no no

10-200 L 0.5- 1 L/min

research, industry, environmental protection, occupational safety gas chlordane and other pesticides NIOSH 5510 GC, ECD

Cd.

8/6/100 mm 100/150 mg

Orbo-44 SUPELCO SUPELCO 50 pieces 257.64 DM

Orbo-43 SUPELCO SUPELCO 50 pieces 285 DM

-

Orbo-42 (large) SUPELCO SUPELCO 50 pieces 257.64 DM

Name of apparatus Model Manufacturer Distributor Price incl. tax

Dimensions (width, height, depth) Sample space size (width, weight, depth) Weight (mass)

SUPELCO

SUPELCO

SUPELCO

(continued)

Company

Tab. 3.

0

b\

Orbo-tube-cutter

elution of both beds (separately) with toluene

yes (after elution)

Maintenance accessories

Method of sample taking

Possibility of taking an aliquot sample

-

Yes no breaking open at sampling location

excessive flow rate and/or collection time

Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

Sample space sealing material Contamination risks during sampling Danger of partial sample loss

glass, silanized glass wool SUPELPAK 20P (20/40 mesh) variable P-containing compounds

Materials in contact with sample

elution of both beds (separately) with acetonitrile, benzol, cyclohexane or methylene chloride yes (after elution)

Orbo-tube-cutter

breaking open at sampling location

-

Yes no

heat, ozone, N02, UV-rays, decompose sample molecules excessive flow rate and/or collection time -

-

glass, silanized glass wool SUPELPAK 20U (20/40 mesh)

anyplace 100-500 min

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collecting beds)

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collecting beds)

anyplace 2h

Yes

-

Sampling location Length of sampling time

Redundancy of sampling procedure (Precaution against sample loss)

Possibility of sample storage in container Possibility of automatic sampling Sampling

elution of both beds (separately) with toluene yes (after elution)

Yes no breaking open at sampling location Orbo-tube-cutter

-

excessive flow rate and/or collection time

glass. silanized glass wool SUPELPAK 20E (20/40 mesh) none known

anyp 1ace 10-40 min

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collecting beds)

> 1 week at 25 C

$

s9ro

$

-r

$

P b

T

p

P-pestizide, OSHA 62

no

no

no

no no no

-

60 -480 L 1 Limin

24 L 0.1 Llmin

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction ofsample into small pieces Cooling of sample -

GC, FPD

HPLC, UV

Analysis method (AAS, XRF, MS. GC, LC)

no no no

3030 L 0.1 - 1 L/min

gdS

gdS

phenol, cresol. OSHA 32

research. industry, environmental protection, occupational safety gas amines and alcohols (NIOSH 5217,S233 andothers) GC, FID

research, industry, environmental protection, occupational safety

Which phases are investigated Which components are investigated

ca. l o g

ca. 25 g

ca. l o g

research, industry, environmental protection, occupational safety

6/4/70 mm 150/75 mg

13/11~50mm 270: 140 mg

6/4/70 mm 100:50 m_g

Area of application

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

Dimensions (width. height. depth) Sample space size (width, weight, depth) Weight (mass)

Orbo-52 (small) SUPELCO SUPELCO 50 pieces 166.44 DM

SUPELCO

Orbo-49 P SUPELCO SUPELCO 10 pieces 360 DM

-

Name of apparatus Model Manufacturer Distributor Price incl. tax

SUPELCO

Orbo-47 SUPELCO SUPELCO 50 pieces 191.52 D M

SUPELCO

(continued)

Company

Tab. 3.

3 s

3

3

Q

N

m

glass, glass fiber filter, polyurethane foam, SUPELPAK 20 P

glass, silanized glass wool, SUPELPAK 70

Materials in contact with sample

excessive flow rate and/or collection time -

Yes no breaking open at sampling location Orbo-tube-cutter

excessive flow rate and/or sample collection time -

Yes no opening of the closure cap at the sampling location elution with toluene yes (after elution)

excessive flow rate and/or sample collection time

-

Yes no

breaking open at sampling location

Orbo-tube-cutter

elution of both beds (separately) with methanol

yes (after elution)

Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

Maintenance accessories

Method of sample taking

Possibility of taking an aliquot sample

elution of both beds (separately) with diluted H,SO, in 10% McOH yes (after elution)

-

high humidity

none known

-

glass, silanized glass wool, silica gel (20/40 mesh), urethane foam

anyplace 3 rnin - 50 h

none known

Sample space sealing material Contamination risks during sampling Danger of partial sample loss

anyplace 1-8h

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collection beds)

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collection beds)

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collection beds)

anyplace 4h

-

17 days

15 days

Sampling location Length of sampling time

Redundancy of sampling procedure (Precaution against sample loss)

Possibility of sample storage in container Possibility of automatic sampling Sampling

G%

$ 9

2

2

h

%

2

9 3

-

ca. 15 g -

yes Pump

-

yes Pump

1-50L 0.05-0.2 L/min -

gas anorganic acid fumes HPLC, IC 3- I00 L (NIOSH 7903) 0.2-0.5 L/min -

no no no

gas amines GC, FID

3-30 L 0.01 - 1 L/min -

no no conly conditionally

Which phases are investigated Which components are investigated Analysis method (AAS, XRF, MS, GC, LC)

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces Cooling of sample

no no no

research. industry, environmental protection, occupational safety gas PCBs. NIOSH 5503 GC, ECD

research, industry. environmental protection, occupational safety

research, industry, environmental protection, occupational safety

yes Pump

ca. l o g

Area of application

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

6/4/70 mm l00/50 mg

7/5/100 mm 400/200 mg

8/6/70 mm 150: 150 mg

Dimensions (width, height, depth) Sample space size (width, weight, depth) Weight (mass)

-

Orbo-60 SUPELCO SUPELCO 50 pieces 200.64 DM

Small sample tube Orbo-53 SUPELCO SUPELCO 50 pieces 237.12 DM

Orbo-52 (large) SUPELCO SUPELCO 50 pieces 196.08 DM

Name of apparatus Model Manufacturer Distributor Price incl. tax

SUPELCO

SUPELCO

SUPELCO

(continued)

Company

Tab. 3.

3

Q

3

P

a.

Yes no

Orbo- tube-cutter elution of both beds (separately) with 3 mM NaHCO,,,, 4 mM Na,CO, yes (after elution)

Yes no breaking open at sampling location

Orbo-tube-cutter

elution of both beds (separately) with diluted H,S04 in 10% MeOH

yes (after elution)

Sample lossitime unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

Maintenance accessories

Method of sample taking

Possibility of taking an aliquot sample

breaking open at sampling location

-

excessive flow rate and/or sample collection time -

excessive flow rate and/or collection time -

-

Sample space sealing material Contamination risks during sampling Danger of partial sample loss

glas, glass fiber filter, urethane foam, activated silica gel (20/40 mesh) none known

glass, sikdnized glass wool, silica gel (20/40 mesh), urethane foam

Materials in contact with sample

anyplace 6 - 500 min

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collection beds)

Yes

high-humidit y

anyplace 3 min-50 h

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collection beds)

Sampling location Length of sampling time

Redundancy of sampling procedure (Precaution against sample loss)

Possibility of sample storage in container Possibility of automatic sampling Sampling

yes (after elution)

elution o f both beds (separately) with hexane

breaking open at sampling location Orbo-tube-cutter

-

DDT, DDE, S-containing compounds excessive flow rate and/or sample collection time

glass, silanized glass wool, urethane foam

anyplace 5 min- 16 h 40 min

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collection beds)

2 months

ss

3

2

91

2

$

5:

ze.

Q

m,

no no

no

no

no

no

I-12L 0.01 -0.2 Limin

1-24 L 0.05-0.15 L/min

50- 100 L 1 L/min

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces

GC, ECD

Analysis method (AAS, XRF. MS. GC, LC)

gas ethyleneoxide

HPLC

gas acidic fumes, (HCI, NH,, Ac-acid, formic acid)

Which phases are investigated Which components are investigated

research, industry, environmental protection, occupational safety gas 2-butanone (methyl ethyl ketone) NIOSH 2500 GC. FID

research, industry, environmental protection, occupational safety

research, industry. environmental protection, occupational safety

Area of application

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

ca. l o g

ca. 20g

ca. l o g

6/4/70 mm 160iXO mg

6j4:lOO mm 400,'200 mg

8/6!100 mm 335 mg/165 mg

Dimensions (width, height, depth) Sample space size (width, weight, depth) Weight (mass)

Orbo-90 SUPELCO SUPELCO 25 pieces 184.68 DM

Orbo-78 SUPELCO SUPELCO 25 pieces 184.68 DM

Small sampling tube Orbo-70 SUPELCO SUPELCO 50 pieces 244 DM

Name of apparatus Model Manufdcturer Distributor Price incl. tax

SUPELCO

SUPELCO

SUPELCO

(continued)

Company

Tab. 3.

a

0 .

3\

glass, sikdnized glass wool, urethane foam, Carboxen 564 acetone, isopropanol

excessive water flow and/or collection time -

glass, silanized glass wool, HBr impregnated Carboxen 564 2-bromethdnol

excessive water flow and/or collection time -

Yes no breaking open at sampling location

glass, silanized glass wool, urethane foam, 5.0% Na,CO, on Chromosorb P (20/40 mesh)

excessive water flow and/or collection time -

Yes no

Materials in contact with sample

elution of both beds (separately) with D M F yes (after elution)

Orbo-tube-cutter

desorption of both beds (separately) with distilled water yes (after elution)

Maintenance accessories

Method of sample taking

Possibility of taking an aliquot sample

Orbo-tube-cutter

breaking open at sampling location

Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

Sample space sealing material Contamination risks during sampling Danger of partial sample loss

none known

anyplace 5 min-20 h

anyplace 7min-8 h

50- 100 min

Sampling location Length of sampling time

Redundancy of sampling procedure (Precaution against sample loss)

elution of both beds (separately) with CS2 yes (after elution)

yes no breaking open at sampling location Orbo-tube-cutter

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (collection beds)

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (collection beds)

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collection beds)

no 6 weeks at 25 “C

no 17 days in darkness

no yes, several days

Cooling of sample Possibility of sample storage in container Possibility of automatic sampling Sampling

B

$

%

k

fi

7

i

3 L (OSHA 69) SO mL/min (OSHA 69)

Sample size (mass or volume) Output capacity Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces Cooling of sample

no no no

gas acetone and other volatile hydrocarbons GC. FID

Which phases are investigated Which components are investigated

Analysis method (AAS, XRF, MS. GC, LC)

research, industry, environmental protection, occupational safety

Area of application

no

no no

-

no no no

varies varies -

24 L 0.1 Limin

GC. FID

vinylacetate, OSHA 51

gas

research, industry, environmental protection, occupational safety gas many types of organic compounds MS. GC, LC

research, industry. environmental protection. occupational safety

ca. 1Sg

ca. l o g

ca. l o g

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

7/S,'I 00 mm 350/175 g

6i4,'100 mm 160/80mg

6/4/70 mm 130/6S mg

Dimensions (width. height. depth) Sample space size (width, weight. depth) Weight (mass)

-

Orbo- 100 SUPELCO SUPELCO 2.5 pieces 29 1.84 DM

-

Orbo-92 SUPELCO SUPELCO 25 pieces 202.92 DM

Small sample tube Orbo-9 1 SUPELCO SUPELCO 25 pieces 196.10 DM

Name of apparatus Model Manufacturer Distributor Price incl. tax

SUPELCO

SUPELCO

SUPELCO

(continued)

Company

Tab. 3.

a

00

m

-

acids, bases, free radical react with vinyl acetate excessive water flow and/or collection time -

-

other adsorbable analytes

Sample space sealing material Contamination risks during sampling Danger of partial sample loss

desorption of both beds (separately) with 1% DMF in CS2 yes (after elution)

Method of sample taking

Possibility of taking an aliquot sample

Orbo-tube-cutter

Orbo-tube-cutter

Maintenance accessories

elution of both beds (separately) varies with 9/5 methylene chloride/methanol yes (after elution) yes (after elution)

breaking open at sampling location Orbo-tube-cutter

-

-

breaking open at sampling location

Yes no

Yes no

-

Yes no

breaking open at sampling location

-

Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

excessive water flow and/or collection time -

glass, silanized glass wool, Carbotrap (20/40 mesh) various

glass, silanized glass wool, HBr urethane foam, Carboxen 564

glass, silanized glass wool, Carbonsiere S 111

Materials in contact with sample

excessive water flow and/or collection time

anyplace varies

anyplace 4h

anyplace 60 min (OSHA 69)

Sampling location Length of sampling time

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collection beds)

yes (depends on pump used) continuous and discontinuous (depends on pump used) yes (2 collection beds)

yes (depends on pump used)

yes (through 2 separate packing beds)

varies

18 days at 17 C

24 h

Redundancy of sampling procedure (Precaution against sample loss)

Possibility of sample storage in container Possibility of automatic sampling Sampling

W m

P

gT

(continued)

Sample size (mass or volume) Output capacity

-

-

-

gas 2.3 dibromo propanol NIOSH (method not yet made public)

Which phases are investigated Which components are investigated

Analysis method (AAS. XRF, MS, GC, LC)

research, industry, environmental protection, occupational safety

Yes Pump

-

ca. 500 L 1 - 5 L,mh

GC, ECD

gas PCBs, pestizides, (ASTM D4 861-88)

research. industry, environmental protection, occupational safety

ca. l o g

ca. 10g -

6/25/1250 mm 76 x 22 mm diameter

6/4/70 mm 100,'50 mg

-

Orbo- 1000 SUPELCO SUPELCO 50 pieces 164.16 D M

-

SUPELCO

Orbo- I0 1 SUPELCO SUPELCO 25 pieces 207.48 DM

SUPELCO

Area of application

Operation of apparatus Power source (voltage) Transportable Necessary accessories (e.g. pump, etc.)

Dimensions (width. height, depth) Sample space size (width, weight, depth) Weight (mass)

Name of apparatus Model Manufacturer Distributor Price incl. tax

Company

Tab. 3.

~~

5 mL- 3 L:min 5 mL-3 Limin

research, industry, environmental protection, occupational safety gas hydrocarbons. CFC, HUW, PCB, acids PAM, amines, alcohols, acrolein, FA. pesticides, ethyleneoxide, phenols. cresols. aromatic substances MS. GC, HPLC

~

electrical battery (storage battery) Yes included in price: holder for small sample bottle, holder for filter cassettes, charging devices 220 V

ca. 1 kg

120/30,'70mm -

Sampling pump PAS 3000 SUPELCO S U PELCO 2844.30DM

SUPELCO

3

3

2

2

a

0

-4

-

Sample space sealing material Contamination risks during sampling Danger of partial sample loss

not known yes (after elution)

Method of sample taking

Possibility of taking an aliquot sample

~~

eluation with 5% diethyl ether in hexane yes (after elution)

Orbo-tube-cutter

Maintenance accessories

~

yes, depends on requirements yes, depends on requirements rinsing, Soxhlet extractable opening thecap at sampling location

Yes no breaking open at sampling location

Sample loss/time unit (e.g. via evaporation, diffusion) Apparatus is disposable reusable Cleaning procedure Maintenance procedure

-

excessive flow rate and/or sample collection time excessive flow rate and/or sample collection time

excessive flow rate and/or collection time

none known

glass, polyurethane foam

glass, silanized glass wool, carbotrap (20140 mesh)

Materials in contact with sample

none known

anyplace depends on the method

anyplace 100- 500 min

anyplace not known

Sampling location Length of sampling time

-

depends on method used

no Yes charging from electric outlet, calibrating charging device, flow rate measuring device

depends on method and adsorber excessive flow rate and/or sample collection time

-

only sampling test tube, sample does not enter the pump

yes (depends on collector used)

yes (2 collection beds)

yes (depends on type of analysis) yes (depends on pump used) discontinuous

30 days

-

Redundancy of sampling procedure (Precaution against sample loss)

no

no

no yes (depends on pump used) continuous and discontinuous (depends on pump used) no

no no

no no

no no

-

yes (depends on pump used) -

Provision for separation of other phases (Particles, drops, gas) Homogenization of sample Reduction of sample into small pieces Cooling of sample Possibility of sample storage in container Possibility of automatic sampling Sampling

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

3 Trace Elements Neec Trace Analysis Istvan Pais

3.1 Problems of Trace Element Analysis 3.1.1 Introduction “Trace elements” are present in living organisms only in small trace amounts, i.e., their concentration is not exceeding 1 pg/g. According to the literature, the concentration of some trace elements is 2 1 ng/g in the tissues of plants, animals, and humans. This concentration ratio requires the use of sensitive analytical methods and accurate analytical applications. In the 20th century, the sensitivity and biological application of analytical methods have brought forth a few revolutionary changes (Jones, 1987). In the 1950s, e.g. the atomic absorption method utilizing flame-ionization was introduced and the accuracy of this method was just a few pg/g (Willis, 1970). The main disadvantage of this method was that it is time-consuming: in the analysis of different elements the lamp must be changed. The accuracy of this method for the most common trace elements, i.e., iron, copper, zinc, and manganese, was relatively sufficient but in spite of that more sensitive methods were required. In the late 1970s, the ICP method was available (Barnes, 1985) and offered several advantages. For most metals the analytical sensitivity was about 1 - 2 ng/g and, using a polychromator, it was possible to identify 20- 30 elements in the same sample with high accuracy. Today, this method is used in various laboratories, because of its speed and wideranging applications for biological analysis. The disadvantages of ICP are: The elements required can only be determined in clean solutions. The biological samples should pass a preliminary wet digestion with various digestive mixtures and techniques (Zunk, 1990).The preliminary ashing methods have some disadvantages too: At low temperatures (below +400 “C)carbon particles remain in the ash, while at higher temperatures (over +500 “C) some elements may disappear. In analytical chemistry, various methods with different sensitivities or detecting limits are used. From a review-paper by J. B. Jones (Jones, 1987) the values for 26 elements are presented in Tab. 1; these are the most important either due to physiological essentiality or to environmental contamination. No general rule exists for the selection of the most appropriate analytical method, because there are various advantages and disadvantages according to the elements to be identified and to the biological environment in which the given element exists (Behne, 1992; Cornelis, 1992a; Fardy and Warner, 1992; Kwiatek et al., 1992; Sanchez et al., 1988; Schramel, 1993; Tolg and Garten, 1985; Valkovic, 1989; Versieck et al., 1987).

74 Tab. 1.

1. Puis

Sensitivities of Different Analytical Methods in pg/g (after Joncs, 1987)

Symbol

AES

NAA

MS

AAS

Chem

XRF

Al As Ba Be Bi

0.5 10 0.1 0.05 0.2 2 0.1

0.004 0.005 0.02

0.002 0.0006 0.002 0.00008 0.002 0.007 0.0005 0.000 5 0.002 0.0002 0.000 5 0.007 0.00 1 0.00006 0.0004 0.003 0.002 0.003 0.002 0.002 0.002 0.004 0.02 0.004 0.0004 0.002

0.8 0.8 0.2 0.02 0.22 0.01 0.07 0.06 0.04

0.0005 0.0 1 0.1 0.008 0.6 0.003 0.003 0.007 0.002 0.05 0.2 0.005 0.0 1 0.003 0.005 0.01 0.004 0.006 0.004 0.2 0.1 0.06 0.4 0.02 0.01 0.1

17

Cd CO

Cr

cu F Fe

HI: I Li Mn Mo Ni Pb Sb Sc Si Sn Te Ti V Zn

1

0.1 I00 1 2

0.005 0.0 1 0.3 0.002 2 0.003 0.002

0.1

0. I 1 0.2 0.02 2 I00 0.5 0.5 50

2 0.5 5

0.00 I 0.1 0.7 0.5 0.007 0.01 0.03 0.03 0.002 0.1

0.06 2.2 0.02 0.02 0.33 0.07 0.1 0.3 0.5 2 1 0.26 0.03 0.9 0.009

0.4 1

2 2.5 0.2 0.7 0.4 1 0.3

0.2 1 0.2 1.5 0.2 2

0.15 0.2

The abovementioncd processcs may be the source of severe contamination and other analytical pitfalls (Caroli, 1992). As is shown in other chapters of this book, the method of sampling, the cleaning of natural biological samples, and the different contamination risks during the whole process cause great uncertainty in the final analytical results. For the ng/g ratio or in lower concentrations, the error rate can reach 100%. The terms utilized in analytical chemistry have also changed. In the 1970s, the term “trace” was equivalent to pg/g and “ultra-trace” was registered as ng/g. Now, and in the near future, “trace” will mean ng/g and “ultra-trace” will mean pg/g or less.

3.1.2 Problems of Biological Analysis The food chain or, better statcd, the “nutritional network” begins with geochemical sources and reaches the biosphere through the transmitting media, including the final station: man.

Truce Element5 Need Truce Anulysis

15

In the second part of this chapter, we will present the analytical constitution of sea water (see Tab. 6), which is the largest source of food for man, and therefore analytical data are important for trace-element science. The analysis of sea water is linked to various problems which must be solved in order to obtain correct results. According to a recent paper (McLaren et al., 1993), the use of the relatively easy method of ICP is problematic in the case of sea water with its extremely high salt concentration. Thus, a t least a ten-fold dilution is necessary to eliminate deposition at the torch and to avoid severe suppressions of salt. This unavoidable dilution results in extremely low concentrations of some element and thus accurate identification is impossible. Today’s analytical techniques provide the possibility of determining single elements in the pg/g scale, but the analytical result has no biological or physiological value. We should know, e.g., the compound form in which the given element exists, the oxidation state and the relative concentrations of 10- 12 other elements in the same system. With this information, we can begin to evaluate the role of the given element in the given biological system. Biological trace element research is probably the most multidisciplinary research field. As Fig. 4 indicates, the cooperation of 10- 12 different scientific fields is required. To illustrate this problem, we quote (after Iyengar, 1991) one of Mertz’s statements: “An analytical chemist should be more than a procurer of data and a life-scientist more than their interpreter.” Bioavailability is also largely a question of scientific evaluation. We can measure, e.g., a high amount of a given element in a food-sample (Cornelis, 1992b), but the digestive tract of an animal or human may only absorb a small percentage of it or, in other cases, the bioavailability may be near to 100%. This is very important for the determination of positive or toxic effects on human health. The resulting problems are contamination of laboratory instruments (Marshall et al., 1991), environment (e.g., laboratory air; Caroli, 1992), and materials used (Moyer et al., 1991). As will be shown, any “indicator organ” gives a relatively acceptable picture of the trace element supplementation of the whole organism, but in most cases - especially during illness - some organs contain higher or lower values than usual. In this context we should analyze various organs, which is possible in plant and animal experiments, but very limited in the case of humans. The so-called autopsy analysis (Lyon and Fell, 1992) can assist in some cases, but not in all. In these situations we must consider whether some kind of contamination is caused by sampling and whether the sample is really representative of the whole organ. As demonstrated by various authors we can find 2 - 4 or more times divergent results in different parts of the same organ (liver, kidney, brain, bones, etc.). In conclusion, we should in our approache always be like Boyle’s “sceptical chemist”, reviewing our analytical methods and the evaluation of analytical data.

16

1. Puis

3.1.3 Indicator Organs in Biological Evaluation Due to the fact that it is difficult to produce analytical samples from live human and (some) animal organs (excluding laboratory circumstances), analytical chemists have tried to identify certain “indicator organs” which can be collected without any damage to the living organism. These analytical data are well correlated with the supply of trace elements and represent a sound basis for the estimation of trace element supplementation. One of the oldest indicator organs is the deck hair of animals, which is accepted as a biological index of geographical region and of environmental pollution as well. In the case of humans, the same is valid for scalp hair (Moro et al., 1992; Robertson, 1987; Zhuang et al., 1991). Trace metal storage in hair has its own time-scale and biodynamics, not comparable to any other storage or excretory process known in human physiology. It was concluded (Gibson, 1984) that during sampling and washing of hair samples by standardized procedures, the effects of any adventitious trace element contamination are small and can be effectively ignored. Takagi et al. (1986) demonstrated that equally treated hair samples (washed and prepared with acetone, three times with deionized water, again with acetone and dried) from 5 countries showed great differences in their trace element concentrations. Copper was much higher in the USA and Canada than in Poland, India and Japan. Lead, arsenic and mercury were high in India; selenium was lowest in Poland, 10-50 times lower than in Japan, the USA, and Canada. According to recent literature (Sturaro et al., 1993a, b) dynamic ion-exchange chromatography is a convenient method for simultaneous identification of different metals in hair samples. Another important indicator organ is the toenail (Wilhelm et al., 1991). The authors concluded that in most cases toenail analysis is not as useful as hair analysis. As recently reported (Ovaskainen et al., 1993), the selenium concentration in toenails is proposed to be a long-term indicator of the human selenium status. The collection and storage of toenails is very easy and according to the cited author “selenium intake contributed significantly to tonail selenium concentrations”. In addition to the advantages mentioned, it should be noted that these results generally provide a picture of about 1 % of the total content of the given trace element in the living organism. In the literature we find the following data (Lyon and Fell, 1992): an adult (70 kg) contains about 80 mg of copper, the copper component in the serum is only about 3 mg. In the whole body there is an estimated 1.5 g of zinc, but the zinc component in the serum is only about 2.5 mg: this means that the serum contains only 0.2% of the total body content of zinc. For humans and domestic animals, the most important indicator organ is blood: usually serum is used, in some cases (e.g., in lead determination) whole-blood samples are analyzed. Heparin or another anti-coagulant compound is used for preservation of serum samples. As reflected in recent literature (Minoris et al., 1992; Caroli, 1992) heparin contains (pg/g) 2- 12 Ba, 0.6 Cu, 3.6 Mn, and 28 Zn. Conse-

Truce Elements Need Trace Analysis

77

Tab. 2. Trace-Element Contents of some Tissues Symbol

Blood-serum PglL

*

Toenails **

Scalp-hair *

w/g 0.1 -0.2 800-1100 800 - 1200 0.5- 1.0 75-120 800 - 1100

Cr Cu Fe Mn Se Zn

* **

= =

0.3 -0.8 15-25 30 - 60 0.5 - 1.5 0.5- 1.0 150-250

7.5 0.46 129

after Iyengar, 1991 after Wilhelm et al., 1991 and Oveskainen et al., 1993

quently, we should calculate with these values. Because the different heparins do not have the same purity, they have to be analyzed and corrections by evaluation are indispensable. Many values for blood sera, and other biological samples, have changed in the last 15 - 20 years. For example, the “normal” manganese value was considered to be 10 - 30 pg/L, but now recent research has demonstrated an average of 0.5 pg/L. Similarly, the chromium value used to be as high as 5 pg/L, today’s literature reports a lower value of 0.1 -0.3 pg Cr/L. The analytical survey of biological samples has to solve a quite new problem: the analytical research at cellular level. According to the most recent literature (Lindh, 1993) the most interesting problem is to screen the distribution of trace elements in individual cells. In immunological research this new knowledge is of outstanding importance. The investigation of the cellular mechanisms requires new techniques and broad cooperative, multidisciplinary coordination among different scientists. The isolation of some metal containing macromolecules from the given single cell is a very complicated task and we should use the finest and most accurate techniques available. We should also keep in mind that the chemical bonds between trace elements and organic molecules is highly correlated with bioavailability and with biochemical activity in different organs, because of the exchange of different ligands and other parameters.

3.2 The Importance of Trace Elements 3.2.1 Introduction The concentrations of trace elements in the dry matter of living organisms are not higher than a few pg/g. In spite of this fact they have an equal or even higher importance in physiological processes than elements with concentration ratios of 1 - 10% or more.

78

Tab. 3.

I . Puis

Elements without Significance in the Biosphere

Noble gases

Radioactive elemcnts

Argon Helium Krypton Neon Radon Xenon

Actinium Polonium Protactinium Radium Thorium Uranium

The earth, and according to our recent knowledge, the solar system contain 88 permanent chemical elements. The noble gases (which do not form compounds) and the strong radioactive elements may be excluded (Tab. 3) for they don’t have any positive physiological function, so that 76 elements remain. It is accepted worldwide that seven non-metals and four metals are called macroelements (Tab. 4), because in most organisms their concentrations are higher than 0.1% : usually the quantities of carbon, oxygen, hydrogen, nitrogen, phosphorus and sulfur exceed 96% in the dry matter of plants. The remaining 65 elements are called trace elements and some of them - due to their extremcly low concentrations - ultra-trace elements. I n Fig. 1, these elements are classified into four main groups. (1) and (2): The so-called generally and the partly essential trace elements (Tab. 5). To avoid any misunderstanding, we accentuate that this classification of elements represents the author’s opinion, and is strictly an arbitrary, free approach. (3): The physiologically beneficial elements, a term that was first defined by the author (Pais, 1990a). The beneficial character of some elements has also been demonstrated by the author and his research group. (4): Elements for which no positive physiological role has been demonstrated yet. Regarding the last sentence above, we recall a remark from Schwarz (Schwarz, 1970) as a basis for future development in trace-element science: to demonstrate Tab, 4. Essential Macroelcments Symbol

Nzme

H 0 C N P S

Hydrogen Oxygen Carbon Nitrogen Phosphorus Sulfur Chlorine Potassium Sodium Calcium M agnesiutn

C1

K Na Ca Mg

Truce Elements Need Truce Anulysis

I,

Ill

111,

IV,

v,

VI, VII,

VIII,

I,

11,

111,

Iv,

V,

vi,

19

VII,

vni,

Generally essential essential Physiologically beneficial Physiological role hardly known

I Partly

*

** Fig. 1. Physiological significance of trace elements.

that an element is essential is a very difficult task, but to exclude it categorically from being essential (or promotive) has no real basis. The nutritional network (theoretically and practically) contains all 88 permanent elements. 30 or more of them are the so-called little-known trace elements, and we should calculate with unknown physiological effects which may be the basis of different plant, animal and human diseases. Because of that it is most urgent to collect and accumulate as much knowledge as possible on these elements.

Tab. 5. Generally and Partly Essential Trace Elements Generally essential

Partly essential

Boron Cobalt Copper Iodine Iron Manganese Molybdenum Nickel Zinc

Arsenic Chromium Fluorine Lithium Lead Selenium Silicon Tin Vanadium

80

1. Puis

3.2.2 Criteria of Essentiality and Beneficiality The rigid rules of essentiality were originally provided by Arnon and Stout (Arnon and Stout, 1939).These rules have been reevaluated several times and the most recent version accepted woldwide is the following, according to Frieden (Frieden, 1985):

( I ) The organism can neither grow or complete its life cycle without the element. (2) The element should have a direct influence on the organism and be involved in its metabolism. (3) The effects of the essential element cannot wholly be replaced by those of any other element. (4) The element is present in tissues of different living organisms in comparable concentrations. ( 5 ) Its withdrawal produces similar physiological or structural abnormalities, regardless of species. (6) Its presence reverses or prevents these abnormalities. (7) These abnormalities are accompanied by specific biochemical changes that can be remedied or prevented when the deficiency is checked. In earlier trace element literature the author found that some scientists preferred the oversimplified classification of “essential” vs. “non-essential” groups. In order to establish a more multifaceted picture, in 1990 he created the following criteria of beneficiality: ( 1 ) By application of the given element, we can recognize promotive roles in plant growth, increase in weight of domesticated animals, or human health conditions. (2) After application of the given element, the activity of one or more enzymes with promotive physiological effects increases. (3) By application of the given element, plants, animals and humans demonstrate improved health conditions or better immunological responses to diseases. (4) By application of the given element, the toxic effects of other elements are decreased or otherwise diminished (“antidotum effect”). ( 5 ) By application of the given element, the intake and utilization of other essential elements can be increased. (6) By application of the given element, some physiological parameters show an advantageous change, c.g., the cholesterol level in blood will be lower.

According to these criteria, titanium, gallium and zirconium - previously regarded as physiologically useless elements - were demonstrated to be beneficial by the author and his research team (Kiss et al., 1985; Novak-Fodor et al., 1992; Novak et al., 1992; Pais, 1983, 1988; Pais et al., 1989; Simon et al., 1989). Horowitz (1988) demonstrated that scandium may have beneficial activity in some physiological processes. According to Liu and other scientists (Liu, 1988; Wu et al., 1985) some rare carth metals also demonstrate beneficial properties in the life of plants, and certain types of fertilizers containing them are widely used in Chinese agriculture.

Trace Elements Need Trace Analysis

81

3.2.3 Criteria of Toxicity The problem of toxicity is also a multifaceted question, and providing a correct answer is not an easy task. Some centuries ago, Paracelsus (Theophratus Bombastus von Hohenheim, 1493 - 1541) provided the main basis on which materials should be considered to be toxic. Based on this traditional opinion, we would like to make an apparently unusual and surprising statement: there is no material (elements and compounds) which is eo ips0 toxic, and yet all materials are toxic! The criterion is solely the concentration or quantity of the substance. We would like to illustrate this situation with Fig. 2, which is well-known in trace element literature. As has been clearly demonstrated, all substances can cause deficiences which appear as deficiency diseases (lethal in some cases), and, on the other hand, in higher concentrations the same element (or compound) causes severe toxicity or may be lethal. We will now try to classify the most important parameters which are meaningful in deciding what is toxic and when: (1) The first parameter is the compound form. Living organisms can rarely take up chemical elements in elemental status, but mostly as compounds. Some criteria for the toxicity of a form are the solubility in water or weakly acid media, or the formation of chelates that can be taken up into the digestive system. Some forms may be quite toxic, while the others usually are not harmful, as the following commonly accepted examples illustrate. Mercury chloride and even more methyl mercury is - between given concentration ratios - highly toxic, but HgS, because it is insoluble, does not represent any danger to living organisms. The soluble

L No growth

Mild Severe deficiency deficiency

Optimum range

Luxury Toxicity consumption

Fig. 2. Effects of trace element concentrations on living organisms.

Concentrat ion Lethality

82

1. Pais

barium compounds are usually very toxic, but BaSO, shows no real toxicity: this compound is used as a contrast-medium in X-ray diagnosis of the digestive tract. (2) Oxidation state. The international literature has provided many arguments that trivalent chromium demonstrates essentiality in most cases and has a very low toxicity, but the hexavalent form, i.e. chromates are usually highly toxic under common conditions. Nevertheless in daily praxis during uptake, the reduction from Cr(V1) to Cr(l1l) mostly proceeds very rapidly and, therefore, toxicity is not serious. The same situation is valid for As(II1) and As(V): the latter generally expresses greater toxic effects than the former. Here it is also important to consider the parameter of the organic compound form as the source of different uptake properties. ( 3 ) Presence or absence of other (synergetic or antagonistic) elements. A very illustrative example is the toxicity of cadmium in smoking. According to recent literature (Bell et al., 1992), one cigarette may contain 1.0 pg, of easily absorbable cadmium. This means that an average smoker has a cadmium uptake of 20- 30 pg/day which is the threshold of strong toxicity. Provided that the zinc and calcium levels in his organism are sufficient, this quantity of cadmium is not harmful. In case of insufficient zinc and calcium levels toxicity increases exponentially. Wilson’s disease is known from literature (Weisner et al., 1987) as an accumulation of copper in cells of the brain. With an increased zinc uptake in daily nutrition, the copper level decreases, but this way of treatment is not the best. Chelate therapy with a simple tripeptide is more appropriate. (4) Mode and manner of intake. Intravenous injection mostly is more dangerous than digestion via the intestinal system: it is known from literature that some elements demonstrate an excellent (nearly 100%) absorption rate from common foods, while others show a smaller percentage. These parameters vary from person to person, from food to food, etc. Therefore it is completely impossible to create a generally valid regulation of element uptake. These uptake processes, i.e., bioavailabilities are in clear correlation with the pH value and the exchange possibilities of ligands. ( 5 ) Duration of action. Usually a single (common) dose of the toxic compound is not very dangerous, but continuous uptake over an extended period of time, is much more harmful. It is quite understandable that the combination of timing and dose of the compound is so important. (6) Defense mechanisms of living systems. In the context of this chapter it is neither possible to deal in detail with metallothioneins (Kiigi and Schaffer, 1988) in human and animal organisms, nor with the role of newly found phytochelatins (Rauser, 1990) in plants, although these factors can also mitigate or diminish toxic effects. The activity of the immune system of different individuals under the influence of some microorganisms or viruses will not be discussed as well. The author does not contend that the given picture of toxicity is wholly complete, but to qualify a compound to be harmful requires the establishment of parameters under which the term toxicity may be judged as valid or not. Concluding our discussion of toxicity, we find it necessary to provide a general overview of what may constitute “too little” or “too much”, and when.

Trace Elements Need Truce Analysis

83

3.2.4 Changes in Element Concentrations In Tab. 6, the chemical composition of earth’s crust, sea water, and some land plants and animals (Pais, 1991) is provided. These data are taken from various sources (e.g., Adriano, 1986; Bowen, 1966) and therefore the figures presented are not absolutely exact. The most frequent and yet physiologically important elements were compiled. As can be seen from these data, the chemical composition of inorganic nature differs much from organic nature. The main differences are: (1) The quantity of hydrogen is 40-60 times higher in the organic than in the inorganic world. The ratio for nitrogen is about 700-2500 times higher in living organisms than in the earth’s crust. A similar difference can be seen in the quantity of carbon: the enrichment factor is > 1000. For most metals with concentrations in the earth’s crust 21% the ratio is decreased by a factor of 5 to 1000 in the organic world. On the other hand, some essential elements demonstrate enrichment in organic material. Tab. 6. Different Element Concentrations Name

Earth w. ratio

Sea water mg/L

Plants

Animals mg/kg

Oxygen Silicon Aluminium Iron Calcium Sodium Potassium Magnesium Titanium Hydrogen Phosphorus Manganese Sulfur Carbon Chlorine Fluorine Chromium Vanadium Zinc Nickel Copper Lithium Nitrogen Cobalt Molybdenum Boron Iodine Selenium

466000 277200 81 300 50000 36 300 28 300 25 900 20 900 4400 1400 1180 1000 520 320 314 300 200 150

132 80 70 65 40 23 15 3.0 0.3 0.09

857000 3.0 0.01 0.01 400 10 500 380 1350 0.001 108000 0.07 0.002 885 28 19000 1.3 0.00005 0.002 0.01 0.0054 0.003 0.18 0.5 0.0003 0.01 4.6 0.06 0.00009

410000 < 5000 < 500 140 I 8 000 1200 14000 3 200 < 1.0 55000 2 300 120 3 400 454000 < 2000 < 40 0.6 1.6 100

3.0 14.0 0.1 30000 <0.5 0.9 50 0.4 0.2

400000 < 6000 < 100 160 <85000 4000 7 400 1000 0.2 70000 < 40000 0.2 5000 465 000 2 800 < 800 0.07 0.1 250 0.8 2.4 0.02 100000 0.03 0.2 0.5 0.4 1.7

84

I. Pnis

(2) The selection in the food chain produced even more surprising samples. Sea water, as demonstrated in Tab. 6, is relatively poor in chemical elements, apart from sodium and chlorine. Some mussels and oysters accumulate a selected group of elements, e.g., chromium, iron, cadmium, mercury, zinc with a factor of 10000 to more than 100000. This is also valid for other sea animals which may be good sources of essential elements and, at the same time, dangerous sources of toxic heavy metals as well. It is well-known from international literature that some Ascidian species contain 0.1 -0.2% of vanadium in their blood, with concentrations as high as 1.5% also possible: i.e., the accumulation rate in this example is close to ten million. In some relation, this is similar to the thyroid gland containing over 80% of the iodine i n the human organism. The abovementioned enrichments were natural processes over a long time during biological evolution, and today the consequences of developed, genetically determined element enrichments also occur. These change usually do not correlate with toxicity. ( 3 ) Hot770 sopiens as a part of nature has initiated more and more changes of the original equilibria of nature. These changes were not very significant before industrialization, and therefore the beginning of the era of anthropogenic changes can be dated from the French Revolution. With the progress of chemical industry, the use of natural energy sources and mechanization have increased rapidly. In this century, and especially in the last two or three decades, several hundred thousand tons of “new” compounds have come into the environment which were formerly unknown. So the time for living beings to adapt to them was too short and consequently, these pollutants became the sources of highly dangerous contaminations. These toxic contaminations have not always had a direct impact, but more often indirect effects. The equilibrium in the food chain is especially vulnerable, as illustrated by the examples provided. In conclusion, these contaminations from anthropogenic sources should be minimized, otherwise the former tolerable equilibria will change in a manner which will be destructive to living systems. Therefore, the direct of indirect toxic effects of trace elements are of great importance in the nutritional network and finally in the future of our world.

3.2.5 The Importance of Trace Elements in the Environment The importance of trace elements is similar to that of chemical bonds. Chemical bonds of thc first order (ionic, covalent and metallic) have a much higher energy of valencies (200- 800 kJ/M), while those of the second order (hydrogen bonds, Van der Waals forces) havc a much lower (0.1 to 40 kJ/M) energy level. These weaker bonding varicties are much more important to living systems because the three-dimensional structures of macromolecules (carbohydrates, proteins, lipids and nucleic acids) are caused by these weaker forces. In biochemical processes this structure geometry is essential for the more important of these processes.

Trace Elements Need Trace Analysis

85

The appropriate level of trace elements, i.e., a good supply of the living organism, is a cardinal question of life. It is a well-known fact that the enzymes and enzyme-systems control the physiological and biochemical processes of the organism and that these enzymes contain trace elements as inbuilt parts of their structure, or that they utilize them as activator agents for their biochemical function. In other words: the “optimum” concentration of trace elements is the basis for the health of living beings, including humans. Before the modern industrialized society existed, the various geographical regions of the world used to have specific concentrations of these elements and therefore, living being had been capable to adapt to the original conditions. With the dramatic changes of these equilibria, adaptation is very difficult or requires a much longer period of time. Therefore, living beings - including humans - are faced with unfeasible conditions. The dramatic changes of the original equilibria are the main source of the crucial problems of environmental contamination. As we have mentioned, the supply with trace elements is less than optimal today. In some regions of the world, certain trace elements show minimum levels, yet in other regions the same elements are available in excess levels, as demonstrated by the following examples. Selenium deficiencies are found in some regions (southwest to northeast) of China (this is the basis of the well-known Keshan disease), yet in other areas of China excess selenium levels are reported. The same is valid for the United States. In seleniferous districts the cultivated wheat contains 150- 200 pg/kg selenium, yet in other districts a deficient selenium supply of agricultural plants exists. Selenium deficiencies have been demonstrated in New Zealand and in the whole of Scandinavia as well. The selenium supply in most European countries (e.g. Germany, Poland) is less than optimal. In the author’s country, Hungary, most geographical conditions provide very good possibilities for agricultural production. Yet deficiencies predominate in some districts, while excess supply of various nutritive elements is present in others. In some countries (mostly in the west), for example, there exists iodine deficiency, while in others, mineral-waters are relatively very rich in iodine compounds. In conclusion we can find regions of deficiency and/or toxicity for all elements in our world. This means, in other words, that some regions of the earth require supplementation of certain nutritive elements, or that the problem of toxicity, which can be the basis for some very dangerous diseases, be addressed. The struggle against environmental contamination is accepted worldwide as a general scientific “religion”, but most people, and even an important number of scientists, have not exactly recognized the consequences of pollution, specifically with trace elements, upon the life cycles and the nutritive supplementation of living beings. The pollution can have a geological origin: e.g. the high arsenic content of drinking water in some regions of Taiwan or southeastern Hungary. If this type of problem exists, chemical approaches can be organized (application of ionexchange resins, etc.) to eliminate or diminish toxic effects by providing - as a minimum - drinking water of acceptable quality. The issue is very similar to cases of pollution of anthropogenic origin: we should eliminate the sources of contamina-

86

I . Puis

tion. Moreover, in other cases, technological improvements of industrial processes should be found, as well as new materials which avoid polluting the food chain with toxic chemical compounds. On the other hand, if we find a situation of deficiency in a certain area of the world, we should try to provide support of the given element in the nutritional network of living beings, applying additives containing trace elements in animal nutrition and pills or roborative drinks in human nutrition. This latter solution poses different problems, mostly because some trace elements show very narrow limits between “daily requirements” and toxic levels. Therefore, all supplementation should be under comprehensive governmental and medical supervision. In some cases, this supplementation is quite simple: in the case of the aforementioned iodine deficiency, cooking salt has been enriched with the appropriate quantity of sodium iodide and thus iodine deficiency could be eliminated from the daily lives of several hundred million people. The author is convinced that these problems of mankind should be addressed as soon as possible, in spite of the fact that the way to a solution is a precarious one. The main problem is that any change in the nutritional equilibrium creates new problems. Yet scientists should not be satisfied with the new “habits” of our modern society: new “compositions” produced worldwide which are - according to the available sourccs - effective against all diseases. We should definitively underscore that these new compositions may have some positive effects, but in most cases are ineffcctivc, and in some cases have negative physiological consequences, as well. Conversely, we should accentuate the main task of international science, which is to create better nutritional possibilities for mankind: ( I ) Broaden the list of elements in “Recommended Daily Allowance” tables; (2) create a complete and realistic list of the toxic concentrations for all elements which are (or may become a part of) the food chain in the near future; ( 3 ) the National Health Institutes of all nations should establish supplementation values, and if deficiencies exist, should take action to provide the required supplementation with nutritive elements in daily diets; (4) scientists from various countries should define concentration limits in blood serum and other indicator organs, between which a patient’s state of health can be determined; (5) by finishing the survey under the control of the National Health Institutes, a supplementation program could be established t o ensure a more appropriate intake of trace elements by all citizens.

3.2.6 Interactions between Different Elements According to the aforementioned program, we should know the interactions of elements much better than we do today. The physiological effects of various trace elements do not solely depend on the given element, they shows rather multifaceted correlation with other nutritive elements.

Trace Elements Need Trace Analysis

Fig. 3. Interactions of toxic heavy metals and several essential elements (Chowdhury and Chandra, 1987).

//

Se-Hg

87

‘Cr

These interactions between various elements are well-known and, simplified, called synergism and antagonism. In fact these correlations are more complicated than can be expressed by these words. From various sources we find very complicated pictures that do not provide for an easy understanding (Markert, 1988). Fig. 3 shows a relatively simple, but well-understood picture from Chowdhury and Chandra (1987). The multifaceted correlation between the three highly toxic heavy metals Cd, Pb, and Hg and some essential elements is demonstrated. We should mention that if the human organism has a good supply of calcium, zinc, or selenium, cadmium toxicity can be decreased or even prevented. There are a great number of papers which discuss how the application of various essential elements in appropriate concentrations can cause some toxic heavy metals to be forced out from binding places (Berr et al., 1992; Odeh, 1992; Schlesinger et a]., 1992; Zalups and Cherrian, 1992). Application of zinc-enriched baker’s yeast to chickens which were fed a daily formula with a high lead content did not show any negative symptoms concerning increase in weight, and concentrations of lead present in the blood decreased as well (Pais et al., unpublished data). The aforementioned case is of much greater importance. Originally, our environments only rarely contained excessively high levels of pollutants like cadmium, lead, mercury, thallium, and beryllium, but with the increasing pollution of nature we can count on these undesirable consequences. If daily feeds are also deficient in calcium, zinc, selenium, etc., severe toxicity will be even greater. Our environment - as a consequence of industrialization, chemization of agriculture, enormous mechanization, acid rain, and other reasons - contains an increasing number of toxic elements and compounds which were not part of the food chain formerly. Therefore, genetic determination and, as mentioned earlier, the capability of accomodation of our immunological system is not able to effectively fight against these formerly unknown factors. According to our opinion, some so-called civilization diseases have similar physiological origins. Today, more and more attention is given to the peroxidation processes and to the free radicals which are in the background of these “new” biochemical processes. In spite of the fact that many details of these biochemical pathways are far from being solved and brought to light, we should determine the main criteria for healthy nutrition in order to avoid these undesirable consequences of our modern diet. The author is deeply convinced that the issue of health and survival is basically related to daily nutrition: to sufficient supplementation with different nutritive elements, especially trace elements.

88

1. Ptris

3.2.7 The lrnportance of Interdisciplinary Trace Element Research The extremely rapid diversification of science into individual disciplines provides increasingly fewer possibilities for scientists to have sufficient knowledge of and information about the neighbouring fields of their respective scientific areas of expertise. Consequently, the importance of team-work has been increasing for the last few decades and today its importance is greater than ever. The author was among the first scientists to propose and emphasize on the importance of interdisciplinary trace element research (Pais, 197I), and according to this concept he organized the Trace Element Committee of the Hungarian Academy of Sciences. On this committee 9 - 10 scientific fields are represented, and in the proceedings of five international symposia, the advantages of multifaceted traceelement research can be recognized (as editor: Pais, 1985, 1987, 1988b, 1990b, 1992). Recently, the same concept has been supported by Iyengar (1992). Trace element science is really a specific field with a multidisciplinary character. If we look, e.g., at the food chain, which starts from geochemical sources and has three transmitter media: soil, water, and atmosphere. Therefore, the knowledge of a minimum of four scientific fields is required: geochemistry, soil science, water science, and some aspects of physics. Turning our attention from the geosphere to the biosphere, at first there is the plant kingdom which requires the knowledge of plant growers, including the modern aspects of plant physiology and plant biochemistry. Then there is the animal kingdom which requires the knowledge of animal husbandry and veterinary sciences. At the final station of the biosphere there are human beings, and our health requires the complete realm of medical science (see Fig. 4). The geosphere and the biosphere definitely require the knowledge and findings of analytical chemistry, which were in some important correlations discussed in the Dlfferent parto of the food-chaln

The requlred rclencer

‘I

Human beings

Animals

Bionphere

t

Plants

t

t

Soil

t

plant cultivation analytical chemistry meteorology physics

Atmosphere

Water

coordination chemistry medical sciences biochemistry animal breeding veterinary sciences

Tranrm1tt.r medla

water sciences soil science

Earth’s crust Origin of trace elements geochemistry

Fig. 4.

lntcrdisciplinary concept trace elemcnt research.

of

Trace Elements Need Trace Analysis

X9

first part of this chapter. Correct and sensitive chemical analyses of different parts of the geo- and the biosphere will form the original basis for all future extensions of knowledge and their consequences. If we know the backgrounds of living processes, we should apply the knowledge of modern up-to-date biochemistry and, because trace elements are fundamental parts of enzymes and enzyme systems, we should apply the knowledge of coordination chemistry (especially in connection with bio-inorganic chemistry). If we examine the nutritional processes of plants, animal and human beings, the bioavailability of different nutritive elements and digestive processes, we will find that these are also based on changes in chelate binding and constitution, as well. In conclusion, trace element science requires the composite knowledge of 10 different scientific fields cooperating in order to fully understand the issues arising from the whole of trace element science. We are convinced that this type of collaboration and worldwide interdisciplinary cooperation are the most important bases to solve the multifaceted problems of the environment and to ensure the future of mankind.

3.3 References Adriano, D. C. (1986) Trace elements in the terrestrial environment. New York: Springer Verlag. Arnon, D. L., Stout, P. R. (1939) The essentiality of some elements in minute quantity for plants and special reference to copper. Plunt Physiol. 14, 371 -375. Barnes, R. M. (1984) Determination of trace elements in biological materials by inductively coupled plasma spectroscopy with novel chelating resins. Review. Biol. Tr. El. Res. 6, 93- 103. Behne, D. (1992) Speciation of trace elements in biological materials: trends and problems. Analyst 117, 555-557. Bell, P. F. et al.( 1992) Microelement concentration in Maryland air-cured tobacco. Commun. Soil Sci. Plant Anal. 23, 1617-1628. Berr, CI. et al. (1993) Selenium and oxygen-metabolizing enzymes in elderly community residents: a pilot epidemiological study. J . Am. Geriatr. Soc. 41, 143 - 148. Bowen, H. J. M. (1966) The biochemistry oftrace elements. London: Academic Press. Caroli, S. (1992) The accuracy syndrome in trace element analysis of biological samples. Microchim. J . 45, 257 - 27 1. Chowdhury, B. A,, Chandra, R. K. (1987) Biological and health implications of toxic heavy metal and essential trace element interactions. Progr. Food Nutr. Sci. 11, 57- 1 1 3. Cornelis, R. (1992a) Quality control in trace element analysis of clinical and biological samples. How good are your data? Review. J . Tr. El. Electr. Hlth. Dis. 6, 129-135. Cornelis, R. (1992b) Use of reference materials in trace element analysis of foods. 43, 307-313. Fardy, J. J., Warner, 1. M. (1992) A comparison of neutron activation analysis and inductively coupled plasma mass spectrometry for trace element analysis of biological materials. J . Radioanal. Nucl. Chem. 157, 239-244. Frieden, F. (1985) New perspectiveson theessential traceelements. J . Chem. Educ. 62,917-923. Gibson, R. S. (1984) The interpretation of human hair trace element concentrations. Sei. Total Environ. 39, 93- 101. Horowitz, C . T. (1988) Is the major part of the periodic system really inessential for life? J . Tr. El. Electr. Hlth. Dis. 2, 135- 144. Iyengar, G. V. (1991) Millestones in biological trace element research. Sci. Total Emiron. 100, 1 - 15.

90

I . Pais

Jones, J. B. (1987) Instrumental methods for trace element determination. In: New re.sulfs in the reseurch qf hirrilly known truci’ e1mzent.c. mid the analytical problems i f trace elerrlent reseurch. Pais, I. (cd.), Budapest: Univ. Press. Kiigi, J. H. R., Schaffcr, A. (I9XX) Biochemistry of metallothionein. Biochem. 27, 8509-8515. Kiss, F. et al. (1985) The effect of titanium and gallium on photosynthetic rate of algae. J . Plti/it N ~ t r 8. . 82.5 -83 1 . Kwiatck, W. M . ct al. (1992) Application of FTIR, PIXE, and EBS for trace element analysis in biological samples. Nuel. Insrr. Meth. P h j x Rex B64. 51 2 - 5 16. Lindh, U. (1993) Nuclear microscopy in trace-element biology - from cellular studies to thc clinic. Nirc.1. Instr. M d i . PIIJ.~.Res. B77, 261 -267. Liu, Z. (1988) The crrects or rare earth elements in growth of crop. In: Neb!, results in the rescurcli of’hurdlv knoicw tracr elements uniltheir role in thefbodchain. Pais, I. (ed.), Budapest : Univ. Press. Lyon, T. D. B., Fell. G. S. (1992) Trace elements in autopsy tissue. Food Chcm. 43, 299-306. Markcrt, B. ( I 988) Interelement correlations in dirferent reference materials. Fres. Z.A d . Chcvn. 332, 630-6.35. Marshall, J. ct al. (1991) Determination of trace elements in solid plastic materials by laser ablation-inductively coupled plasma mass spectrometry. J . A n d . Atom. Spectr. 6, 145- 1 50. McLaren, J. W. et al. (1993) One-line method for the analysis of sea-water for trace elements by inductively coupled plasma mass spectrometry. J . Anal. Atom. Spectr. 8, 279 - 2x6. Minoia, C. e l al. (1992) Trace element reference values in tissues from inhabitants of the European Community. I l l . The control of preanalytical factors in the biomonitoring of trace elements in biological fluids. Sci. T o t d E~iuiron.120, 63 - 79. Moro, R . et al. ( I 992) Trace elements in full-term neonate hair. J . Tr. El. E k t r . Hlth. Di.c.6,27 3 I Moyer, T. P. et al. (1991) Blood-collectioii device for trace and ultra-trace metal specimens evaluated. Clin. C17em. 37, 709-714. Novik-Fodor, M. ct al. (1992) Preparation of zirconium-containing yeast. In: N e w per.c.piwii;rs in the resrwrch of’htrrrlljk n o i c ~trui~’~ l e m ~ n t . Pais, c . . I. (ed.), Budapest: Univ. Press. Novhk, F. M. ct al. (1992) Zirconium. als ein neues Spurcnclement. Experiment niit Flefen. 111: I.?. Ar/w;rsrirguq: Mcngoi- und Spiireni~I~~t~~erztu. Anke, M., Groppel, B., Giirtler, M., Griin, M., Lombeck, I.. Schneidcr, H. J . (eds.), Oberlungwitz: Mugier Druck. Odeh, M . (1992) The role of zinc in acquired immunodeficiency syndrome. J . Intertiat. Med. 231. 463 -469. Ovaskaincn, M. J. ct al. (1993) Toenail selenium as an indicator o f selenium intake among niiddlc-sgcd m c n in an area with low selenium. A m . J . C’lin. Nutr. 57. 662-665. Pais, I. (1971) Chemistry, agricuture and human race (in Hungarian) Tudom. Meziigazd. 9(6) 9- 13. Pais. I. (1983) The biological importance of titanium. J . Plunt Nutr. 6, 3 - 131. Pais, 1. (ed.) (1985) N c w r.rs1rlt.c.in the rrseurch qf’ h u r ~ t l )known ~ trace cle?nents. Budapest: Univ. Press, pp. 1 - 2 9 , Pais, I . (cd.) ( I 987) Ncri, rc~.c.ultsin the reseirrch of‘ h a r d l ~knoitw ~ trace elerrients crnd the trntrl~ytic~rl prohli~nisid ~ Y L I W cliviirnt research. Budapest : Univ. Press, pp. 1 - 309. Pais, 1. (1988a) Importance of hardly known trace elements and convincing data on the beneficial character o f titanium. I n : Proc. Internat. C o n g r m Plant Pli,~.’siol.Sinha, S. K., Sane, P. W., Bhargava. S. C., Agrawnl, P. K. (cds.), New Delhi: Neo Art Press, Vol. 2. pp. 1172- 1177. Pais, 1. (ed.) (1988 b) Ncn’ rr.srr1t.s in the reseurcli ofhurdly k n o w i ~trace elements trnd their role in thcJ food clitrin. Budapest: [Jniv. Press pp. 1-219. Pais, I . et al. (1989) The rolc of titanum ascorhate in animal nutrition and reproduction. In: 1’1. 117rertwt. Tr. El. SIVn7p., Jena. Anke, M., Biiumann, W., Briiuiilich, H., Briickner, Ch., Ciroppel, B., G r i n , M. (cds.), Oberlungwitz: Mugler Druck. Pais, I. (1990a) The problematic of essentiality and bcncficiality of trace elements. In: Pro(.. I V . Irrtc~rnrrt.Tr. El. S J W ~Budapest, ~., Pais, I. (ed.), Budapest: Univ. Press. Pais, 1. (cd.) (1 990 b) Neii. results qf’ hurdly kriown trace rlrments und their importance in thc Intcrnu/ionul Grosphere and Riospheri. Progrrrmnie. Budapest: Univ. Press pp. 1 - 350. Pais, I . (1991) Thc criteria of csscntiality, beneliciality and toxicity. What is too little and too much‘?I n : C.yc/ing o/nir/ritioc c1iwiciit.s ingeo- andhiosphere. Pais I. (ed.), Budapest: Univ. Press. Pais, I . (ed.) ( I 992) Ncic’pi,r.spei.tiue.sin the research nf’hrrrtllyknonw truce elenimts. Budapest: Univ. Prcss pp. I -230. -

Trace Elements Need Trace Analysis

91

Rauser, W. E. (1990) Phytochelatins. Annu. Reo. Biochem. 59, 61 -86. Robertson, J . (1987) Trace elements i n human hair - a significant forensic future? Trends Anal. Chem. 6, 65-69. Sabbioni, E. et al. (1992) Trace element reference values in tissues from inhabitants of the European Community. 11. Examples of strategy adopted and trace element analysis of blood, lymph nodes and cerebrospinal fluid in Italian subjects. Sci. Total Emiron. 120, 39-62. Sinchez, R. T. et al. (1988) The value of trace elements as a nutritional status index: a statistical analysis. Tr. El. Med. 5, 150- 153. Schlesinger, D. et al. (1992) Effect of zinc-fortified formula on immunocompelence and growth of malnourished infants. Am. J . Clin. Nutr. 56, 491 -498. Schramel, P. (1993) Trends and demands in and on trace element analysis in the biomedical and environmental field. J . Radioanal. Nuc/. Chenz. 168, 21 5-221. Schwarz, K. (1970) Control of environmental conditions in trace element research: an experimental approach to unrecogzized trace element requirements. In : Truce element metabolism in animals. Mills, C . F. (ed.), Edinburgh-London: E. & S. Livingstone Publ. Simon, L. et al. (1989) Influence of gallium on photosynthetic pigments and peroxidase activity of Chlorella pyrenoidosa. J . Plant Nutr. 12, 1123- 1140. Sturaro A. ct al. (1993a) Simultaneous determination of trace metals in human hair by dynamic ion-exchange chromatography. Anal. Clzini. Acta 274, 163 - 170. Sturaro A. et al. (l993b) Spurious interferences in the ion chromatographic determination of trace elements i n human hair. C'hromatogr. 35, 675-678. Takagi, Y. et al. (1986) Trace elements in human hair: an international comparison. Bull. Emiron. Contam. Toxicol. 36. 193 - 800. Tolg, G., Gartcn, R. P. H. (1985) Great fear of small amounts of elements - The significance of analytical chemistry in our modern industrialized community as exemplified by trace element analysis. Angew. Cliem. 24, 485 -494. Turnlund, J . R. (1 989) The use of stable isotopes in mineral nutrition research. J . Nutr. 119.7 - 14. Valkovic, V . (1989) Application of nuclear analytical techniques in the study of trace element role in biology and medicine. Nucl. Instr. Meth. Phys. Res. B40-41, 848-852. Versieck, J . et al. (1987) Accuracy of biological trace element determination. Review B i d . Tr. El. Rcs. 12, 45 - 54. Versieck, J. et al. (1988) Certification of a second-generation biological reference material (freeze-dried human serum) for trace element determination. Anal. Chim. Acta 204, 63 -75. Weisner, B. et al. (1987) CSF copper concentration: a new paramcter for diagnosis and monitoring thcrapy of Wilson's disease with cerebral manifestation. J . Neurol. Sci. 79, 229-237. Wilhelm, M. et al. (1991) Monitoring of cadmium, copper, lead and zinc status in young children using toenails: comparison with scalp hair. Sci. Total Enuiron. 103, 199-207. Willis. J. B. (1970) In: Analyticu1,Jlame spectroscopy Mavrodineau, R. (ed.), New York: Springer Verlag pp. 525 - 598. Wu, Z. et al. (1 985) The effect of rare earth clemcnts in nodulation and nitrogen fixation of soybean plants. In: Proc. hiternrrt. Con/:, Bcijing, Vol. 2. Xu, G., Xiao, J. (eds.), Beijing: Sci. Press. Zalups, R. K., Cherian, M. G . (1992) Renal metallothionein metabolism after a reduction of renal mass. 11. Effect of zinc pretreatment of the renal toxicity and internal accumulation of inorganic mercury. To.\-icol. 71, 103- 117. Zhuang, G. et al. (1991) Study of the correlation on trace elements in human hair and internal organs by NAA. J . Radioanal. Nucl. Cliem. 149, 305 - 3 15. Zunk, B. (1990) MikrowellenaufschluR zur Bestimmung von Spurenelementen in Pflanzenniaterial. Anal. Chin?. Acta 236, 337-343.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

4 Error Estimation in Environmental Sampling and Analysis Michael H . Ramsey

4.1 Introduction Sampling the environment can never be perfect, there will always be some errors in the process of sampling. There are two approaches possible to address this situation. Traditionally every effort is made to take the most representative sample possible, and not to consider the effects of any errors incurred in the sampling process. The alternative is to attempt to measure the errors introduced by the process of sampling. It is only when the sampling errors are measured that these estimates can be propagated through to give a reliable interpretation of the variations of concentrations between sites. For example, a representative sample of soil from a contaminated site may give a lead analysis of 600 pg . g- ',which is over a guideline value of 500 pg . g- '. This appears to support an unambiguous conclusion that the site is contaminated with lead. Confirmation of this conclusion may be inferred from a repeated chemical analysis of 610 pg . g-' lead, and an assurance that the method of chemical analysis has a systematic error of < 10%. If, however, a second representative sample is taken from the same site and analyzed giving a lead concentration of 200 pg . g- then the conclusion of contamination would seem to be more ambiguous. The representivity of sampling is therefore uncertain and needs to be estimated before conclusions can be drawn as to the lead concentration of the site, of if the sampling procedure needs to be improved. The aims of this chapter are to provide the ideas and the methodology that will enable an environmental geochemist to estimate the errors in both of the measurement processes, i.e., of sampling and analysis. It also aims to provide criteria to establish whether these errors are at acceptable levels, that will not invalidate subsequent environmental interpretation. A distinction is drawn between quality control (QC) and error estimation in sampling. In QC there is a threshold for an error below which data is acceptable for interpretation, whereas in error estimation the values of the estimates are transferred to the interpretation and expressed as limitations on that interpretation.

',

4.2 Basic Concepts and Terminology Before discussing measurement errors in detail it is important to establish the meaning of various terms. The word sample is commonly used with several different meanings. Here it is used only for a portion of material taken in the process of sampling, with the intent

M . H . R(irnwy

94

of being representative. The prepared material in the laboratory ready for analysis is called a test nzateriul and the portion of this material that is prepared for an individual analysis is called a test portion. The precision of a method is a measure of its repeatability, and expresses random error, regardless of it closeness to the true value. It is usually expressed as either an estimate of standard deviation (s) or coefficient of variation (CV = IOOs/c) at a concentration (c). Precision is not fixed for a method, but varies with concentration (Fig. 1). Precision can also be expressed as variunw ( 5 ’ ) which is simply the square of the standard deviation. Variance has the advantage of being additivc when arising in different parts of a method, as discussed below. The closeness to the “true” value achieved by a method is called the uccurucj’ and is an expression of the systematic error, quantified as the bius of the method. The “true” value of an analyte concentration at an environmental site can never be known cxactly, but can only be estimated. The bias can be expressed in two ways, either as the absolute bias (XOhs - XI,,,) of the mean observed value (Tub,) compared with the “true” value (Xtnle), or as relative bias, expressed as a percentage (lOOIXOb, - Xtrue]/.ftr,,e). Bias usually changes as a function of concentration and can usefully be characterized by plotting a number of measured values against their “true” values of concentration. The bias may be “translational” in nature, which is a fixed absolute value (e.g., pg . g- ’) over the entire range of concentration. Alternatively the bias may be “rotational” in nature, which is a fixcd relative percentage of the concentration value. In practise thc bias of a method usually has components of both of these types of bias. These components can be estimated by linear regression, where the slope and intercept coefficients estimate the rotational and translational components of the bias, rcspectively .

A

(I

Concentration

Fig. 1. Schematic representations of thc changc of precision of an analytical method with concentration. ( A ) When precision is exprcsscd as the standard deviation of the method (.sc) at any concentration (c) is considered to be a linear function defined by the intercepts s o and slope k. (B) Whcn the precision is expressed as the cocfficicnt of variation (100 s/c) the same relationship givcs an almost constant value a t high concentrations. The precision riscs rapidly as the concentration falls towards s,, to give a value of 33% at the detection limit (c,, = 3s0), and a value of 5% at ten times the detection limit (assuming k = 0.01).

Error Estimation in Sampling

95

Three other data quality indicators have been defined by the US Environmental Protection Agency (EPA) as representativeness, completeness and comparability (Smith et al., 1987), but these are less fundamental and less widely used and have therefore not been considered further.

4.3 Sampling Error in Context The initial objective of most environmental geochemistry is to estimate the true values of analyte concentration at a site. In the absence of any measurement error, the true geochemical mean concentration of the analyte (2,) has a real geochemical variation described by the standard deviation (s,) or the geochemical variance (s:). In practise there are always measurement errors. Considering the systematic errors in isolation, the overall mean observed (ZtOt)is then given by =

2,

+ B, + B,

(1)

where the observed mean is inaccurate due to the sampling bias (B,) and the analytical bias (B,). For the random errors, the variance of the practical observations (s&)is given by: 2 Sto,

+ sf + sf

= s;

(2)

where the sampling variance (s.:) and the analytical variance (s:) both overprint the true geochemical variance to some extent. The measurement variance (sf) is therefore given as sf = sf

+ sf

(3)

this simplifies Equation (2) to give ,;s

=

s,"

+ sf.

(4)

Measurement error is therefore defined here to have only two components. Sampling error encompasses all the sources of error in the field, such as bias in site selection (by accident or design) or contamination of the sample by the sampling equipment. It also includes all sample preparation errors in the laboratory, before the sample is split for the measurement of analytical error. The other type ofmeasurement error is defined as analytical error, which can include sample splitting, subsequent sample preparation, decomposition and analytical determination. Equations (1) and (2) therefore cover a broad range of errors under these two categories, and could be rewritten to include more detailed steps if required. The estimation of the sampling variance from Equation (2), requires the use of analysis of variance (ANOVA) on the data from a suitably designed experiment (Section 4.4.2). The use of the square of the standard deviation in Equation (3) gives rise to a strong dominance for the larger terms. A large sampling variance in the equation will dominate measurement variance, so that any reductions in the analytical variance will be ineffective in reducing the total measurement variance. This supports

96

M . H . Ramxj

mathematically the commonly held belief that a highly precise analysis is wasted on a badly taken sample. Conversely if the analytical variance is relatively high, then improvements in the precision of sampling may not reduce the measurement precision significantly overall.

4.4 Methods for Estimating Quality of Measurements Methods for measuring the quality of chemical analyses have been developing over several decades, and some consensus is now evident although no method is universally agreed. Estimating the quality of environmental sampling however, is a relatively new objective and fcw methods have even been described, and none widely agreed. There are, however, great similarities between the errors originating in sampling and those from chemical analysis. These two sources of error are also inextricably linked in the consideration of measurement error as a whole. It is useful therefore to examine methods of estimating analytical errors as they are both analogous and complementary to methods for estimating sampling errors.

4.4.1 Measuring Analytical Precision It appears a straight forward task to measure analytical precision. If precision is just reproducibility, then it should be sufficient to analyze one test material ten times and quote the standard deviation or coefficient of variation. One problem with this approach arises because the precision may not be constant for all samples, and one estimate of precision is therefore not representative of that for a large batch of test materials. One factor that will affect the precision is the concentration, a s discussed in Section 4.2. A second potential problem with this approach arises from the cost of making ten analyses for one sample. This will become overwhelming if all test materials have to be analyzed ten times to allow for the fact that the precision may be different for each test material. A better solution for the measurement of analytical precision was proposed by Thompson and Howarth (1976). The method uses the analysis of test materials in duplicate to estimate the precision. The proportion of the total number of test materials that need to be duplicated depends on the application but is typically 10% for large batches of samples. The use of a range of test materials chosen at random, rather than just one, compensates for any change of precision between the test materials. The obvious objection to this approach however, is that a standard deviation estimated from two measurements is very imprecise. The standard error (sc) on this estimate of the standard deviations (s)is k 50% (given by se = s/l/in = s/2). The answer to this objection is to integrate the information from a large number of duplicate analyses, together with a model of how precision changes with

Error Estimation in Sumpling

91

concentration, to give an overall estimate of the analytical precision of the method. In studies with large numbers of duplicate analyses ( >50) a regression method is used to estimate the two coefficients of the regression model. For smaller studies with limited numbers of duplicate analysis ( <50) the precision cannot be estimated by regression, but can be assessed by comparison with a results expected for a model of known precision (e.g. 5% CV), discussed below. A similar approach for the estimation of precision from determinations on duplicated analyses is by the use of analysis of variance (ANOVA). The traditional ANOVA equations for the case of n duplicate analyses on k multiple samples at one site, gives the analytical variance (s,") as: s,

=

Total of the sums of squares but the sample means Total of the degrees of freedom

SOS

-~

df

Where there are a series of measurements xij on the i-th replicate analyses, of the j-th sample at a single site. The use of this equation can be explained for an example where the lead concentration in soils has been measured on duplicate test portions ( n = 2) taken from five samples ( k = 5 ) at a single site (Tab. 1). The total sum of squares (SOS) is therefore 108 (54 + 54) and the degrees of freedom (df) are k ( n - 1) = 5(2 - 1) = 5. The analytical variance is therefore = 4.65 pg . g-I. 108/5 = 21.6, and the analytical standard deviation is For duplicate analysis Equation (5) can be simplified to give k

2* sf

(0.5 * [xl

-

~

~

1

)

~

i= 1

=

(6)

._

k

A similar method is applicable when ANOVA is applied to the estimation of both sampling and analytical variance (Section 4.4.2). The confidence interval on this estimate of s, can also be estimated from the values in Tab. 2. Tab. 1. Calculations for Estimation of Precision of Analysis for Lead in Soil (pg . g- ') Sample No. (j)

Anal 1

Anal 2

xlJ

X2j

Mean Xi

1 2 3

4 5 Sum

112 95 96 I10 110

113 98 85 103 1 04

112.5 96.5 90.5 106.5 107.0 102.6

Xli

-xi

0.5 I .5 5.5 3.5 3.0

(Xlj - Z i ) 2

(XZj - Xi)2

0.25 2.25 30.25 12.25 9.00

0.25 2.25 30.25 12.25 9.00

54.00

54.00

98

M . H . Rrrrnwy

‘lab. 2. Multipliers to Calculate 95% Confidence Intervals of Estimates of Analytical Precision (from Davies and Goldsmith 1977) ~

No. dups

No. dups.

L2

12 I ~

1 2 3 4 5 6 7 8 9 10

0.45 0.52 0.57 0.60 0.62 0.64 0.66 0.68 0.69 0.70

~~

Ll

L2

0.72 0.74 0.77 0.78 0.80 0.82 0.85 0.89 1.oo

1.65 1.55 1.44 1.39 1.34 1.28 1.22 1.14 1 .00

~

12

31.9 6.28 3.73 2.87 2.45 2.20 2.04 1.92 1.83 1.75

IS 20 24 30 40 60

I20 nc,

The lower confidence interval is calculated from L , * s, and the upper from L , * sl,. In this example with five duplicate analyses the 95% confidence interval on the estimate ofanalytical precision is 0.62 * 4.65 to 2.45 * 4.65 that is 2.88 to 11.39 pg g - I . The limitation of this method is that is does not make any allowance for the variation of precision with increasing concentration. One approach to overcome this limitation, discussed below, is the use of Robust ANOVA in which a small percentage (< 10%) of high concentration values can be “accommodated” within the ANOVA without undue influence on the estimates of the variance of the main population. Such frequency distributions are not uncommon in environmental investigations. Another option, especially where there is a large variation in the analyte concentration, is the regression method of Thompson and Howarth (1976) that can estimate this variation, for large numbers of duplicates ( > 50). For small numbers of duplicate analyses the precision can be tested against a standard model of precision (e.g., 5% (Is)) using a control chart. The duplicates in this example (Tab. 1) can be 100 50

VI

+

3

E 10 c

E5

Fig. 2. A precision control chart for the assessment of analytical precision against ;I inodcl prccision of 5% (Thompson and Howarth 1976). The duplicate analyses from Tab. 1 are plotted as the absolute difl‘erencc values against the mean value of each pair. The relatively equal distribution ot’ point on either side of the median line suggest that the precision is approximately equal to 5% ( I s ) o r 10Y0 (2s).

+

9J

n ac J u

P I Lc

0.5

0.1

1

10 100 Mean o f duplicate r e s u l t s

1000

Error Es:J.timationin Sampling

99

site

sample 1 Fig. 3. The experimental design required for the estimation of sampling and analytical precision using analysis of variance (ANOVA).

sample 2

A/*\

anal 1 anal 2

anal 1 anal 2

plotted on such a chart using the absolute difference between the two analyses against the mean value (Fig. 2). The approximately even scatter of the point on either side of the median (50%) line, and the lack of > l o % of the values above the 90% line, suggests that the precision of the analysis is approximately 5% (IOOs/c):Although this is a very small number of points with which to test precision by this method, this result agree closely with the result by calculation (4.65%). Generally if more than half of the points lie above the median line (or > 10% above the 90% line) then the precision can be judged to be worse than 5%. Conversely if more than half of the points lie below the median line (or >90°/0 below the 90% line) then the precision can be judged to be better than 5%. Where concentrations are close to the analytical detection limit of the method, the precision tends to deteriorate rapidly towards lower concentrations. In this case a threshold concentration can be specified above which the data does conform to the 5% model.

4.4.2 Measuring Sampling Precision It would seem a reasonable assumption that sampling precision could be estimated by a method analogous to that described for estimating analytical precision. This is only partially true, because the sampling precision cannot be estimated directly from such a procedure. Geochemical measurements always introduce analytical variance that overprints the sampling variance. The solution to the separation of sampling error is to use a specific experimental design followed by ANOVA of the three levels of analytical, sampling and geochemical variance. The experimental design, shown in Fig. 3, should be supported by clearly specified protocols for the methods of sampling, sample preparation and chemical analysis. Duplicate samples taken at many sites are preferable to multiple replicate sampling at a few sites. This is analogous to the case for duplicate analyses, and gives the best estimate of the sampling precision over all sites. The protocol used for the siting of the duplicated sample is critical for the realistic estimation of the sampling precision. The distance between the locations at which the duplicate samples are taken should reflect the uncertainty in the position of the location of the site in respect of its neighbors. An example could be an initial reconnaissance survey to estimate the heavy metal contamination of top soil over in an area of several km2 of wooded country. If a sampling traverse is laid down with a spacing of 100 m it is unlikely that any site can be located to within less than 5 m. Thus a duplicated sample must

I00

M . ff . R u t m q

be located 5 m from the initial sample if it is to accurately reflect the uncertainty in the metal concentration at that site, as it is specified. This applies no matter how many subsamples are taken to constitute the single sample. By contrast if a detailed follow-up soil survey has a spacing of 10 m, then the uncertainty in location of one site relative to the others may only be 0.5 m, and the spacing of the duplicate sample should reflect that. The detailed procedure for taking the sample must be specified as exactly as possible. In this example of top soil sampling, the factors that need to be specified might be:

(1) the degree to which any vegetation should be removed, (2) the depth interval to be sampled (e.g., 0- 10 cm), (3) the area of soil to be subsampled for a sample (e.g., 4 m’), (4) the number and design of subsamples to be taken, (5) prevention of subsoil contamination by top soil (e.g., smearing of top soil on removing auger from hole), (6) cleaning of sampling equipment between sites, (7) contingency plan if sample cannot be taken in specified way. However, in the real world there will always be unforeseen problems, such as obstacles (trees, dense undergrowth) or abnormal conditions (roads, paths or fences) at the specified site. The measurement of sampling error is designed to quantify the inevitable errors introduced in the processes of real sampling, rather than to assume that the sampling has perfectly adhered to the protocol on every occasion. Both of the two field samples at each site are prepared as separate test materials, and both analyzed in duplicate (Fig. 3). All four analyses for each site, should be placed anonymously at random positions in the analytical batch or batches. In this way errors due to time-based changes in the performance of the laboratory (e.g., instrumental drift) can be accurately represented in the estimates of the precision. The instrumental measurements should be stored in an unrounded form (with random error apparent in at least the last digit) and with no concentration estimates suppressed (i.e., those less than zero, or less than a detection limit). If rounded or suppressed values are included, erroneous and often optimistic estimates of the measurement precisions will result. Once the four analytical measurements are tabulated, there are two possibilities for evaluating the sampling precision. A rapid way of inspecting the precision is to plot the absolute differences against the mean values on Precision Charts as described above for analytical precision. This must be done separately for the analytical duplicates and the sample duplicates. The samping precision is tested approximately using the average analysis from the analytical duplicates as the best estimate of the concentration in each of the sample duplicates. The precision tested is not the measurement precision simply the sampling precision, but this is sufficient or for some applications. A more rigorous assessment of sampling precision is to the estimate the variances using ANOVA. Many statistical software packages can perform the classical nested one-way A N O V A required. Classical A N O V A is however based on four assumptions (Eisenhart 1947). One of these is that the data should be normally distributed.

lm,

Error Estimation in Sampling

101

However, in environmental investigations a positively skewed frequency distribution is more common. The most reliable estimates of the variance can then be made using Robust ANOVA (Ramsey et al. 1992).A special program for Robust ANOVA, written by Ripley, has been published (Analytical Methods Committee 1989). A version of this program adapted for this application is available from the author. This program takes the data in four columns A l S l A2S1 A1S2 A2S2, where for example, A2S1 is the second analysis of the first sample. An example of the output file, although not elaborate, gives all the basic information required: _ _ _ element e . . . , . . . . . No of element in data file Classical results: mean = 1 sigma values (geochem, sampling, analysis) - s, ss, a, sigma (total) - stat

Maximum analytical variance (4%)

Variances

0 Geochemical

.

95.7% 3.2y0

Analytical 1.1%’

Fig. 4. Proportions of variance contributed by measurement processes of sampling and analysis in two geochemical surveys for Pb. (a) A survey ofCornish stream sediments. Comparison with the maximum allowable variances (1) shows immediately that both variances are within specification and that the geochemical information is thereforc reliable. (b) A survey in urban park land soils. Comparison with the maximum allowable variances (1) shows immediately that the technical variance for exceeds the specification and that the geochemical information is therefore very unreliable.

Measurement 4.3%

Maximum analytical variance (4%) L

Maximum measureme variance (20%)

Variances

0 Geochemical

56.7% Sampling 43.2%-

(b)

I02

M . H . Kirrn.wr.

Robust results: mean = R sigma values (geochem, sampling, analysis) - sg, ss, a, sigma (total) stat . ~

The relevance of variance estimates depends on their relative sizes. This can best be shown graphically, using a pie chart, showing the proportions of each of the contributing variances (s,",s?, a:) as fractions of the total variances (&). An example is shown in Fig. 4. This permits the limits for the precision of sampling and analysis to be judged in relation to the geochemical variability that has also been measured. This application of ANOVA to the measurement of sampling error has concentrated on spatial variability, with an example in soil sampling. It can also be applied to sampling error due to temporal variation, which can be a problem in rapidly changing systems such as river waters. In that case duplicate samples should be taken at the same site but at different times separated by intervals that are of a similar duration to the average time variation between samples taken in the rest of the survey.

4.4.3 Targets for Acceptable Levels of Precision in Sampling and Analysis The traditional way of expressing the target for analytical precision is to use some fixed coefficient of variation. Trace element determinations in environmental samples are given a target of typically 5% (i.e., 10% at 95% confidence). This target does not however bear any relation to the scale of variation of the element in the environment that is to be measured. If some soils are grossly contaminated with the clement by several orders of magnitude over the normal background level measured in most of the soils, then this precision target is more stringent than is required to map the extent of the contamination. If, however, the objective is to identify different sources of lead contamination by isotope ratios, which differed by only 5%, then this analytical precision would be inadequate. What is required is to set the targets for measurement precision as fractions of the geochemical variability that is to be characterized. Sampling precision targets can be set in a similar way, but neither can bc set indcpendently because of the inter-relationship shown in Equations (2), (3) and (4). The target suggested for the combined measurement error is that it should not exceed 20% of the total variance (Ramsey et al. 1992). This limit is not theoretically based, but derived from practical cxperience of the interpretation of geochemical data. When the measurement variance exceeds this limit it begins to perceptibly distort the environmental interpretation of geochemical data. An example of a survey for lead in soil with 43.3% measurement variance (Fig. 4b) shows a very erratic estimation of the spatial distribution of lead (Fig. 5). The target for the analytical precision can be expressed in two ways. That it should contribute less than 20% to the measurement variance, ensures that it does

Error Estimution in Sampling

I03

predicted by the high measurement variance I0

not dominate this variance. This figure is also empirically derived in a way analogous to that for the measurement variance. In the example of the lead in soil survey the analytical variance contributed only 6% to the measurement variance, showing that the sampling variance is the main cause of the poor measurement variance. Alternatively the analytical precision target can be set at less than 4 % of the total variance (20% of 20%). This is satisfactory, but ignores the fact that the analytical error may be dominating the measurement error. Minimum targets can also be set for the measurement variances. If the analytical variance is less than 1 % of the measurement variance then it is more precise than is required, and a less expensive analytical method could be employed without contributing significantly to the measurement variance. Similarly if the measurement precision is contributing less that 1% to the total variance the measurement technique is more precise that required. In this case either the sampling protocol can be made simpler ( e g , fewer subsamples in the composite sample), or the analytical precision requirement eased. An orientation survey is required therefore if the design of the sampling and analysis protocol is to be optimized for any particular analyte in a given environment. In a multi-analyte survey the protocol must be designed for the analyte with the most stringent sampling and analytical requirements.

4.4.4 Measuring Analytical Bias Bias is more difficult to estimate than precision, because it is defined by comparison with the “true” value of the element concentration. True values are however never known, particularly for environmental sampling and analysis. There must therefore

104

M . H . Hurnsey

always be a surrogate for the true value. As is the case for precision estimation, the methodology is much more advanced for estimating analytical bias, than for sampling bias. A simple but limited test for bias at low concentrations is the reagent blank or control sample. In the laboratory a blank vessel can pick up contamination from the environment, reagents, equipment or instrumental memory effects. The mean value of the blanks can be used as onc estimate of bias, and the standard deviation of the results as a more realistic detection limit (ix., 3s) than those measured on purely instrumental variability. The main methods for the estimation of analytical bias is by the analysis of certified reference materials (CRMs). For example, the bias of the determination of Pb in soil can be estimated by analyzing a CRM that has both a certified value for Pb, and an overall composition that is closely matched to the samples. For example, the CRM, BCR 143 is a sludge amended soil has a certified P b concentration (X,) of 1333 pg . g-', and a 95% confidence limit of 39 p g . g- '. When this CRM was analyzed for P b five times using an acid digest (HNO, + HCIO,) and determination by ICP-AES, the concentrations determined were: 1171, 1342, 1374, 1236 and 1312 pg . g-' Pb. The mean of these measured values (2,) is 1287 pg . g-'. The standard deviation (s,) = 82.58 pg . 6 -l. bias

=

X,

%bias

=

bias

- Zc

* lOO/X,

= =

1287-1333 -46

* 100/1333

=

-46 pg . g-'

=

-3.45%

Although the bias can always be calculated, it is not necessarily statistically significant. This can be checked using a t-test.

The tabulated value of t is 2.776 ( n = 4, p = 0.05). The calculated t value must exceed this value if it is to indicate significant bias. Because the calculated t value (1.246) is less than the tabulated value (2.776) then the calculated bias of this analytical method is not statistically significant. There are problems with this method. CRMs are only available for a small number of sample types and are often only certified for a limited number of elements. Moreover the certified values are not the true value and are thercforc themselves subject to uncertainty. A more involved test (an unpaired t-test) can be applied that also allows for this uncertainty. Environmental materials are relatively well represented by CRMs but the diversity of sample types and analytes required mean that no suitable CRM currcntly cxists for the many combinations of sample type and analyte. This is particularly true for certified values on specified species of elements and those portions of element extracted by selectivc reagents. A further complicating factor is that the bias of an analytical method often changes at different concentrations. Estimates of bias need to be made therefore

Error Estimution in Sampling

I05

with at least two CRMs, and ideally more, spanning the full range of expected analyte concentrations. Where larger number of CRMs have been analyzed, regression of measured on accepted concentration values can be used to further characterize the bias into its translational and rotational components, as a function of concentration. Analytical bias can also be estimated by two other methods. In the “comparative analysis” method, some or all of the test materials are analyzed by different analytical methods and ideally in different laboratories. The idea is that if different methods (or laboratories) agree then there is a greater confidence that the results are accurate, i.e., not biased. Problems often arise when the methods (or laboratories) do not agree, and subjective judgements must be made to decide on which methods give less biased analyses. The third method of estimating analytical bias is to use spike recovery. This depends on the addition of a known mass of analyte to the sample prior to analysis. The recovery of this spike is calculated from the ratio of the difference between the analysis of the spiked and unspiked samples over the known concentration of the spike. Low spike recovery implies that analyte may have been lost during preparation (e.g., by volatilization) or that its response has been suppressed by an instrumental inter-element interference (e.g., a matrix effect). Good recovery does not necessarily however demonstrate small bias for the method, as the spike will often be in a form different from the analyte present in the test material. Spike recovery has the advantage that the matrix used is matched to the samples, unlike CRMs. It is often used therefore as part of a quality control protocol in addition to the use of CRMs ( e g , EPA, 1987). Once the bias has been estimated for a set of determinations by one or more of these methods, the question arises as whether to “correct” the analyses and remove the bias. The detection of bias is problematic, but the exact quantification of the bias at all concentration levels is rarely possible. The correction of bias is at best, inexact and at worst, merely adding further to the overall uncertainty, and is therefore not advisable. Conversely uncorrected analyses with soundly based estimates of the bias are a firm basis for rigorous interpretation of environmental measurements.

4.4.5 Estimating Sampling Bias Sampling bias is the most problematic component to estimate, and therefore least addressed, of all of the measurement uncertainties in environmental geochemistry. Many of the methodologies developed for estimating analytical bias can in theory be adapted for the estimation of sampling bias. The blank sample certainly has a place in field sampling. For water sampling, an empty sample container prepared in the standard way, can be taken into the field and filled on site with high purity water, that has been filtered and acidified in the same way as the samples. The analysis of such field blanks can give evidence of

106

M . H . Ramsey

contamination from the environment or the equipment and reagents. The concept of a control site is often used in experimental design to act as a background area against which to compare the test site. The selection of an “uncontaminated” site for this purpose, that is matched to the test site in every other characteristic is often difficult to achcive completely. Generally, there are no sampling equivalents of CRMs for the estimation of bias. One exception is the use of gases of known composition for air sampling, but sampling pure gas can never be quite the same as sampling air with complications such as particulate matter. There are no reports of reference fields with known levels of soil contamination for the validation of methods of soil sampling, although these are a theoretical possibility. To certify these fields perhaps several independent workers could sample the field by different sampling designs and protocols. A best estimate of the average metal concentration of the field, and perhaps its spatial distribution, could be established as a certified values with confidence intervals. Closely related to this approach is the use of comparative sampling and analysis. If, for example, three total independent workers, using different sampling and analytical techniques, could agree on the analyte concentration at a site then perhaps the bias of a fourth worker could be measure against that agreed value. Spike recovery to assess sampling bias in ambient air is already in use (Smith et al. 1987). Adapting this idea to water sampling is feasible, but problematic for heterogeneous media such as soils (Jackson 1987). Questions arise as to the chemical form of the element that should the spike be added, how homogeneity could be established, and how stability could be maintained.

4.5 Targets for Acceptable Levels of Bias in Sampling and Analysis Acceptable levels of bias are not often specified for geochemical analysis, and when targets are quoted (e.g., < l o % ) they are rarely rigorously derived. If improvements are to be made to reduce bias and if zero bias is impossible, then a target requirement is desirable. Such a target depends on the objective of the environmental sampling, and the ramifications of undetected bias. There are application of environmental sampling, such as mineral exploration in which it has been said that precision is more important than accuracy (Rose et al. 1979). If the objective is to identify areas of “high” concentrations then only poor precision will obscure this. In this context accuracy is of secondary importance in establishing confidence in how high the peak concentration really is. Analogous to this is the environmental survey to establish the source of pollution, in which the spatial pattern is more important than the absolute concentration level. In this case poor measurement precision is more important to identify the source, rather than the accuracy. This is demonstrated in Fig. 5., were there peaks of P b concentration top

Error Estimation in Sampling

107

soil are close to road at 100 and 1900 m along the traverse, but other “spurious” peaks occur at 400 and 1700 m. Accuracy becomes important when comparing the estimated concentrations to external values, such as background levels, other studies or regulatory limits. Thus if the objective of the survey shown in Fig. 5 was to delineate the areas over a 500 pg . g- regulatory trigger concentration, then accuracy becomes of prime importance. The absolute accuracy required may be legally specified, for example in EPA QA/QC protocols bias on RMs is required to be within f 2 0 % and spike recovery, within +25%. The target for the accuracy can be based however, on a particular objective. If the objective is to delineate the area with > 500 pg . g Pb then, if the precision is sufficient to map the variation with confidence (i.e., measurement variance <20% total variance), and so long as the bias is estimated, then the appropriate contour can be used. If the bias is estimated to be - 15%, then in that case the contour of 425 c(g. g-’ measured Pb can be used to estimate the position of the 500 pg . g- contour. There is a distinct difference apparent here between the quality control approach, and that of error estimation in this example. The EPA quality control approach would accept data with bias <20%, whereas the estimated bias of -15% can be incorporated into the interpretation of the concentration data.

’

4.6 Conclusions The estimation of errors in environmental sampling and analysis is a fundamental requirement for sound interpretation of data. Analytical quality control is well established in many laboratories but the estimation of values for precision and accuracy rather than simply asking whether an analysis is within certain quality limits, is a more powerful approach. These error estimates can then be used to constrain the environmental interpretation of the concentration data. Sampling errors are rarely estimated, often because they are not considered the responsibility of the analytical laboratory. Integrated schemes for the estimation of both sampling and analytical errors can be devised that can estimate both precision and accuracy. Moreover the integrated scheme allows the comparison of the errors in the measurements with the natural variation of the geochemistry, to set realistic limits for the quality of data that is required for a particular investigation.

4.7 References Analytical Methods Committcc (1989) Robust statistics - How not to rcjcct outlicrs. Part 2: Inter-laboratory Trials. Anuljst 114, 1699- 1705. Davis, 0. L., Goldsmith, P. L. (1977) Stutistical Methods in Research and Production. Longman, Harlow, England. Eisenhart (1947) The assumptions underlying the analyses of variance. Biornetrics 3, 1-21,

I ox

M . H. R~inwe.~.

EPA, 1987, Statement of Work for the Contract Laboratory, Program SOW 787, Exhibit E. U.S. Dept. of Commcrce, National Technical Inrormation Service, Springfield V.A. 22 161, USA. Jackson, L. P. (1987) Sampling and analysis of hazardous and industrial wastes: special quality assurance and quality control considerations. - In: Principles of Enuironmental Sampling, Keith. L. H. (ed.), American Chemical Society, Washington. Smith, F., Kulkarni, S., Mycrs, L. E., Messner, M. J. (1987) Evaluating and prescnting quality assurance sampling data. In: Principles of’EnnuironrnentcrlSiimpling, Keith. L. H. (ed.), American Chemical Society, Washington. Thompson, M., Howarth, R. J. (1976) Duplicate analysis in geochemical practice. Part I : Thcoretical approach and estimation of analytical reproducibility. Af?NIy.(.I101, 690 - 698. Ramscy, M. H., Thompson M., Halc. M. (1992) Objectivc evaluation of precision rcquiremcnis f o r geochemical analysis using robust analysis of variance. J . Geoclzem. E.xpl. 44, 23 - 36. Ramscy, M. H. (1993) Sampling and analytical quality control (SAX) for improved error estimation in the measurement of heavy metals in the environment, using robust analysis of variance. Appl. Geocheni. 2, 149- 153. Rose, A . W., llawkes, H . E., Webb, J . S . (1979) Geochemisfry in M i n i J r d Explorrrtion. Academic Press.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

5 Estimation with Varying Detection Limits Willium G. Warren

5.1 Introduction Commonly, the measurement of a trace substance on a sample unit is reported as being below a detection limit. It is clear, for example, that if one or more concentrations in a random sample of size n units fall below a detection limit, the average of the remaining measurements will overestimate the mean concentration for the population under study. The population standard deviation will, likewise, be underestimated. The situation is further complicated if the detection limit differs for each sample unit. Common practice is to substitute a constant for the below detection limit values; typical values for the constant are zero, half the detection limit or even the detection limit itself. Such a simplistic approach clearly has its limitations (see, for example, El-Shaarawi and Esterby, 1992). Somewhat better approaches include the estimation of below detection limit values through (log) probability plots, uniform distribution substitution for below detection limit values, or the use of the median and interquartile range instead of the mean and standard deviation. However, if the detection limits are known quantitatively, and the form of the distribution of the concentration over the units in the population can be assumed, for example, that the concentrations are normally, or lognormally distributed, the standard statistical procedure of maximum likelihood estimation of the distributional parameters, e.g., the mean and variance, can be applied. In spite of publications such as El-Shaarawi and D o h (1989), this approach does not appear to be widely employed by toxicologists and chemists. With observations falling below a single detection limit, the data are referred to, in statistical parlance, as type I singly censored samples. Estimation of the distributional parameters in this situation dates back to Cohen (1959). He, and many subsequent authors, including El-Shaawari and Dolan (1989), treat the case of a single detection limit rather than multiple limits. (Note that the term “doubly censored” as employed by Cohen (1959) means that observations above a prescribed value are censored as well as those below a smaller prescribed value.) El-Shaawari and Naderi (1991) consider multiply censored data but their model differs slightly from that considered below in that, in their paper, the number of observations that fall between successive detection limits is taken as known. In the present paper, it is not known whether an observation that falls below its detection limit, falls above or below the smaller detection limits of other individuals. Sometimes measurements are made for more than one substance on each sample unit and the degree of association of the concentrations of these substances may

W . G . Wcirrcn

I10

then be of interest. For example, the correlation matrix of a suite of variables could be used to determine if the variables are clustered in some meaningful or interpretable manner, perhaps throwing light onto the origin of pollutants. If, for one or more of the variables, there is a single value below a detection limit and the remaining values all fall above that limit, either Kendall’s or Spearman’s rank Correlations can be calculated uniquely. If there is more than one belowdctcction-limit value and/or, in the case of varying detection limits, there arc measured values below the detection limits recorded for other individuals, it will be possible, in general, to place upper and lower limits on the rank correlations; this may be useful if the upper and lower limits are sufficiently close to each other. The point is that there is information in the detection limit values that can bc exploited. In particular, under the assumption of bivariate normality, maximum likelihood estimates of the means, variances and the correlation coefficient can be obtained. This is the primay focus of the present paper. In what follows the requisite theory will be outlined and illustrated. The approach can be extended to situations of other than bivariate normality.

5.2 Methodology 5.2.1 The One-Dimensional Case We begin by considering the one-dimensional case. Let x, i = 1, 2, ..., n be independent, identically distributed random variables having a normal distribution with mean ,u and variance 0 2 . Let us suppose, without loss of generality, that the first n , of these are observed but that the remaining n2, n = n , + n 2 , fall below detection limits, lj, j = n 1 + 1, n , 2, ..., n. The likelihood of the sample is then

+

whence

c ni

log ( L ) = - n , log

(271)

-

n1 log

(0) -

(1/202)

(Xi -

p)2

I1 2

+ C log ( Q ( ( l j

-

P)/c))

I

where @(.) denotes the cumulative distribution function of the standard normal. N ( 0 , l), i.e. x

@(.XI

j”

= ~

rx

(

I/P) exp [ - t2/21 d t .

Estimution with Detection Limits

Ill

The maximum likelihood estimates are then found by equating the partial differentials of log (L),with respect to p and (T, to zero and solving the resulting pair of equations for p and (T, or 0 2 .Thus, with ej = (xj - p ) / ( ~ ,

and

a 1% a.

(L)-

-

(

T

i

f (T

lj -

~~

j

P q(oj)

(T

@(Oj)

where q(.)denotes the probability density function of the standard normal. For covenience we shall write F(Oj) for cp(Bj)/@(Bj). In the case of n2 = 0, i.e. n1 = n, the likelihood equations reduce to the usual maximum likelihood estimates n

fi

=

Ci x i / n = x

and n

d2 =

Ci ( x i - $)'/it.

However, when n2 > 0 the likelihood equations cannot be solved explicitly for p and 0 2 . Let y' = [p (TI and g' = [a log (L)/apa log (L)/ao]. Then the NewtonRaphson method of solution has Wk+l

= Iflk - H p ' ( Y ) k ) g ( W k )

where H is the Hessian or matrix of second partial derivatives of log ( L ) ;i.e. log ( ~ ) / a ~ a2 log (L)/aa ap a2

log (Lyap ari1 a2 log ( L ) / a d '

2a 2

Note that

aF(e) aP

q o ) ao do

ap

and d F ( 4 - d ( P ( w @ ( a- P'(4 dB do @(O)

~-

V2(Q

Q2(0)

Since q ( 0 ) = ( 1 / p ) exp [-d2/2], ~ ' ( 0= ) -&p(B).

aFo = - F ( B ) [e + F(@]ao . -

aP

i%

It then follows that

I12

W . G . Wurrm

Similarly for aF(fl)/aa.Then, specifically

, . . . can be generated. The approach requires an initial guess, yo,from which y ~ , tp2, With a satisfactory initial guess, these willconverge to th; sdution of the likelihood equations. A reasonable initial guess will, in general, be the mean and standard deviation of the sample values with the below detection limit values replaced by, say, half the detection limit. However, as the proportion of below detection limit values increases so does the difficulty of finding satisfactory starting values and several attempts may be required before a converging sequence is obtained. Examples are given below. One advantage of this approach is that the variance-covariance matrix of the estimates is estimated by the negative of the inverse of the Hessian evaluated with p = i; and cr = 6.This is, of course, an approximation which is, however, correct asymptotically.

5.2.2 The Bivariate Case We now turn out attention to the bivariate case. We suppose for the time being that below detcction values occur with only one of the two substances (variates) measured. Then, if all y1 values of variate x fall above detcction limits, but only n , values of variate y, under bivariate normality the likelihood of the sample is L = L , L , where

n'

1

nr

L,

=

2na1a,

1/1-

~

Q2

~ X [-Q(xi, P ~J/'(2(1 -

e2)1

Estimation with Detection Limits

113

with, here,

and e denotes the correlation coefficient. Explicit expressions for the first and second partial derivatives of log ( L )are given in the Appendix. More generally, the data can be partitioned into four subsets; firstly, those individuals in which both variables fall below their respective detection limits, l j and mj, say. Let there be n I 1 such individuals. Next those individuals in which one variable, x, falls below its detection limit, lj, but the other variable, y, does not; let there be n l o such individuals. Likewise let their be no, individuals for which variable y falls below its detection limit, mj, but the other variable does not. Finally let there be no, = n - n , , - no, - n , individuals for which both variables fall above their respective detection limits. The likelihood of the sample is then the product of four components, specifically

,

This will not be pursued further other than to remark that the development parallels that given above, although the expressions will be somewhat more complex and involve the bivariante normal integral (available in some computer software packages) as well as the standard normal integral.

5.3 Examples For example, Tab. 1 lists the ovary concentrations of several organic compounds for which one or more values fell below reported detection limits. Summary statistics of these data have been reported by Hellou et al. (1993).

114

W . 6'. W~irrm

Tab. 1. Ovary Concentrations of Organic Compounds with at least one Value below Dctcction Limits ___

~~

Compound pp-DDE Heptachlor epoxide Arochlor 1254 pp-DDD pp-DDT Y-HCH op-DDD Endrin Arochlor 1260

Concentration

1.0 0.4 4.9 1.0 1.8 0.5 0.7 0.5

1.7 0.4 5.1 1.6 2.7 0.7

1.1

1.8

3.8 0.4 6.5 1.5 3.0 0.7

2.0 0.3 1.6 2.4 3.2 0.6

1.0

1.4

1.4

0.5

0.5 2.5

0.8 2.7

0.5

2.7 0.4 4.3 2.2 2.1 0.6 1.3 0.8

0.2 3.2 1.9 2.8 0.3 2.3 0.7 1.7

4.3 0.4 7.7 2.5 4.2 0.8 1.3 0.9 2.4

1.9

1.0 1.0 4.2 1.2 2.0 0.9 0.9 0.9 2.6

1.4 0.6 6.3 2.2 2.1 0.6 1.0 0.7 2.7

2.3 0.6 9.0 2.6 3.6 0.9 1.5 1.0

2.6

Note 1: PCB was measured as Archlor standards. Note 2: The detection limits of those values failling bclow the limits are given in bold face.

Tab. 2 presents the mean, mu, and standard deviation, so, estimated solely from the data that fall above the detection limits, and estimates m , and s, obtained by utilizing the detection-limit values, along with the estimated standard errors (sen,, se,) of the latter, as measures of their precision. Tab. 2. Estimates of the Mean and Standard Deviation and their Standard Errors Tor Organic Compounds with at least one Value below Detection Limits Compound

n

mu

So

m1

sr,,

pp-DDE Heptachlor epoxide Arochlor 1254 pp-DDD pp-DDT ~J-HCH op-DDD Endrin Arochlor 1260

1

2.24 0.52 5.61 1.96 2.70 0.73 1.60 0.70 2.40

1.11 0.20 1.14 0.46 0.80 0.14 0.41 0.16

2.01 0.48 5.18 1.81 2.59 0.61 0.84 0.52 0.6 I

0.41 0.07 0.72 0.19 0.21 0.08 0.35 0.1 I I .23

1

1 2 2 3 6 7 9

-

s

,

.YE,

1.28 0.23 2.26 0.53 0.79 0.28 0.79 0.22 1.08

0.3 I 0.06 0.55 0.15 0.20 0.06 0.31 0.09 0.8 1

whcrc n dcnotcs thc number, out of 10, of observations that rall bclow their detection limits.

The ovary concentrations of other compounds for which all observations fell above detection limits are given in Tab. 3. Note that the values in a column are from the same fish and that the column order is the same as in Tab. I . Tab. 3. Ovary Concentrations of Organic Compounds with n o Values below Detection Limits Compound 1-HCH Oxychlordane Trans-chlordane Trans-nonachlor Cis-nonachlor HCB Dieldrin

Concentration 1.8 1.4 2.0 3.5 1.4 1.9 1.6

'

2.8 2.0 2.5 5.5 2.0 2.0 1.5

2.8 2.5 1.8 7.8 1.7 4.1 1.5

2.3 1.3 1.8 4.8 1.3 3.0 0.9

1.6 1.6 2.6 4.9 1.9 4.1 1.0

2.6 2.1 2.8 7.5 2.2 4.4 1.4

3.6 2.5 3.9 9.2 2.5 5.4 2.4

2.6 2.8 2.6 5.6 2.3 2.0 2.0

3.2 2.0 2.0 4.0 1.6 1.4 1.9

3.7 4.0 3.1 8.1 3.6 3.0 1.9

Estimation with Delrction Limits

I I5

Tab. 4. Estimates of the Mean and Standard Dcviation and their Standard Errors for Organic Compounds with no Values below Detection Limits Compound

m1

Se,

r-HCH Oxychlordane Trans-chlordane Trans-nonachlor Cis-nonachlor HCB Dieidrin

2.70 2.22 2.5 1 6.09 2.05 3.13 1.61

0.21 0.24 0.20 0.58 0.20 0.40 0.14

,

se,

0.65 0.75 0.63 1.83 0.63 1.25 0.43

0.15 0.17 0.14 0.4 1 0.14 0.28 0.10

s

The maximum likelihood estimates of the mean and variance (or standard deviation) of these variables when obtained jointly with the maximum likelihood estimate of the mean and variance (standard deviation) of the variables of Tab. 1 are the same as the maximum likelihood estimates of the variables when obtained in isolation. This, in effect, reduces the set of likelihood equations to be solved from five equations in five unknowns to three equations in three unknowns. The estimates, along with their standard errors, for the variables of Tab. 3 are given in Tab. 4. The standard error estimates of Tab. 4 are generally, but not universally, less than the conventional moment-based estimates. This should not be surprising for an asymptotic approximation applied to a sample of size as small as ten. Notwithstanding, the differences are relatively small, generally being of the order of 10- 15%. Tab. 5 presents the maximum likelihood estimates of the mean and standard deviation of the variables of Tab. 1, i.e. those variables which contain one or more values below detection limits. Unlike the estimates for those variables with no below-detection-limit values, these estimates differ with which of the variables of Tab. 3 they are jointly estimated. In effect, the correlation with each Tab. 3 variable provides somewhat different information to assist in the esimation of the parameters of each Tab. 1 variable. This problem could be avoided if a single multivariate distribution, rather than a set of bivariate normal distributions, was assumed. In general the differences are small and, certainly, inconsequential, so that the added complexity entailed by a model of, say, eight dimensions would appear to be unjustified. Tab. 5. Estimates of the Mean and Standard Deviation and their Standard Errors for Organic Compounds with at least one Value bclow Detection Limits when Estimated Jointly with Compounds with no Values below Detection Limits Compound

Compound

pp-DDE

m2

se,

s2

se.5

2-HCH Oxychlordane Trans-chlordane Trans-nonachlor Cis-nonachlor HCB Dieldrin

1.9929 2.006 1 2.01 67 2.04 17 2.0095 2.0389 1.9983

0.41 85 0.4095 0.4012 0.3855 0.4072 0.3874 0.4153

1.3064 1.2816 1.2612 1.2159 1.2752 1.2210 1.2968

0.3214 0.3094 0.2983 0.2756 0.3061 0.2783 0.3183

Average

2.1049

0.4036

1.2654

0.301 I

I16

W. G . Wurrrn

Tab. 5. (continued) Compound

Compound

Heptachlor epoxide

U-HCH Oxychlordane Trans-chlordane Trans-nonachlor Cis-nonachlor HCB Dieldrin Average

0.4766 0.4786 0.4789 0.4786 0.4789 0.4787 0.4788 0.4784

0.0753 0.0740 0.0738 0.0740 0.0738 0.0739 0.0737 0.074 I

0.2353 0.2315 0.23 10 0.23 16 0.2310 0.2313 0.2309 0.2318

0.0574 0.0559 0.0557 0.0561 0.0557 0.0558 0.0550 0.0559

Arochlor 1254

2-HCH Ox ychlordane Trans-chlordane Trans-nonachlor Cis-nonachlor HCB Dieldrin Average

5.2.144 5.2282 5.1903 5.2094 5.2120 5.1857 5.1926 5.2047

0.6955 0.6849 0.7144 0.6993 0.6970 0.7188 0.7 I I7 0.7031

2.1898 2.1600 2.243I 2.2007 2.1946 2.2546 2.2365 2.21 13

0.5089 0.4938 0.5370 0.5143 0.5100 0.5463 0.5299 0.5200

pp-DDD

U-HCH Ox ychlordane Trans-chlordane Trans-nonachlor Cis-nonachlor HCB Die 1d r i n Avcragc

1.7955 1.7706 1.7782 1.7933 1.7722 1.8145 1.7814 1.7865

0.1953 0.1964 0.1854 0.1882 0.1870 0. I876 0.200 I 0.1944

0.5888 0.5952 0.5728 0.5755 0.5787 0.5710 0.5949 0.5824

0.1509 0 1507 0. I365 0.1408 0.1381 0.1405 0. I559 0.i 448

pp-DDT

z-HCH Oxychlordane Trans-chlordane Trans-nonachlor Ciwwnachlor HCB Diendrin Average

2.5997 2.5598 2.5714 2.5933 2.5670 2.5965 2.5452 2.5761

0.257 1 0.27 13 0.26 18 0.2526 0.2665 0.2585 0.28 10 0.264 I

0.7769 0.8049 0.7963 0.7720 0.8009 0.7833 0.8266 0.7947

0 I846 0.201 3 0.1903 0.1794 0.1957 0.1858 0.2 145 0.1931

y-HCH

z-HCH Oxychloi ddnc Trans-chlordane Trans-nonachlor Cis-nonachlor HCB Dieldrin Avcragc

0.6407 0.6165 0.6147 0.6183 0.6163 0.6124 0.6025 0.6173

0.0619 0.0743 0.0765 0 0745 0.075 1 0.0781 0.08 I4 0.0745

0.1927 0.2221 0.2264 0.2222 0.2235 0.2302 0.2399 0.2224

0.0450 0.0599 0.0637 0.0612 0.06 14 0.0662 0.068 1 0.0608

1-HCH Oxychlorddne Trans-chlordane Trans-nonachlor Cis-nonachlor HCR Dieldrin Average

0.8438 0.8 194 0.7121 0.7891 0.7750 0.8244 0.8777 0.8059

0.3534 0.3676 0.4322 0.3946 0.3919 0.3678 0.3364 0.3777

0.7897 0.8 I35 0.8945 0.8506 0.8452 0.8141 0.7595 0.8239

0.3161 0.3298 0.38 16 0.3581 0.3479 0.3225 0.2970 0.3361

~~

op-DDD

m2

W,

s2

,Y5

1 I7

Estimation with Detection Limits Tab. 5.

(continued)

Compound

Compound

Endrin

a-HCH Oxychlordane Trans-chlordane Trans-nonachlor Cis-nonachlor

0.5539 0.528 1 0.4996 0.4135 0.5094

Average Oxychlordane

Arochlor 1260

se,

s2

Se,

0.0986 0.1109 0.1378 0.1810 0.1308

0.2034 0.2139 0.2230 0.28 I8 0.2184

0.0725 0.0837 0.0995 0.1523 0.093 1

0.5005

0.1318

0.228 1

0.1002

1.0925

0.8360

0.8428

0.5342

m2

For Arochlor 1260, with 9 out of 10 values below detection limits, convergence was attained only in conjunction with oxychlordane; for endrin, with with 7 values below detection limits, convergence was not attained in conjunction with HCB and dieldrin. As might be expected, as the number of below detection limit values increases so does the variability of the estimates obtained in conjunction with the various compounds. Finally, in Tab. 6, we present the estimates of the correlation between the variables of Tab. 1 and 3, along with estimates of their standard error. Note the tendency for the standard errors to increase with the number of below-detection-limit values. Estimates and Standard Errors of the Correlation between Organic Compounds with and without Values below Detection Limits Tab. 6.

Compound

Compound a-HCH

pp-DDE Heptachlor epoxide Arochlor 1254 pp-DDD pp-DDT y-HCH OP-DDD Endrin Arochlor 1260

0.633 0.191 0.694 0.165 0.686 0.170 0.512 0.236 0.569 0.218 0.923 0.050 -0.028 0.340 0.445 0.417

OxyTransTransCischlordane chlordane nonachlor nonachlor 0.381 0.270 0.391 0.268 0.866 0.080 0.713 0.164 0.574 0.222 0.634 0.194 0.215 0.330 0.190 0.528 0.569

0.391 0.270 0.105 0.3 13 0.592 0.205 0.842 0.098 0.719 0.161 0.237 0.307 0.578 0.240 -0.174 0.651 0.392

0.278 0.292 0.04 1 0.3 17 0.664 0.178 0.762 0.136 0.789 0.122 0.364 0.282 0.434 0.286 -0.696 0.290

0.860 0.084 0.205 0.303 0.780 0.125 0.864 0.084 0.632 0.199 0.447 0.261 0.471 0.268 -0.090 0.700

HCB

Dieldrin

0.655 0.187 -0.388 0.269 0.174 0.307 0.546 0.223 0.691 0.168 -0.214 0.307 0.687 0.186

0.431 0.259 0.727 0.149 0.789 0.119 0.386 0.302 0.484 0.268 0.780 0.132 -0.186 0.350

I18

W . G. Warren

5.4 Discussion Although, for the most part, the method has apparently been successfully applicd to samples of size 10, except when the number of below detection limit values approaches the sample size, this should not be taken to imply that samples of such a small size will generate meaningful results. Indeed, the standard errors are, in general, uncomfortably large. A sample of 10 fish was used for illustration simply becausc that was the only sample available to the author at the time. The procedurc should, however, be quite viable for larger sample sizes, say 30 or more, with up to 20% of the observations below detection limits. As noted above, the method can be applied, in theory, when are are considerably more than 20% of the observations below detection limits, although it becomes progressively more difficult to calculate the estimates and their quality degrades. O n the other hand, it is probably better to utilize the information that is available than to ignore it. It has been assumed that the data comprise a random sample from a (bivariate) normal distribution. Clearly the approach could be applied to lognormally distributed data. It must be remembered that the estimated correlation coefficient then applies to the (logarithmically) transformed variables. Similar considerations apply if other transformations to normality are employed, in particular the Box-Cox transformations. Linearity is not maintained when transforming back to the original scale. Acknowledgement. The author wishes to express his appreciation to Dr. J. Hellou for drawing his attention to the problem and for providing the revelant data.

5.5 References Cohcn, A. C. Jr. (1959) Simplificd estimators Tor the normal distribution when samples arc singly censored or truncated. T i ~ h n o r n i ~ ~1,i c21 s 1-231. El-Shaarawi, A. H.. D o h , 0. M. (1989) Maximum likelihood estimates of water quality concentrations from censored data. Can. J . Fish. Ayunt. Sci. 46, 1033- 1039. El-Shaarawi, A . H., Naderi, A. (199 I ) Statistical inference from multiply censored environmental data. Enr:iron. Monit. A.we.r.7. 17, 339 - 347. El-Shaarawi, A . H., Esterby, S. R. (1992) Replacement of censored observations by a constant: an cvaluation. Wmer Res. 26, 835-844. Hellou, J., Warrcn, W. G., Paync, J. F. (1993) Organochloridcs including polychlorinatcd biphenyls in muscle, livcr and ovaries ofcod (Gadus rriorhirrr). Arch. En'noironrn.Contmn. To.ric. 25,497 - 505.

5.6 Appendix For completeness, we here give explicit expressions for the elements of R (w)and H ( y ) where, for the bivariate case, w' = [rn1p26,02~] and the log likelihoodis ~

log ( L ) =

-(@I

+ n2/2) log (2x)

-

@

log

(61) - @,

log

(62)

Estimation with Detection Limits

119

Let xi = ( x i - ,ul)/al, yi = (yi- p2)/a,, X j = (xj - pl)/ol and L j = ( l j - ,u2)/c2. As before F(Oj) = cp(Oj)/@(Oj) and, for convenience, we shall write K(Oj) for F(Oj). Then

k13

=

o:(I

-

[$ (@I: e’) i

n2

-

[2(1 - e2)X j

2Xi) j

I

+ &l2XjF(Oj)K(Oj)

+e

vm

F(Oj)

E,ytimation with Detection Limits

121

h45 =

In spite of their appearance, the above equations can be readily programmed in any one of a number of common computer languages. Code in SAS IML is available on request to the author.

Part I11 Examples for Sampling A. Air

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

6 Particle and Gas Measurements on Filters John C. Wutson and Judith C. Chow

6.1 Introduction

-

Particles with aerodynamic diameters less than 50 pm (termed Total Suspended Particulate matter, or TSP) can become suspended into the atmosphere, and those with aerodynamic diameters less than 10 pm (termed P M can transport over large distances (Lancaster and Nickling, 1993) and enter the human respiratory system (Ferris et al., 1979; Miller et al., 1979). Particles with aerodynamic diameters less than 2.5 pm are most effective at scattering light and have a great effect on visibility (Maim, 1979) impairment and the Earth’s radiation balance (Charlson et al., 1992). The adverse health impacts of suspended particles have caused the United States to adopt National Ambient Air Quality Standards (NAAQS) for PM,, of 50 pg/m3 for an annual average, and 150 pg/m3 for a 24-hour average (Federal Register, 1987a). Similar health-based standards have been adopted, or are being considered, in other countries. In addition to mass concentration and particle size, knowledge of the chemical composition of suspended particles is important. This chemical composition is of special concern for health, since toxic chemicals such as lead, arsenic, beryllium, cadmium, chronium, copper, selenium, zinc, acidic species, and certain organic compounds can cause both short-term and long-term medical problems. Different chemicals affect the transfer of radiation through the atmosphere in different ways. The application of receptor models (Watson et al., 1989, 1990, 1991a; Chow et al., 1993a) also requires chemical characterization of these particles to apportion ambient concentrations to their sources for the development of emissions reduction strategies. Several chemical constituents of suspended particles form in the atmosphere from gases which are emitted by pollution sources. The gaseous precursors of particulate sulfate, nitrate, and ammonium are sulfur dioxide (SO,), oxides of nitrogen (NO and NO,, the sum of which is designated NO,), and ammonia (NH,). It is often necessary to measure the concentrations of these gases along with the particulate phase in order to understand their sources, formation mechanisms, and health hazards. Aerosol samples are most often acquired by drawing ambient air through filter material using a pump. The particles remain deposited on the filter, while all gases pass through the filter. When the filter is impregnated with an absorbing solution, or when the filter material has specific gas-absorbing properties, quantitative measures of gases as well as particle phases are possible. After sampling, filters can be submitted to laboratory analyses for the desired chemical compounds. Chemical analysis of filter deposits cannot be separated from the methods used to obtain the

I26

J . G. Wutson arid J . C. Choic,

sample. Investigators sometimes collect long records of particulate filter samples, store them in envelopes or file boxes for many years, then send them to a general chemical laboratory to obtain some numbers for different chemical constituents. These invcstigators are often disappointed when they must defend these numbers as representing what was in the air at the time of sampling. Sampling for chemical analysis requires stringent attention to filter media, to sample handling, to sample storage, and to the sampler used to obtain the filter deposits. When this chemical analysis is intended for source apportionment receptor modeling, the use of sequential sampling systems, particle and gas sampling systems with denuders, and portable samplers may be needed. When chemical analysis of samples is anticipated, the first consideration should be which analysis methods will be applied, and how samples can be obtained to meet the necds of those analyses. This chapter provides an overview of filter-based particle and gas sampling systems. These systems consist of more than the mechanical devices used to acquire the sample. The laboratory analyses to be applied, the types of filters which are amcnable to those analyses, thc minimum deposits needed on these filters, the sampling hardware which extracts pollutants from the atmosphere onto the filters, and the procedures which assure the accuracy, precision, and validity of the acquired atmospheric concentrations must all be considered. Though the chapter is brief given the breadth of the subject area, extensive references are identified to provide more depth in each topic area.

6.2 Filter Analysis Methods Several chemical analysis methods have been adapted to the quantification of particulate and precursor gas concentrations in the atmosphere, and some understanding of these methods is necessary to obtain samples which can be appropriately submitted. Lodge (1989) and Appel (1993) provide extensive summaries of the principles, procedures, and results of these methods applied to the analysis of suspended particles. Tab. 1 compares minimum detectable concentrations achievable by different analysis methods for several chemicals. The values in Tab. 1 are nominal, and actual detection limits should be supplied by the laboratory performing the analysisprior to sampling, so that sample durations and flow rates can be adjusted to acquire sufficient samples for the intended analyses. Thc most common aerosol analyses can be divided into the categories of mass, elements, ions, and carbon.

6.2.1 Mass Gravimetric analysis is used almost exclusively to obtain mass measurements of filters in a laboratory environment. Gravimetry determines the net mass by weighing the filter before and after sampling with a balance in a temperature- and relative

127

Measurements on Filters

Tab. 1. Analytical Measurement Specifications for Air Filler Samples ~

Species

ICPI AESh Be Na Mg AI Si P S

c1 K Ca

sc Ti

v

Cr Mn Fc

co Ni

cu Zn Ga As Se Br Rb Sr Y Zr Mo Pd

0.06 NA 0.02 20 3 50 10

NA NA 0.04 0.06 0.3 0.7 2 0.1 0.5 1 2 0.3 1

42 50 25 NA NA 0.03 0.1 0.6 5 42

Ag

1

Cd In Sn Sb 1 Cs Ba La Au Hg TI Pb Ce

Sm Eu

0.4 63 21 31 NA NA 0.05 10 2.1 26 42 10 52 52 0.08

Hf Ta

26

16

Minimum Dctcction Limit in ng/m7" _ _ _ . - -- - - AA AA Fldmeh Furnaceh lNAAb PIXEg XRF' 2d 0.2* 0.3 30 85 IOO,000 NA NA 24 Id 50 95 52 2 1

4 6" 5 4 1

52 100 100

NA NA 4 300 1000 31 10

4 1

31 31 31 NA NA 8" 2,000 21 500 21 10 NA 2,000 21 2,000 2,000

0.05 < 0.05 0.004 0.0 I 0.1 40 NA NA 0.02 0.05 NA NA 0.2 0.01 0.01 0.02 0.02 0.1 0.02 0.001 NA 0.2 0.5 NA NA 0.2 NA NA 0.02 NA 0.005 0.003 NA 0.2 0.2 NA NA 0.04 NA 0.1 21 0.1 0.05 NA NA NA NA NA

N A ~ 2 300 24 NA NA 6,000 5 24 94 0.001 65 0.6 0.2 0.12 4 0.02 NA 30 3 0.5 0.2 0.06 0.4 6 18

NA NA NA NA 0.12 4 0.006 NA 0.06 1

0.03 6 0.05 NA NA NA NA 0.06 0.0 1 0.006 0.0 1 NA

NA 60 20 12 9 8 8 8 5 4

NA 3 3 2 2 2 NA 1

I I 1

I 1

1 2 2 NA 3 5 NA NA NA NA NA NA NA NA NA NA NA NA NA 3 NA NA NA NA NA

NA NA NA 5 3 3 3 5 3 2 NA 2 1

I 0.8 0.7 0.4 0.4 0.5 0.5 0.9 0.8 0.6 0.5 0.5 0.5 0.6 0.8 1

5 6 6 6 8 9 NA NA 25 30 2

I 1 1 NA NA NA NA NA

____

lCb

ACh

TORh

NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA

NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA

NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA NA

1.G. Watson and J . C. Chow

I28

Tab. 1. (continued) Minimum Detection Limit in ng/m3"

Species

W Th U

c1NO;

so:

NH;

oc EC

31 63 21 NA NA NA NA NA NA

1,000 NA 25,000 NA NA NA NA NA NA

NA NA NA NA NA NA NA NA NA

0.2 0.01 NA NA NA NA NA NA NA

NA NA NA NA NA NA NA NA NA

NA NA 1 NA NA NA NA NA NA

NA NA NA 50 50 50 NA NA NA

NA NA NA NA NA NA

50 NA NA

NA NA NA NA NA NA NA 100

100

Minimum detection limit is three times the standard deviation of the blank for a filter of 1 mg/cm2 arcal density. ICP/AES = Inductively Couplcd Plasma with Atomic Emission Spectroscopy. AA = Atomic Absorption Spcctrophotometry. PIXE = Proton Induced X-ray Emission Analysis. XRF = X-ray Fluorescence Analysis. INAA = Instrumental Neutron Activation Analysis. IC = Ion Chromatographic Analysis. AC = Automated Colorimetric Analysis. TOR = Thermal Optical Reflectance Analysis. Concentration is based on the extraction of Ij2 of a 47 mrn filter in 15 mL of deionizcd-distillcd water, with a nominal flow rate of 20 L/min for 24-hour samples. Concentration is based on 13.8 cm3 deposit area for a 47 mm filter substrate, with a nominal flow rate of 20 L/min for 24-hour samples with 100 s radiation time. Harrnan (1 989). Fernandez ( I 989). OIrnez (1989). Eldred (1993). Not Available.

humidity-controlled environment. PM,, reference methods (Federal Register, 1987b) require that filters be equilibrated for twenty-four hours at a constant (within +_ 5%) relative humidity (RH) between 20% and do%, and at a constant temperature (within f3 "C) between 15 "C and 30 "C. These are intended to minimize the liquid water absorbed by soluble compounds and to minimize the loss of volatile species. Nominal values of 30% RH and 15 "C to 20 "C best conserve the particle deposits during sample weighing. The main interference in gravimetric analysis of filters results from electrostatic charges, which induce on-gravimetric forces between the filter and the balance (Engelbrecht et al., 1980). The charge can be removed from most filter material by exposure to a low-level radioactive source prior to and during weighing. Accurate gravimetric analyses require the use of filters with low dielectric constants, high filter integrity, and inertness with respect to absorbing water vapor and other gases.

Measurements on Filters

129

6.2.2 Elements As noted above, the most common interest in elemental composition derives from concerns about health effects and the utility of these elements to trace the sources of suspended particles. Instrumental neutron activation analysis (INAA), atomic absorption spectrophotometry (AAS), inductively coupled plasma with atomic emission spectroscopy (ICP/AES), photon-induced x-ray fluorescence (XRF), and proton induced x-ray emission (PIXE) have all been applied to elemental measurements of aerosol samples. AAS and ICP/AES are also appropriate for ionic measurements when the particles are extracted in deionized distilled water (DDW). Since air filters contain very small particle deposits (20 to 100 pg/cm2), preference is given to methods which can accommodate small sample sizes. XRF and PIXE leave the sample intact after analysis so that it can be submitted to additional examinations by other methods. In INAA (Dams et al., 1970; Zoller and Gordon, 1970; Olmez, 1989), a sample is irradiated in the core of a nuclear reactor for periods ranging from a few minutes to several hours. The neutron bombardement chemically transforms many elements into radioactive isotopes. The energies of the gamma rays emitted by these isotopes identify them, and therefore their parent elements. The intensity of these gamma rays is proportional to the amount of the parent element present in the sample. Different irradiation times and cooling periods are used before counting for different time periods with a germanium detector. INAA does not quantify some of the abundant species in ambient particulate matter such as silicon, nickel, tin, and lead. While INAA is technically nondestructive, sample preparation involves folding the sample tightly and sealing it in plastic, and the irradiation process makes the filter membrane brittle and radioactive. These factors limit the use of the sample for subsequent analyses. In AAS (Ranweiler and Moyers, 1974; Fernandez, 1989), the sample is first extracted in a strong solvent to dissolve the particle deposits; the filter, or a portion thereof, is also dissolved during this process. A few milliliters of this extract are introduced into a flame where the elements are vaporized. Most elements absorb light at certain wavelengths in the visible spectrum, and a light beam with wavelengths specific to the elements being measured is directed through the flame to be detected by a monochromater. The light absorbed by the flame containing the extract is compared with the absorption from known standards to quantify the elemental concentrations. AAS requires an individual analysis for each element, and a large filter or several filters are needed to obtain concentrations for a large number of the elements specified in Tab. 1. AAS is a useful complement to other methods, such as XRF and PIXE, for species such as beryllium, sodium, and magnesium which are not well-quantified by these methods. Airborne particles are chemically complex and do not dissolve easily into complete solution, regardless of the strength of the solvent. There is always a possibility that insoluble residues are left behind and soluble species may co-precipitate on them or on container walls. In ICP/AES (Fassel and Kniseley, 1974; McQuaker et al., 1979; Lynch et al., 1980; Harman, 1989), the dissolved sample is introduced into an atmosphere of

130

J . G. Watson rind J . c'. Chow

argon gas seeded with free electrons induced by high voltage from a surrounding Tesla coil. The high temperatures in the induced plasma raise valence electrons above their normally stable states. When these electrons return to their stable states, a photon of light is emitted which is unique to the element which was excited. This light is detected at specified wavelengths to identify the elements in the sample. ICP/AES acquires a large number of elemental concentrations using small sample volumes with acceptable detection limits for atmospheric samples. As with AAS, this method requires complete extraction and destruction of the sample. In XRF (Dzubay and Stevens, 1975; Jaklevic et al., 1977) and PIXE (Cahill et al., 1990; Eldred, 1993), the filter deposit is irradiated by high energy x-rays (XRF) or protons (PIXE) which eject inner shell electrons from the atoms of each element i n the sample. When a higher energy electron drops into the vacant lower energy orbital, a fluorescent x-ray photon is released. The energy of this photon is unique to each element, and the number of photons is proportional to the concentration of the element. Concentrations are quantified by comparing photon counts for a sample with those obtained from thin-film standards of known concentration. Emitted x-rays with energies less than -4 keV (affecting the elements Na, Mg, Al, Si, P, S, C1, and K) can be absorbed in the filter, in a thick particle deposit, or even by large particles in which these elements are contained. Very thick filters also scatter much of the excitation radiation or protons, thereby lowering the signal-tonoise ratio for XRF and PIXE. For this reason, thin membrane filters with deposits in the range of 10 to 50 pg/cm2 provide the best accuracy and precision for X R F and PIXE analysis. XRF and PIXE are also most accurate for particles with geometric diameters less than 10 pm and for elements which emit x-rays with high energies ( > 4 keV). Dzubay and Nelson (1975) have developed mathematical corrections for this absorption which depend on a general knowledge of the particle size distribution on the sample. These corrections are more accurate when size-selective inlets with known transmission characteristics are incorporated into the particle sampling system. Most of these elemental analysis methods are applied to a portion of the filter. Therefore, it is important to acquire a sample with a uniform deposit of known area across the filter so that the amounts measured can be scaled to the entire deposit.

6.2.3 Water Soluble Ions Aerosol ions refer to chemical compounds which are soluble in water. The water-soluble portion of suspended particles associates itself with liquid water in the atmosphere when relative humidity increases, thereby changing the light scattering properties of these particles. Different emission sources may be distinguished by their soluble and non-soluble fractions. Gaseous precursors can also be converted to their ionic counterparts when they interact with chemicals impregnated on the filter material.

Meusuremmts on Filters

131

Several simple ions, such as sodium, magnesium, potassium, and calcium, are best quantified by AAS as described above. In practice, AAS has been very useful for measuring water-soluble potassium and sodium, which are important in apportioning sources of vegetative burning and sea salt, respectively. Polyatomic ions such as sulfate, nitrate, ammonium, and phosphate, must be quantified by other methods such as ion chromatography (IC) and automated colorimetry (AC). Simple ions, such as chloride, may also be measured by these methods along with the polyatomic ions. All ion analysis methods require filters to be extracted in DDW and then filtered to remove the insoluble residue. The extraction volume needs to be as small as possible, otherwise the solution becomes too dilute to detect the desired constituents. Each square centimeter of filter should be extracted in no more than 2 m L of solvent for typical sampler flow rates of 20 to 30 L/min and sample durations of 24 hours. This often results in no more than 20 mL of extract which can be submitted to the different analytical methods, thereby giving preference to those methods which require only a small sample volume. Sufficient sample deposit must be acquired to account for the dilution volume required by each method. When other analyses are to be performed on the same filter, the filter must first be sectioned using a precision positioning jig attached to a paper cutter. For rectangular filters (typically 20.32 cm x 25.40 cm), a 2.0 cm x 20.32 cm wide strip is cut from the center two-thirds of the filter. Circular filters are usually cut in half for these analyses, so the results need to be multiplied by two to obtain the deposit on the entire filter. Filter materials for these analyses must be chosen that can be easily sectioned without damage to the filter or the deposit. IC can be used for both anions (fluoride, phosphate, chloride, nitrate, sulfate) and cations (potassium, ammonium, sodium) with separate columns. Applied to aerosol samples, the anions are most commonly analyzed by IC with the cations being analyzed by a combination of AAS and AC. In IC (Small et al., 1975; Mulik et al., 1976, 1977; Butler et al., 1978; Mueller et al., 1978; Rich et al., 1978; Small, 1978), the sample extract passes through an ion-exchange column which separates the ions in time for individual quantification usually by a conductimetric detector. Prior to detection, the column effluent enters a suppressor column where the chemical composition of one element is altered, resulting in a matrix of low conductivity. The ions are identified by their elution/retention times and are quantified by the conductivity peak area or peak height. IC is especially desirable for particle samples because it provides results for several ions with a single analysis and it uses a small portion of the filter extract with low detection limits. AC applies different colorimetric analyses to small samples volumes with automatic sample throughput. The most common ions measured are ammonium, chloride, nitrate, and sulfate (Butler et al., 1978; Mueller et al., 1978; Fung et al., 1979; Pyen and Fishman, 1979). Since IC provides multi-species analysis for the anions, ammonium is most commonly measured by AC. A peristaltic pump introduces air bubbles into the sample stream at known intervals to separate samples in the continuous stream. Each sample is mixed with reagents and subjected to appropriate reaction periods before submission to a colorimeter. The ion being measured usually reacts to form a colored liquid. The liquid absorbance is related

I32

J . G . Wutson and J . C. Chow

to the amount of the ion in the sample by Beer’s Law. This absorbance is measured by a photomultiplier tube through an interference filter which is specific to the species being measured. The major sampling requirement for analysis of water-soluble species is that the filtcr material be hydrophilic, allowing the water to penetrate the filter and fully extract the desired chemical compounds. Small amounts of ethanol or other wetting agents are sometimes added to the filter surface to aid the extraction of hydrophobic filter materials, but this introduces the potential for contamination of the sample.

6.2.4 Organic and Elemental Carbon Three classes of carbon arc commonly measured in aerosol samples collected on quartz-fiber filters : (1) organic, volatilized, or non-light absorbing carbon; (2) elemental o r light-absorbing carbon; (3) carbonate carbon. Carbonate carbon (i.e., K,CO,, Na,CO,, MgCO,, CaCO,) can be determined on a separate filter section by measurement of the carbon dioxide (CO,) evolved upon acidification (Johnson et al., 1981).Though progress has been made in the quantification of specific organic chemical compounds in suspended particles (e.g., Rogge et al., 1991), sampling and analysis methods have not yet evolved for use in practical monitoring situations. Many methods have been applied to the separation of organic and clemental carbon in ambient and source particulate samples (McCarthy and Moore, 1952; Mueller et al., 1971; Lin et al., 1973; Patterson, 1973; Gordon, 1974; Grosjean, 1975; Smith et al., 1975; Appel et al., 1976, 1979; Kukreja and Bove, 1976; Merz, 1978; Rosen et al., 1978; Daisey et al., 1981; Dod et al., 1979; Johnson and Huntzicker, 1979; Macias et al., 1979; Malissa, 1979; Weiss et al., 1979; Cadle et al., 1980a, b ; Cadle and Groblicki, 1982; Heisler et al., 1980a, b; Johnson et al., 1980, 1981; Pimenta and Wood, 1980; Mueller et al., 1981; Novakov, 1981, 1982; Gerber, 1982; Heintzenberg, 1982; Huntzicker et al., 1982; Muhlbaier and Williams, 1982; Rosen et al., 1982; Tanner et al., 1982; Stevens et al., 1982; Wolff et al., 1982; Japar et al., 1984). Comparisons among the results of the majority of these methods show that they yield comparable quantities of total carbon in aerosol samples, but the distinctions between organic and elemental carbon are quite different (Countess, 1990; Hering et al., 1990). The definitions of organic and elemental carbon are operational and reflect the method and purpose of meassurement. Elemental carbon is sometimes termed “soot”, “graphitic carbon” or “black carbon”. For studying visibility reduction, light-absorbing carbon is a more useful concept than elemental carbon. For sourcc apportionment by receptor models, several consistent but distinct fractions of carbon in both source and receptor samples arc desired, regardless of their light-absorbing or chemical properties. Differences in ratios of the carbon concentrations in thcsc fractions form part of the source profile which distinguishes the contribution of onc source from the contributions of other sources. Light-absorbing carbon is not entirely constituted by graphitic carbon, since there are many organic materials which absorb light (e.g., tar, motor oil, asphalt, coffee).

Measurements on Filters

133

Even the “graphitic” black carbon in the atmosphere has only a poorly developed graphitic structure with abundant surface chemical groups. “Elemental carbon” is a poor but common description of what is measured. For example, a substance of three-bond carbon molecules (e.g., pencil lead) is black and completely absorbs light, but four-bond carbon in a diamond is completely transparent and absorbs very little light. Both are pure, elemental carbon. Chow et al. (1993b) document several variations of the thermal (T), thermal/ optical reflectance (TOR), thermal/optical transmission (TOT), and thermal manganese oxidation (TMO) methods for organic and elemental carbon. The TOR and T M O methods have been most commonly applied in aerosol studies in the United States. The thermal/optical reflectance (TOR) method of carbon analysis developed by Huntzicker et al. (1982) has been adapted by several laboratories for the quantification of organic and elemental carbon on quartz-fiber filter deposits. While the principle used by these laboratories is identical to that of Huntzicker et al. (1982), the details differ with respect to calibration standards, analysis time, temperature ramping, and volatilization/combustion temperatures. In the TOR method (Chow et al., 1993b), a filter is submitted to volatilization at temperatures ranging from ambient to 550 “C in a pure helium atmosphere, then to combustion at temperatures between 550 “C to 800 “C in a 2% oxygen and 98% helium atmosphere with several temperature ramping steps. The carbon which evolves at each temperature is converted to methane and quantified with a flame ionization detector. The reflectance from the deposit side of the filter punch is monitored throughout the analysis. This reflectance usually decreases during volatilization in the helium atmosphere owing to the pyrolysis of organic material. When oxygen is added, the reflectance increases as the light-absorbing carbon is combusted and removed. Organic carbon is defined as that which evolves prior to re-attainment of the original reflectance, and elemental carbon is defined as that which evolves after the original reflectance has been re-attained. By this definition, “organic carbon” is actually carbon that does not absorb light at the wavelength (632.8 nm) used, and “elemental carbon” is light-absorbing carbon (Chow et al., 1993b). The thermal optical transmission (TOT) method applies the same thermal/optical carbon analysis method except that transmission instead of reflectance of the filter punch is measured. Thermal methods (T) apply no optical correction and define elemental carbon as that which evolves after the oxidizing atmosphere is introduced. The thermal manganese oxidation (TMO) method (Mueller et al., 1982; Fung, 1990) uses manganese dioxide (MnO,), present and in contact with the sample throughout the analysis, as the oxidizing agent, and temperature is relied upon to distinguish between organic and elemental carbon. Carbon evolving at 525 “C is classified as organic carbon, and carbon evolving at 850 “C is classified as elemental carbon. Carbon analysis methods require a uniform filter deposit because only a small portion of each filter is submitted to chemical analysis. The blank filter should be white for light reflection methods, and at least partially transparent for light transmission methods. The filter must also withstand very high temperatures without melting for the combustion methods.

134

J . G.

Wfitsan

id J . C. Chaw

6.3 Filter Media Particle sampling filters consist of a tightly woven fibrous mat or of a plastic membrane which has been penetrated by microscopic pores. Tab. 2 identifies several filter types and manufacturers which have been successfully used in different sampling systems when chemical characterization is desired. Each of the particulate analysis methods described above places different demands on the filter used to collect the sample. Sometimes these demands are contradictory, and a single filter samplc cannot be used for all of the desired analyses. General criteria which must be considered when selecting filter media are: Mechanical stability: The filter must remain in one piece, lie flat in the sampler filter holder, and provide a good seal with the sampling system to eliminate leaks. A brittle filter material may flake and negatively bias mass measurements. PM, , reference methods (Federal Register, 1987b) specify weight losses or gains due to mechanical or chemical instability of less than a 5 pg/m3 equivalent. For many studies, a 1 pg/m3 tolerance is imposed. If the filter is to be divided into more than one portion, the filter must allow precise and accurate sectioning. The pure quartz-fiber filters listed in Tab. 2 are very brittle, and portions of their edges often become attached to the filter holder. This biases mass measurements. Ringed Teflon membranes are stretched between a ring, and these curl when they arc cut in half or when a punch is removed. Ringed Teflonmembrane filters should not be used when filter sectioning is required. - Temperature stability: The filter must retain its porosity and structure in the presence of temperatures typical of the sampled airstream and of the analysis methods. Plastic filters, for example, may melt in the presence of hot exhaust from an industrial source. Filters that melt during thermal carbon analysis can encapsulate the deposit, making it unavailable for combustion and detection. All of the membrane filters have a plastic base, and should not be used to sample air streams with temperatures exceeding 50 "C. All but the pure quartz-fiber filters will melt when subjected to the temperatures commonly applied for thermal carbon analysis. - Chemical stability: The filter should not interact chemically with the deposit, even when submitted to strong extraction solvents. It should not absorb gases that are not intended to be collected. When gas absorption is desired, the filter material should absorb those gases at near 100% cfficiency. PM,, reference methods require a filter alkalinity of less than 25 microequivalents/gm to minimize absorption of sulfur dioxide (SO,) and nitrogen oxides (NO,). Coutant (1977), Spicer and Schumacher (1977), and Meserole et al. (1976, 1979) tested a variety of filter materials for these species and found substantial adsorption for all of these species on glass-fiber filters, with a minor adsorption of nitric acid on quartz-fiber filters. Eatough et al. (1990) and McDow and Huntzicker (1990) demonstrate evidence of organic vapor adsorption on quartz-fiber filters. Demuynck (1975) and Charell and Hawley (1981) show the large effect on mass of water vapor adsorption on cellulose-fiber filters.

-

-

Compatible Analysis Methods"

Gravimetry, light transmission, XRF, PIXE, INAA, AAS, ICP/AES, IC, AC

ICPIAES, IC, AC, T, TOR, TMO, TOT

Gravimetry , XRF, PIXE AA, ICPIAES for some metals, IC, AC, T, TOR, TMO, TOT

Ringed Teflon membrane, Gelman (Ann Arbor, MI), Teflo", R 2PJ 047, R2PJ037

Pure quartz-fiber, Pallflex (Putnam, CT), 2500 QAT-UP

Mixed quartz fiber, Whatman (Hillsboro, OR), QM/A # 1861865

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

-

Quartz fibers with 5% borosilicate content. White opaque surface, diffuses transmitted light. Some batches can melt at 500 "C. Effects on thermal carbon analysis are unknown. Becomes brittle when heated. Low flow resistance.

-

Mat of pure quartz fibers. White opaque surface, diffuses transmitted light. Edges flake in most filter holders. Melts at >900 "C Moderate flow resistance.

Thin membrane stretched between polymethylpentane ring. White surface, nearly transparent. Minimal diffusion of transmitted light. Cannot be accurately sectioned. 1.2, 2.0, and 3.0 pm pore sizes. Melts at -60 "C. High flow resistance.

Physical Characteristics

Particle Sampling Filters and Relevant Physical and Chemical Characteristics

Filter Type, Major Manufacturer, and Example

Tab. 2.

0

0

0

0

0

0

0

0

0

0

0

Contains large and variable quantities of Na, Al, and Si in all batches. Variable levels of other metals are found in many batches. Passively adsorbs organic vapors. Adsorbs little HNO,, NO,, and SOz. Low hygroscopicity. High blank weight.

Extensively washed during manufacturelow blank levels for ions. Contains large and variable quantities of A1 and Si. Some batches contain other metals. Passively adsorbs organic vapors. Adsorbs little HNO,, NO2, and SO,. High blank weight.

Usually low blank levels, but several contaminated batches have been found. Made of carbon-based material, so inappropriate for carbon analysis. Inert to adsorption of gases. Low blank weight.

Chemical Characteristics

2

3 2

0

0,

9

2

2

$E:

$

Gravimetry, XRF, PIXE, INAA, AAS, ICPIAES, IC, AC

Backed Teflon membrane, Gelman (Ann Arbor, MI).

Gravimetry, XRF, PIXE, INAA, AAS, ICP/AES, IC, AC

IC, AC

Cellulose fiber, Whatman (Hillsboro, OR), # 1441047

Nylon membrane, Gelman (Ann Arbor, MI), # 66509

F2996-25

“Zefluor”,

Compatible Analysis Methods”

(continued)

Filter Type. Major Manufacturer, and Example

Tab. 2.

0

0

0

0

0

0

0

0

0

0

0

0

0

0

Thin membrane of pure nylon. White opaque surface, diffusion transmitted light. 1 pm pore size. Melts at -60 ‘ C . High flow resistance.

Thick mat ofcellulose fibers, often called a “paper” filter. White opaque surface, diffuses transmitted light. Low sampling efficiencies, < 70% for some variations of this filter. Burns at elevated temperatures (exact temperature depends on nature of particle deposit). Low flow resistance.

Thin membrane mounted on thick polypropylene backing. White opaque surface, diffuses transmitted light. Melts at -60 ‘C. High flow resistance.

Physical Characteristics

~~

0

0

0

0

0

0

0

0

0

0

0

0

Low blank weight.

so,.

High HNO, collection efficiency. Adsorbs low levels of NO, NOz, PAN, and

Low blank levels. Made of carbon-based material, so inappropriate for carbon analysis. Absorbs gases, especially water vapor. Most appropriate for absorbing gases such as HNO,, SO,, NH,, and NO2 when impregnated with reactive chemicals. High blank weight.

Usually low blank levels. Made of carbonbased material, so inappropriate for carbon analysis. Inert to adsorption of gases. Higher background levels lor XRF and PIXE than TefloR owing to greater filter thickness. Low hygroscopicity. High blank weight.

Chemical Characteristics

~

-

C

F c,

o\

W

Gravimetry, light transmission, XRF, PIXE, INAA, AAS, ICP/AES, IC, AC

Glass fiber, Gelman (Ann Arbor, MI), Type AIE

Analysis methods are identified in Tab. 1

Gravimetry

Teflon-coated glass fiber, Pallflex (Putnam, CT), TX40HI20

a

XRF, PIXE electronmicroscopy

Polycarbonate capillary pore membrane, Nuclepore (Pleasanton, CA), # 111129.

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

Borosilicate glass fiber. White opaque surface, diffuses transmitted light. Melts at -500 "C Low flow resistance.

-

Thick mat of borosilicate glass fiber with a layer of Teflon on the surface. Glass fiber supporting Teflon is shiny. Melts at 500 "C Low flow resistance.

Smooth, thin, polycarbonate surface with straight through capillary holes. Light green surface, nearly transparent. Minimal diffusion of transmitted light. Low sampling efficiencies, < 70% for some larger pore sizes. Retains static charge. 0.2,0.3,0.4, and 8.0 pm uniform pore sizes. Melts at -60 'C. Moderate flow resistance.

0

0

0

0

0

0

0

0

0

0

0

High blank levels. Absorbs HNO,, NO,, SO,, and organic vapors. Low hygroscopicity. High blank weight.

Low hygroscopicity. High blank weight.

so,.

Low blank levels for ions. (Glass backing and carbon content make it less suitable for elemental and carbon analyses. Inert to adsorption of HNO,, NO,, and

Low blank levels. (Made of carbon-based material, so inappropriate for carbon analysis.) Low hygroscopicity. Low blank weight.

138

-

J . G. W u m n und J . C . Chow

Particle sampling effeciency: The filter must collect most of the particles in the air being drawn through it, regardless of particle size or flow rate. Some looselywoven fiber filters or membrane filters with large pores allow an excessive fraction of particles to pass through the filter with the air stream. P M , rcference methods (Federal Register, 1987b) specify sampling efficiencies in excess Of 99%, measured with 0.3 pm diameter D O P (dioctyl phthalate) particles. Lippmann (1989) and Lee and Ramamurthi (1993) tabulate the sampling efficiencies for several filter media with different pore size and flow rates. Most filter materials allow less than 1% of the submitted particles to be transmitted, regardless of particle size. The exceptions are cellulose-fiber filters and etched polycarbonate-membrane filters, which have efficiencies lower then 50% for some porosities, pore sizes, and particle sizes. Lower porosities and pore sizes generally result in higher sampling efficiencies.These characteristics also increase flow resistance, however. Flow resistance and loading capacity: Flow resistance refers to the amount of pressure drop across a filter required for a given flow rate. The larger the pressure difference required for a constant flow rate, the greater the flow resistance. As the filter acquires particles, its flow resistance increases because pores and fibers become clogged. Lippmann (1989) and Lee and Ramamurthi (1993) tabulate these resistances for several different filter types. Membrane filters generally have higher flow resistances and lower loading capacities than fiber filters. Lower resistances and higher capacities can be attained by increasing the filter size, increasing the porosity or pore size, increasing the number of pores (in a membrane filter), and decreasing the filter thickness. Decreased flow resistance is often gained at the expense of decreased sampling efficiency. Blank concentrations: The filter material should not contain significant and highly variable concentrations of the chemicals which are being sought by analysis. Each batch of the unexposed filters should be examined for blank concentration levels prior to field sampling. These will be measured as if they were part of the particulate deposit. Cost and availability: Lee and Ramamurthi (1993) provide cost comparisons for different filter materials, but these vary substantially with the filter size, the quantity purchased, and the current competition. Ringed Teflon-membrane filters are typically the most costly ( ~ $ 4 . 5 0for each 47 mm diameter filter), with cellulose-fiber and glass-fiber filters ( - $0.25 for each 47 mm diameter filter) being the least expensive. The cost of the filter is generally a small fraction of the cost of analysis, and the validity of the measurement should not be compromised because one filter is “cheaper” than another. Filters are not always kcpt in stock, and cvcn when thcy arc, the lcad timcs for acccptancc tcsting and preparation can require one month or more. Filters should be procured well in advance of a monitoring program and in sufficient quantity to last the duration of the study.

,

-

-

-

Teflon-membrane and quartz-fiber filters are those which arc most commonly used for the particulate chemical analyses described above, though cellulose-fiber filters lend themselves nicely to impregnation for absorbing gaseous precursors, and etched polycarbonate-membrane filtcrs are best suited for microscopic analyscs (Casuccio et al., 1983).

Measurements on Filters

139

Filters require treatment and representative chemical analyses before they can be used (Chow, 1987). Excessive blank levels and filter interferences discovered during or after the completion of several important air quality studies have compromised their results. At least one filter from each lot (typically 100 filters) received from the manufacturers should be analyzed for all species to verify that pre-established specifications have been met. Lots should be rejected if they do not pass this acceptance test. Each filter should be individually examined over a light table prior to use for discoloration, pinholes, creases, or other defects. In addition to laboratory blanks, 5% to 10% of all filters should be designated as field blanks to follow all handling procedures except for actual sampling. I n addition to acceptance testing, some filters require pretreatment before sampling. As noted above, quartz-fiber filters may adsorb organic vapors with time. Blank quartz-fiber filters should be heated in air for at least three hours at -900 "C prior to acceptance testing analysis. Sets of filters with levels exceeding 1.5 pg/cm2 for organic carbon and 0.5 pg/m3 for elemental carbon should be re-fired or rejected. Pre-fired filters should be sealed, and stored in a freezer prior to preparation for field sampling. Nylon-membrane filters absorb nitric acid over time. Extraction efficiency tests have shown that the sodium carbonate ion chromatograph eluent is needed to remove nitrates from the active sites of the nylon filter. Blank nylon-membrane filters should be soaked in 0.015 M sodium carbonate (Na,CO,) for four hours, then rinsed in DDW for 10 minutes, soaked overnight in DDW, rinsed three times in DDW, then dried in a vacuum oven at 60 "C for 5 to 10 minutes. Sets of washed nylon filters with nitrate levels exceeding 0.05 pg per cm2 should be rejected. Pre-washed nylon filters should also be sealed and refrigerated prior to preparation for field sampling. At times, batches of Gelman ringed Teflon-membrane filters have yielded variable (by up to 100 pg/47 mm filter over a few days) blank masses (e.g., Tombach et al., 1987). As the time between manufacture and use increases, this variability decreases. Since Gelman has minimized its long-term inventory of these filters, and is manufacturing them on an as-ordered basis, this variability has been observed with greater frequency, though it is not widely reported. A onemonth storage period in a controlled environment, followed by one week of equilibration in the weighing environment, appears to have reduced the variability to acceptable (within & 15 pg per filter for re-weights of 47 mm and 37 mm diameter filters) levels. In areas with significant secondary aerosol contributions to PM,., or PMIo, Whatman 41 cellulose-fiber filters can be impregnated with gas-absorbing solutions to collect gaseous ammonia (NH,) and sulfur dioxide (SO,). Several impregnation solutions have been used to absorb these gases. These solutions differ with respect to their reactive components and with respect to their formulations. The criteria which must be met by the impregnation solution are: (1) availability of pure reagents; (2) stability of the impregnation solution composition before and after impregnation; ( 3 ) low degree of hazard or toxicity; (4) lack of interferences with other pollutants being sampled or with analytical methods; ( 5 ) minimal effects of environmental factors such as temperature and water vapor content.

140

J . G. Wutson und J . C . Chow

Sulfuric acid (Okita and Kanamori, 1971; Knapp et al., 1986), oxalic acid (Ferm, 1979, Shendrikar and Lodge, 1975; Ohira et al., 1976), phosphoric acid, sodium carbonate (Ferm, 1986), and citric acid (Stevens et al., 1985) have been used as active agents in the sampling of ammonia on a variety of substrates. Citric acid impregnating solutions best meet the criteria described above. Fung (1988) tested the ammonia absorption capacity of Whatman 41 cellulose-fiber filters impregnated with 0.13 pg of citric acid and 0.024 pg of glycerine. These filters absorbed more than 4000 pg of ammonia with better than 99% efficiency. Tests a t temperatures ranging from -20 "C to 25 "C and at high and low relative humidities showed sampling efficiencies for ammonia in excess of 99%. Potassium carbonate or sodium carbonate with glycerine have been used in impregnating filters for sampling of sulfur dioxide, nitric acid, and organic acids (Forrest and Newman, 1973; Johnson and Atkins, 1975; Anlauf et al., 1985; Daum and Leahy, 1985; Hering et al., 1993; Tanner et al., 1993). The carbonate in the impregnating solution presents interferences to both the ion chromatographic (IC) and automated colorimetric (AC) analyses of extracts from these filters, however. In ion chromatography, the carbonate interferes with the nitrate peak and broadens the sulfate peak. In colorimetric methylthymol-blue analysis, the reaction of the MTB-Ba complex needs to be acidic and the carbonate raises the pH. Triethanolamine (TEA) has been used as an absorbent for nitrogen dioxide (NO,) and to measure aerosol acidity (Dzubay et al., 1979). When used as a solution in a bubbler, TEA is a US. EPA equivalent method (No. EQN-1277-028) for monitoring nitrogen dioxide. Alary et al. (1974),Ohtsuka et al. (1978), Gotoh (1980), and Knapp et al. (1986)have applied TEA solutions to filter media such as Whatman 3 1 chromatographic paper for the collection of nitrogen dioxide. The TEA is usually mixed with glycol or glycerine to improve its absorbing capacity (Doubrava and Blaha, 1980). Peroxyacetyl nitrate (PAN), organic nitrates, and sulfur dioxide are also collected by this substrate, and the nitrogen-containing compounds will appear as nitrate during analysis. TEA oxidizes in air and light, so impregnated filters must be stored in the dark in sealed containers.

6.4 Aerosol Sampling Systems As noted in the introduction, sampling systems for particles and gases have been designed and applied for a number of purposes. Most of the comercially-available samplers, however, have been oriented toward the measurement of P M l o mass to determine compliance with U S . ambient air quality standards. These samplers must pass several requirements (Federal Register, 1987b) to be qualified as reference o r equivalent P M measurement methods. These requirements include wind-tunnel testing of the P M l o inlet, collocated testing, and reliability measures. Tab. 3 identifies the sampling systems which have passed these test and their status as reference or equivalent methods. Hering (1 989) and Perry (1989) describe several other commercially-available systems for ambient aerosol sampling.

Measurernents on Filters

141

Tab. 3. P M , , Reference and Equivalent Sampling Methods Designation No.

Sampler Description

Reference/ Equivalent Method”

Federal Register Citation

Notice Date

RFPS- 1087-062

Wedding & Associates’ PM Critical Flow High-Volume Sampler (using Wedding PM inlets and Wedding & Associates’ critical flow device).

Reference method

Vol. 52, 37366

10/06/87

RFPS-I 287-063

Refcrence Sierra-Andersen (SA) method or General Metal Works (GMW) Model 1200 PM,, High-Volume Air Sampler System (using SA- or GMW1200 PM,, Size-selective inlets and any of the following air samplers: SAUV-IOH, SAUV11H, GMW-IP-10, G M W-I P- 10-70, G M W-lP-10801, GMW-IP-10-8000).

Vol. 52, 45684 Vol. 53, 1062

12/01/87 01/15/88

RFPS-I 287-064

Referencc Sierra-Andersen or General method Metal Works Model 321-B PM High-Volume Air Sampler System (using SA- or GMW-321.B PM,, size-selective inlets and any of the following air samplers: SAUV-lOH, SAUV-I IH, G M W-IP-10, G M W-TP- 10-70, GMW-IP-10-801, GMWIP- 10-8000).

Vol. 52, 45684 Vol. 53, 1062

12/01/87 Ol/l5/88

RFPS- 1287-064

Reference Sierra-Andersen or General method Metal Works Model 321-C PM,, High-Volume Air Sampler System (using SA- or GMW321-C PM,, sizeselective inlets and any of the following air samples: SAUV-IOH, SAUV-IlH, G M W-IP-10, G M W-IP- 10-70, G M W-IP-10-801, G M W-IP10-8000).

Vol. 52, 45684 Vol. 53, 1062

12/01/87 01/15/88

RFPS- 1287-064 (continued)

RFPS-0389-071 Orcgon D E Q Medium-Volumc PM,, Sampler (using SA-254 PM,, inlet, and 47 mm Teflonmembrane and quartz-fiber filters).

Reference method

Vol. 54, 12273

03/24/89

,,

,,

142 Tab. 3.

J . G. Watson and J . C. Chow

PM,, (continued)

Designation No.

Sampler Description

Reference/ Equivalent Method"

Federal Register Citation

Notice Date

RFPS-0389-073

Sierra-Andersen Models SA-241 and SA-241M or General Metal Works Models G241 and GA-241M PM,, Dichotomous Samplers (using SA-246B or G246 PM,, inlet, 2.5 pn virtual impactor assembly, and 37 mm PM, and coarse [PM minus PM, filter holders).

Rcference mcthod

Vol. 54, 31 247

07/27/89

,,

EQPM-0990-076

Andcrsen Instruments Equivalent Model FH621-N PM,, Beta mcthod Attenuation Monitor (using SA246b PM,, inlet, FHlOl vacuum pump assembly, FH102 acccssory kit, FH107 roofflangc kit, and FH125 7ero and span PM mass foil calibration kit).

Vol. 55, 38387

09/18/90

EQPM- 1090-079

Rupprecht & Patashnik TEOM Equivalent Series 1400a PM,, Monitor mcthod (using SA-246b PM inlet and internal tapered element oscillating microbalance).

Vol. 55, 43406

10/29/90

,"

a

Code of Federal Rcgulations (1988). (40CFR, Part 50, Appendix J)

Watson and Chow (1993) observed that commercial systems are not entirely adequate for many sampling programs, though well-tested and commercially-available size-selective inlets, sampling manifolds, filter holders, flow controllers, and pumps can often be assembled into cost-effective and reliable systems which are tailored to specific measurement objectives. Several of these systems, with references describing them and the projects in which they were applied, are identified in Tab. 4.

6.4.1 Size-Selective Inlets Size-selective inlets arc needed to define the particle size fraction being sampled. Air is drawn through these inlets to remove particles which exceed a specified aerodynamic diameter prior to exposure of the filter to the air stream. Inlets are charaterized by sampling effectiveness curves showing the fraction of spherical particles of unit density which penetrate the inlet (Watson et al., 1983). This sampling effectiveness is summarized by a 50% cut-point, d5", the diameter at which half of the particles pass through the inlet and the other half are deposited in the inlet. This cut-point changes for different flow rates, so every inlet requircs a constant

Southern California Air Quality Study Sampler (SCA Q3' 1

California Institute of Technology Sampler

SA-254 mediumvolume inlet (PM,,) 35 out of 113

22

AIHL cyclone

47 mm Teflon membrane 47 mm quartz fiber 47 mm nylon membrane

Gelman stainless steel in-line Geman stainless steel in-line

Tefloncoated aluminum and glass Stainless steel and aluminum

47 mm Teflon membrane 47 mm quartz Fiber

47 mm Teflon membrane 47 mm quartz fiber

Gelman stainless steel in-line

Stainless steel and aluminum

47 mm Teflon membrane 47 mm quartz fiber

Nuclepore polycarbonate open-face

Stainless steel and alumin um

113 out of 1,130

16.7

47 mm Teflon membrane 47 mm quartz fiber

47 mm Teflon membrane 47 mm quartz fiber

Nudepore polycarbonate open-face

Nuclepore polycarbonate in-line

Aluminum and copper

113

47 mm Teflon membrane 47 mm quartz fiber

Aluminum and polyvinyl chloride

Nuclepore polycarbona te in-line

Aluminum and copper

113 out of 1,130

Filter Media

113 out of 1,130

Filter Holders

Sampling Surface

Flow Rate (Ljmin)

SA-246 low-volume inlet (PM,,)

Size Classifying SA-320 high-volume Isokinetic inlet (PMI5) Sequential Aerosol Sampler (SCISAS) Bendix 240 cyclone (PM2.5)

Western Region SA-320 high-volume Air Quality inlet (PMI5) Study Sampler (WRAQS) Bendix 240 cyclone

Inlet and Particle Size (pm)

Examples of Size-Classified Ambient Aerosol Sampling Systems

Sampling System

Tab. 4.

So1om on et al., 1989

Rogers et al., 1989

Tombach et al., 1987

Descriptive Reference

Option to Fitz et al., add 20 cm flow 1989 homogenizer Wolff et al., 1991

Sequential sampling

Features

W P

c

2 2

2

-

0

$

2

T

52

Stacked Filter Unit (SFU)

Modified Sequential Filter Sampler (SFS)

Interagency Monitoring of Protected Visual Environments (IMPROVE)

35 out

Bendix 240 cyclone (PM 2 . 5 )

Tefloncoated aluminum

Polycarbonate

20 out of 113

10

Large-pore etched polycarbonate filters (2-3 pm)

(PM2.5)

Bendix 240 cyclone

Nuclepore pol ycarbonate open-face

Nuclepore pol ycarbonate open-face

Aluminum

20 out of 113

D)

24

Modified AIHL cyclone (PM,.,)

SA-254 mediumvolume inlet (PM

Nuclepore polycarbonate open-face Nuclepore polycarbonate open-face

Aluminum

18

Nuclepore polycarbonate open-face

quartz fiber nylon membrane Teflon membrane quartz fiber

47 mm Teflon membrane 47 mm quartz fiber 47 mm nylon membrane 47 mm impregnated cellulose fiber 47 mm etched polycarbonate membrane 47 mm Teflon membrane

25 mm 47 mm 47 mm 47 mm

25 mm Teflon membrane

Eldred et al., 1990

Descriptive Reference

Uses large-pore Cahill et al., etched 1990 polycarbonate filters as PMz.5 inlet

Option to Chow et al., add nitric acid 1993d denuders in the sampling stream Sequential sampling

Nitric acid denuders can be placed in inlet line

Option to add 20 cm flow homogenizer

47 mm Teflon membrane 47 mm quartz fiber 47 mm impregnated quartz fiber 47 mm nylon membrane 47 mm etched polycarbonate 25 mm teflon membrane 25 mm quartz fiber

Gelman Stainless steel and Savillex PFA Teflon in-line

Tefloncoated aluminum and Teflon

Aluminum

Features

Filter Media

Filter Holders

Sampling Surface

Wedding low-volume inlet (PM,,)

of 113

Flow Rate (L/min)

Inlet and Particle Size (pm)

(continued)

Sampling System

Tab. 4.

47 mm Teflon membrane 47 mm etched polycarbonate membrane 47 mm quartz fiber

37 mm Teflon membrane 37 mm impregnated quartz fiber

SierraAndersen open-face ring

Glass

Sierra37 mm Teflon membrane Teflon37 mm nylon membrane coated glass Andersen open-face rings

10

4

Teflon-coated New York University glass (PM2.4 Medical Center Sequential Acid Aerosol Sampling System(NYUMC/ SAASS)

Teflon-coated 33 rain cap (PM,,) and virtual impactor (PM,.,)

47 mm Teflon membrane 47 mm nylon membrane

Savillex PFA Teflon open-face University Research Glassware open-face

PFA Tefloncoated aluminum Tefloncoated aluminum

20 out of 113

47 mm Teflon membrane 47 mm impregnated cellulose fiber

Savillex open-face

Aluminum

20 out of 113

Harvard/EPA Glass impactor Annular (PM2.5) Denuder System (HEADS)

Versatile Ambient Particulate Sampler (VAPS)

California Acid SA-245 mediumDeposition volume inlet (PM,,) Monitoring Program Dry Deposition Sampler (CADMP) Teflon-coated Bendix 240 cyclone (PM,,,)

Chow ct al., 1993c

Includes annu- Stevens et al., lar denuders 1993 to capture nitric acid, nitrous acid, and sulfur dioxide; and polyurethane foam (PUI) to collect organic compounds Includes sodium Koutrakis carbonate et al., 1988; coated denuders 1991; 1992 to collect acidic gases (e.g., nitric acid, nitrous acid, sulfur dioxide, organic acids) and citric acid coated denuders to collect ammonia Sequential Thurston sampling et al., 1992

lncludes nitric acid denuders Sequential sampling

0

146

J . G. Wutson and J . C. Chow

flow to retain a given 50% cut-point. Hering (1989) identifies several available inlets with various particle sizing characteristics. For P M tested inlets are available for high volume ( - 1000 Ljmin) sampling (Wedding et al., 1977; McFarland et al., 1980; Wedding and Weigand, 1985), medium volume ( 100 Ljmin) sampling ( O h and Bohn, 1983; Wedding et al., 1983),and low volume ( N 20 Ljmin) sampling (Wedding et al., 1982; McFarland and Ortiz, 1984). For PM2,5,tested inlets are also available for high volume (Willeke et al., 1975), medium volume (Chan and Lippmann, 1977; Watson, 1979; Mueller et al., 1983); and low volume (McFarland et al., 1978; John and Reischl, 1980; John et al., 1983a, b) sampling. Several inlets can be placed in a series, in the form of a “cascade impactor”, to obtain more detailed size distributions of chemical concentrations (Berner et al., 1979; Hering et al., 1979a, b; Marple et al., 1981; Raabe et al., 1988). These inlets must be regularly cleaned to prevent the resuspension and transmittal of the large particles they have collected. John and Wang (1991) proved that reentrainment and disaggregation of particles occurs on the impaction surface of highvolume impactor-type inlets, even though these surfaces are often oiled or greased to assure specified sampling effectiveness. Woods et al. (1986) conducted extensive wind tunnel tests on clean and dirty, greased and ungreased high-volume P M l o inlets, which showed changes in d50 as the inlets become dirtier. The deposits apparently affect the size-selective properties of the inlets when they are not regularly cleaned. N

6.4.2 Sampling Surfaces Most samplers are manufactured from aluminium, plastic, or galvanized steel owing to their availability and economy. These materials can absorb some gases (Hering et al. 1988; John et al., 1986), especially nitric acid, which can change the equilibrium of volatile particles on a filter with the surrounding air. When this happens, volatile particles (e.g., ammonium nitrate) may evaporate, thereby biasing the result of the measurement. John et al. (1986) show that surfaces coated with perfluoroalkoxy (PFA) Teflon can pass nitric acid with 80% to 100% efficiency. They also show that the aluminium surfaces common to many samplers and inlets have an almost infinite capacity for absorbing nitric acid vapor while transmitting particles with high efficiency. Several of the samplers in Tab. 4 have applied the coating methods described by John et al. (1986) when gases and particles are sampled. Plastic surfaces can acquire an electrical charge which might attract suspended particles, though the dimensions of most ambient sampling systems are sufficiently large that this attraction is negligible (Rogers et al., 1989).

6.4.3 Filter Holders Wherever possible, filters should be loaded into and unloaded from filter holders in a clean laboratory environment rather than in the field. This minimizes filter contamination and protects the samples during transport. Lippmann (1989) and

Measurements on Filters

147

Watson and Chow (1993) describe several different types of filter holders which are used in aerosol sampling. Filter holders are configured as open-faced, with no constrictions upstream of the filter surface, or in-line, with a small diameter pipe opening into a small chamber into which the filter is mounted. In-line holders often concentrate the particles in the center of the substrate, and this will bias the results if analyses are peformed on portions of the filter. Tombach et al. (1987) and Fujita and Collins (1989) show differences as high as 600% between chemical measurements in the middle and at the edges of filters samples with in-line filter holders. Open-faced filter holders are a better choice for ambient aerosol sampling systems. All PM," reference samples use open-face filter holders. Filter holders used for aerosol sampling are made of stainless steel (e.g., Gelman Instrument Co., Ann Arbor, MI), aluminium (e.g., Graseby-Andersen Instruments, Atlanta, GA), polycarbonate ( e g Nuclepore Corporation, Pleasanton, CA), polyvinyl chloride (e.g., Graseby-Andersen Instruments, Atlanta, GA), or molded PFA Teflon (e.g., Savillex, Inc., Minnetonka, MN). The metal holders are not useful for sampling gases on absorbing substrates owing to their potential interaction with these gases. Tab. 4 identifies several of the commercially-available filter holders which have been used in different sampling systems.

6.4.4 Pumps and Flow Controllers Air is passed through the sampling substrates by means of a vacuum created by a pump. Rubow and Furtado (1989) describe commercially available air pumps, their capacities, and operating principles. Rogers et al. (1989) have found that a 3/4 horsepower carbon vane pump is sufficient to draw in excess of 120 L/min through a Teflon-membrane filter with 1 pm pore size. Smaller pumps can be used for lower flow rates and filter media with lower resistances. Pump capacity, filter media, flow controllers, and inlet flow requirements must be specifically matched for each sampling system. Regardless of the pump used, the quantity of air per unit time must be precisely measured and controlled to determine particle concentrations and to maintain the size-selective properties of the sampling inlet. Four general methods of flow control are used in particle sampling systems: (1) manual volumetric; (2) automatic mass; (3) differential pressure volumetric; (4) critical orifice volumetric flow controls. For manual control, a valve is located in the sample line between the filter and the pump. This valve is adjusted until the desired reading is obtained from a low-resistance flow meter attached to the front of the filter. Flow rates which are manually controlled often decrease over a sampling period as the collection substrate loads up and presents a higher flow resistance. For mass flow control, the gas stream in back of the filter is directed over a thermal anemometer which measures the heat transfer between two points in the gas stream; the heat transfer is proportional to the flux of gas molecules. Since heat transfer depends on the density of the gas, which is a function of temperature

I48

J . G. Wutson and J . C. Cliow

and pressure, mass flow is not always equivalent to volume flow. Wedding (1985) estimates potential differences in excess of 10% between mass and volumctric measurements of the same flow rates, depcnding on temperature and pressure variations. For differential pressure volumetric flow control, a constant pressure drop is maintained across an orifice by a diaphragm-controlled valve located between the filter and the orifice (Chow et al., 1993~).The diaphragm is exposed to the pressure difference between the orifice and thc pump. As this pressure difference increases (as it does when filters load up), the diagphragm opens the valve and allows more air to pass. For critical orifice flow control, a small circular opening is placed in the flow line between thc filtcr and the pump. When the downstream pressure is lower than 50% of the pressure upstream of the orifice, the air velocity in the orifice attains the speed of sound and remains constant, even when the filter loads up. Wedding et al. (1987) have developed a “critical throat” which uses a diffuser arrangement to recover much of the energy which is normally expended in back pressure behind a critical orifice. This design allows higher flow rates to be obtained with a given pump than does a simple critical orifice.

-

6.4.5 Sampler Configurations The commercial market for aerosol sampling is largely driven by the need to determine compliance with U.S. ambient air quality standards. Since these standards specify mass concentration within a specitic particle size range (PM J, the majority of the reference and equivalent samplers identified in Tab. 5 are designed for this purpose. They are not entirely adaptable to particle chemical characterization. High-volume PM samplers are the most widely used; these acquire P M l o mass measurements at more than 2000 sites in the U S . on a sampling schedule of every sixth day. These samplers use a low pressure blower to draw air through 20.3 x 25.4 cm fiber filters. Procedures for these samplers are well established (e.g., Watson et al., 1989). Frequent inlet cleaning is necessary for accurate size sampling by these units, and filters must be carefully handled if chemcial analysis is anticipated. Whatman QM/A quartr. fiber filters which have been submitted to acceptance testing can be used in these samplers for many chemical analyses. As noted above, these filters contain large amounts of sodium, aluminium, and silicon, so these specie3 cannot be measured with this system. The thickness of the fiber filter also raises the background i n X R F and PIXE analysis, thereby decreasing the sensitivity of these analyses. The low-volume dichotomous sampler is often used with appropriate filter media when elemental, ionic, and carbon analyses are desired. This sampler separates particles into PM, and coarse (PM,, minus PM2 5 ) size fractions. Flow rates are controlled by a differential pressure regulator. Ten percent of the PM2,5are sampled on the coarse particle filters, and corrections must be made (Evans and Ryan, 1983)

Measurements on Filters

149

to the coarse particle measurements to compensate. Several dichotomous samplers may be collocated to accommodate different filter types. John et al. (1988) describe how dichotomous samplers can be adapted for nitric acid sampling. The sequential filter sampler (SFS) equipped with a medium-volume P M inlet was originally designed in the late 1970s for use in the Sulfate Regional Experiment (SURE, Mueller et al., 1983) and the Portland Aerosol Characterization Study (PACS, Watson, 1979) and has been applied in over a dozen subsequent studies of P M l o and visibility impairment. The SFS consists of an aluminum plenum to which the P M l o inlet is attached. U p to 12 sampling ports within the plenum are controlled by solenoid valves which divert flow from one channel to the next by means of a programmable timer. These ports accept filters which have been pre-loaded into open-faced 47 mm Nuclepore filter holders. The sample flow can be divided for simultaneous collection on two or more filter media. The differential pressure volumetric flow controller splits the flow between filters and maintains a constant flow rate despite filter loading. The SFS is especially useful when less than 24-hour average samples are sought. Two types of continuous filter monitors have achieved equivalence .status for PM monitoring with hourly averages: the Tapered Element Oscillating Microbalance (TEOM) and the Beta Attenuation Monitor (BAM). These monitors have potential for providing samples which can be chemically analyzed. The TEOM (Patashnick and Rupprecht, 1991; Rupprecht et al., 1992) utilizes a replaceable filter, usually Teflon-coated glass-fiber, mounted on the narrow end of a hollow tapered tube. The wider end of the tube is fixed, while the narrow end oscillates in response to an applied electic field. Air is drawn through any desired inlet, then through the filter and the tapered tube, to a flow controller. The filter loading causes a mass change which is detected as a change in the frequency of oscillation of the tapered element. The filter is only about 0.5 cm in diameter. While it might be submitted to chemical analysis, the deposit on it is small and the analytical sensitivity would be low. The filter is usually changed weekly, so analysis for a 24-hour period would not be possible. The TEOM draws air through this filter at a flow rate of 3 L/min. This flow is extracted from a total flow of 16.7 L/min which is drawn through an SA-246 PM,, inlet. The make-up air flow of 13.7 L/min can be diverted through one or more larger filters which can then be submitted to chemical analysis after sampling. There is no reason that a sequential sampling feature, similar to that of the SFS, could not be added to allow sequential 24-hour filter samples, or every sixth day samples, to be taken between weekly maintenance visits. The BAM uses the attenuation of beta rays (moderately high energy electrons) emitted by a radioactive source when they pass through an aerosol filter deposit to indicate the mass of that filter deposit (e.g., Lillienfeld and Dulchinos, 1972; Husar, 1974; Lillienfeld, 1975; Macias and Husar, 1976; Lillienfeld, 1979). Continuous P M monitoring systems consist of a filter tape, which is first drawn across the path between the beta emitter and a detector to measure blank attenuation, then across a sampling area in which ambient air is deposited on the tape, and then across the detection path to measure the combined attentuation of the filter and the deposit. The beta attenuation is caused by the inelastic collision of the incident

,

150

.J. G . Watson and J , C. (Ilio~c.

clcctrons with thc orbital electrons of the atoms in the samples for cnergies less than 1 MeV. The filter spots are approximately 2 inches in diameter, and depending on the filter media used, these might be appropriate for chemical analysis. This has not yet been attempted, however, and development is required before these filter deposits can be considered for chemical Characterization. Though it is not yet designated a PM reference sampler, a mini-volume portable survey sampler (Air Metrics, Springfield, OR) has been developed which can be deployed in spatially-dcnse P M sampling networks. These battery-powercd units can be hung from power poles and building walls and do not require complicated sampler siting efforts (Watson et al., 1991b). They can be placed in and around fugitive dust sources to help locate the most significant contributors to high P M caused by dust cmissions. The portable survey samplers consist of a pump, timer, tubing and fittings, removable filter holder, flow meter, impactor inlet, and battery pack. All of these are packaged in a plastic cylindrical enclosure which is about 25 cm in diameter and 50 cm long. A carrying bale allows the sampler to be hung from a hook or hanging bracket. The sampler weighs about 15 pounds, most of which is the battery. It can be located and removed from elevated locations with a grappling pole. Substantially greater flexibility for sampling and analysis exists when thcrc is no need to determine compliance with with P M t o standards, and Tab. 4 identifies several sampling systems which have been designed for specific aerosol monitoring purposes. Watson and Chow (1993) provide brief descriptions for most of the samplers identified in Tab. 4, and the cited references provide detailed hardware and performance specifications. Several of the samplers identified in Tab. 4 use denuders and absorbing filters to acquire gases as well as particles. Denuders are tubes with gas-absorbing surfaces which are placcd between the inlet and the filters. Gases have much higher diffusion coefficients than particles, so they diffuse to the denuder surfaces while the particles pass through the denuder to the filter. On some of these denuder systems, nitric acid, sulfur dioxide, and other gases are absorbed on the inner surfaces of the denuder inlet and arc removed for analysis by washing with an extraction solution (Stevens ct al., 1990).

,

6.5 Sampling and Analysis Procedures The selection of appropriate analysis methods, filter media, and sampling hardware must be complemented with detailed sample handling and analysis procedures if accurate, precise, and valid measurements are to be attained. Fig. 1 shows a flow diagram of the different operations which are applied in a typical aerosol characterization monitoring program. Each box represents a set of actions which must be taken as part of the overall measurement process. The following phenomena have been found to bias particle measurements in the past:

Measurements on Filters

Acceptance Testing

Teflon

Transmittal to Laboratory

Post-Sampling Gravimetry and Babs

Disassemble Filter Packs

IfrC03 Impreg.

Citric Acid Impreg.

fq

I

1

I

Mass and Babs Concentration Calculations

Refrigerated Storage

Selection for Chemical Analysis

05cm Quartz

Quartz

X-Ray Fluorescence (40 Elements1

Washing

Pro-firing

112 Filter

Optical

in

Refiectance

l5mlDDW

(OH. OC, EH, EC)

1

Autometed Colorimetry ( N H ~ Atomic Absorption (Naf K9 ion Chromatography (CI ;NO;. SOTI

151

Assurance

Nylon

112 Filter Extracted in 10 mi

Kfio3

4C03 Cellulose

Extract in 10 ml 0.1% q 2 and 1:11 DDW Dllutlon

I

Ion

Ion

Chromatography IHNO3Igl NO$

Chromatography (S02)

Citric Acid Cellulose

Whole Filter Extracted in 1 0 ml Citrlc Acid

1

Automated Colorimetry INH3)

I

Fig. 1. Flow diagram for filter proccssing and chemical analysis activities for a particlc/gas sampling system.

I52 -

-

-

-

-

-

J . G. Wditson und J . c‘. C‘lioiv

Passive deposition: Dust can deposit on filters prior to and following sampling. This bias can be evaluated by placing “dynamic blank” filters through which no air is drawn in the sampler along with the sampled filters and submitting them to the same chemical analyses. It can be minimized by using samplers which do not expose the filters directly to the atmosphere and by minimizing the passive time in the sampler before and after sampling. Re-entrainment: Large particles collected in the size-selective inlets may bc resuspended in the air stream for sampling on the filter. Frequent cleaning and greasing of inlet impaction surfaces will minimize biases to chemical concentrations. Recirculation: Pump exhaust contains fragments of its brushes and armatures (Countess, 1974). Most high-vacuum pumps have outlet filters which should be installed and changed at least quarterly. New pumps should be broken in for at least 48-hours, and the exhaust filter should be replaced, prior to taking the first sample. A piece of clothes-dryer duct can be attached to the exhaust pipe to direct pump exhaust away from the sampler inlet. Volatilization: Chemical compounds which change particle/gas equilibrium in different environments may lose ammonium nitrate and certain organic compounds (Witz and Wendt, 1981; Witz et al., 1990).The most accurate monitoring of these species involves denuder-type sampling systems. Volatilization can be minimized by removing samples soon after sampling, storing them in sealed containers under refrigeration, and keeping them in coolers for transport between the sampling site and laboratory. Particle loss during transport: Particles can fall off filters when samples have large deposits and receive rough handling during movement from the field site to the laboratory (Dzubay and Barbour, 1983). Shorter sample durations and lower flow rates may be required in very polluted environments, especially those in which fugitive dust is a large contributor, to prevent overloading. Careful handling during transport will also minimize the loss of particles from the filter surface. Filter contamination: Various chemicals can be inadvertently introduced to the filter during handling. This can be minimized by acceptance testing a few blank filters from each manufacturing lot to assure that handling prior to receipt has not caused contamination. Subsequent contamination can be reduced by loading and unloading filters in a laboratory setting, keeping them in containers before and after sampling, and eliminating contact of the filters with bare hands.

6.6 Summary Particle sampling on filters is the most practical method currently available to characterize the sizes, chemical compositions, and precursor gas concentrations for ambient aerosol. Ambient aerosol sampling systems consist of a combination of monitoring hardware, filter media, laboratory methods, and operating procedures

Measurements on Filters

I53

which are specifically tailored to different monitoring objectives. No single sampling system can meet all needs, and it is often necessary to adapt existing sampling components to the specific situation being studied. Examples of successful sampling systems which can be copied or modified to meet these specific needs have been identified.

6.7 References Alary, J., P. Bourbon, P. Chevin, C. Delaunay, J. Escalassan, and J. D. Lepert (1974) New Method of Determination of Nitrogen Dioxide in Polluted Atmospheres Derived from the GriessSaltzmann Method. Water, Air, Soil Pollut. 3, 555 (in French). Anlauf, K. G., P. Fellin, H. A. Wiebe, H. I. Schiff, G . I. Mackay, R. D. Braman, and R. Gilbert (1985) A Comparison of Three Methods for Measurement of Atmospheric Nitric Acid and Aerosol Nitrate and Ammonium. Atmos. Enuiron. 19 (2), 325-333. Appel, B. R., P. Colodny, and J. J. Wesolowski (1976) Analysis of Carbonaceous Materials in Southern California Atmospheric Aerosols. Enuiron. Sci. Technol. 10, 359. Appel, B. R., E. M. Hoffer, E. L. Kothny, S. M. Wall, M . Halik, and R. L. Knights (1979) Analysis of Carbonaceous Material in Southcrn California Atmospheric Aerosols-2. Environ. Sci. Technol. 13, 98. Appel, B. R. (1993) Atmospheric Sample Analysis and Sampling Artifacts. In: Aerosol Measurement: Principles, Techniques and Applications, K. Willeke and P. A. Baron (eds.). Van Nostrand, Reinhold, New York, NY, pp. 233-259. Berner, A,, C. H. Lurzer, L. Pohl, 0. Preining, and P. Wagner (1979) The Size Distribution of the Urban Aerosol in Vienna. Sci. Total Environ. 13, 245-261. Butler, F. E., R. H. Jungers, L. F. Porter, A. E. Riley, and F. J. Toth (1978) Analysis of Air Particulates by Ion Chromatography: Comparison with Accepted Methods. In: /on Clwomatographic Analysis of Environmental Pollutants, E. Sawicki, J. D. Mulik, and E. Wittgenstein (eds.). Ann Arbor Science Publishers, Inc., Ann Arbor, MI, pp. 65-76. Cadle, S. H. and P. J. Groblicki (1982) An Evaluation of Methods for the Determination of Organic and Elemental Carbon in Particulate Samples. I n : Particcilute Carbon: Atmospheric Lijk Cycle, G. T. Wolf and R. L. Klimisch (eds.). Plenum Press, New York, pp. 89- 109. Cadle, S. H., P. J. Groblicki, and D. P. Stroup (1980a) An Automated Carbon Analyzer for Particulate Samples. A n d . Chem. 52, 2201 -2206. Cadle, S. H., G. J. Nebel, and R. L. Williams (1980b) Measurements of Unregulated Emissions from General Motors’ Light-Duty Vehicles. Paper 790694, Society of Automative Engineers Trunsactions 87, 238 1 - 2401. Cahill, T. A,, R. A. Eldred, P. J. Feeney, P. J. Beveridge, and L. K. Wilkinson (1990) The Stacked Filter Unit Revisited. In: Transactions, Visibility and Fine Particles, C . V. Mathai (ed.). Air & Waste Management Association, Pittsburgh, PA, p. 213. Casuccio, G . S., P. B. Janocko, R. J. Lee, J. F. Kelly, S. L. Dattner, J. S. Mgebroff (1983) The Use of Computer Controlled Scanning Electron Microscopy in Environmental Studies. J . Air Pollution Control Assoc. 33, 937 - 943. Chan, T. and M. Lippmann (1977) Particle Collection Efficiencies of Sampling Cyclones: An Empirical Theory. Enuiron. Sci. Technol. 11 (4), 377. Charlson, R. J., S. E. Schwartz, J. M. Hales, R. D. Cess, J. A. Coakley, Jr., J. E. Hansen, and D. J. Hoffmann (1992). Climate Forcing by Anthropogenic Aerosols. Science 255, 423 -599. Charell, P. R. and R. G. Hawley (1981) Characteristics of Water Adsorption on Air Sampling Filters. Am. Ind. Hyg. Assoc. J . 42, 353. Chow, J. C. (1987) Inorganic Analysis Methods. In: Pacific Northwest Source Profile Library: Suurce Sumpling and Analytical Protocols, J. Core and J. Houck (eds.). Oregon Department of Environmental Quality, Portland, OR.

154

J . G. Wiztson uncl .I. C'.

C/101t,

Chow, J. C.,J. G . Watson,D. M . Ono,andC. V. Mathai (1993a)PM,,Standardsand Nontraditional Particulate Source Controls: A Summary of the A & WMA/EPA International Specialty Conference. J . Air Wuste M o n u g ~ Assnc,. . 43, 74- 84. Chow, J. C., J . G. Watson, L. C. Pritchctt, W. R. Pierson, C. A. Frazicr, and R. G. Purccll (3993b) The DRI Thcrmal/Optical Reflectance Carbon Analysis System: Description, Evaluation and Applications in U.S. Air Quality Studics. Atmos. Enuiron. 27A, 1185- 1201. Chow, J. G., J . Ci. Watson, J . L. Bowcn. A. W. Gertler, C. A. Frazier, K. K. Fung, and L. Ashbaugh ( 1 9 9 3 ~ A ) Sampling System for Reactive Species in the Western U.S. I n : Sumpling unil Anci1,vsis oJ' A irhorrw Pollrrtcitits, E. Winegar. (ed.) American Chemical Society, Washington. DC', pp. 209 - 228. Chow, J. C., J. G. Watson. D. H . Lowenthal, P. A. Solomon, K. Magliano, S. D. Ziman. and L. W. Richards (1993d) Planning Tor SJVAQS/AUSPEX Particulate Matter and Visibility Sampling and Analysis. I n : Plmning und Manriging Air Quu1it.v Modeling rrritl Meu.surCni(wt.s Stu(1ie.v: A Puywctii)e Tlrrough SJVAQS/ACISPEX, P. A. Solomon and T. Silver (eds.). Air 62 Waste Management Association, Pittsburgh. PA, i n press. Code of Federal Regulations (1988). 40 CFR Part 50, Appendix J . Countess, R. J. (1974) Production of Aerosol by High Volume Samplers. J . Air Po//. Control Assor. 24, 605. Coutant, R. W. (1977) Effect or Environmental Variables on Collection of Atmospheric Sulfate. Etwiron. Sri. Trriinol. 11. 873. Daisey. I. M., R. J. McCaffrcy, and R. A. Gallogher (1981) Polycyclic Aromatic Hydrocarbons and Total Extractable Particulate Organic Matter in the Arctic Aerosol. Atnios. Enniron. 15, I353 - 1363. Dams, R., J. A. Robbins, K. A. Rahn, and J. W. Winchester (1970) Non-Destructive Neutron Activation Analysis of Air Pollution Particulates. A n d . Chetti. 42, 861. Daum, P. 11. and D. F. Lcahy (19x5) The Brookhaven National Laboratory Filter Pack System for Collection and Determination of Air Pollutants. Brookhavern National Laboratory, Upton, NY. Demuynck, M. (1975) Dctermination of Irreversible Absorption of Water on Air Sampling Filters. Ani. Itid. Hj'g. A.s.voc. .J. 42, 353. Dod, R. L., H . Roscn, and T. Novakov (1979) A~niosplirricAerosol Rescwch Annual R ~ p o r tfiir 1977- 1978. Document #LBL-8696, Lawrence Berkeley Laboratory, Berkeley, CA. Doubrava, J., and F. Blaha (1980) Mixture for Impregnation of Air Filters. Czochoslotwkiti Ptrtcnr, C S 182164. Dzubay, T. G. and K. 0. Nelson (I975) Self A bsorption Corrections for X-Ray Fluorcscencc Analysis of Aerosols. Aduivtrrrczs in X - Ruy Anu/ysi.s 18, 61 9. Dzubay, T. G. and R. K . Stevens (1975) Ambient Air Analysis with Dichotomous Sampler and X-Ray Fluorescence Spectrometer. Ennrirori. .%i. Tc.cifno/.9, 663 - 667. Dzubay, T. G., G. K . Snyder, D. J . Reuttcr, and R. K . Stevens (1979) Aerosol Acidity Dctcrmination by Reaction with I4C Labelled Atnine. A m o s . Ennciron. 13, 1209- 1212. Dzubay, T. G. and R. K . Barbour (1983) A Method to Improve the Adhesion of Aerosol Particles on Teflon Filters. JAPCA 33, 692. Eatough, D. J . , N. Aghdaic, M. Cottam, T. Gammon, L. D. Hansen, E. A. Lewis, and R (1990) Loss o f Semivolatile Organic Compounds from Particles During Sampling on Filters. In: Trtoisrrr~tiot~.~: Visihilitj untl Fine P u r t i i ~ l ( ~C. , ~ V. , Mathai (ed.). Air & Waste Management Association. Pittsburgh, PA, pp. 146- 156. Eldred, R. A , , T. A. Cahill, L. K . Wilkinson, P. J . Fccney, J. C. Chow, and W. C. Malni (1990) Measurement of Fine Particles and Their Components in the NPS/IMPROVE Network. I n : Tran.snrtions: Visihilifj, ~ n i d fin^ Pivtirtirles, C . V. Mathni (ed.). Air & Waste Management Association, Pittsburgh, PA, pp. 187- 196. Eldred, R. A. ( I 993) Personal Communication. Crockcr Nuclear Laboratory. University of California, Davis, CA. Engelbrecht, D. R., T. A. Cahill, and P. I. Fccney (1980) Electrostatic Effects on Gravimetric Analysis of Membrane Filters. J . Air Poll. Control A,WJC.30, 391 -392. I)vans. J . S. and 1'. B. Ryan (1983) Statistical Uncertainties in Aerosol Mass Concentrations Measured by Virtual Impactors. A i w s o l Sri. Tc,c/inol.2, 53 1 - 536.

Meusuremeiits on Filters

I55

Fassel, V . A. and R. N. Kniseley (1974) Inductively Coupled Plasma-Optical Emission Spectroscopy. A n d . Chem. 46, 1 I 10. Federal Register (19874 Revisions to the National Ambient Air Quality Standards for Particular Matter: 40 C F R parts 51 and 52. Federul Register 52, 24634. July 1 . Federal Register (1987b) Ambient Air Monitoring Reference and Equivalent Methods: 40 C F R part 53. Federul Register 52, 24724, July 1. Ferm, M . (1979) Method for Determination of Atmospheric Ammonia. Atmos. Enuiron 13, 1385- 1393. Ferm. M. (1986) A NaC0,-Coated Denuder and Filter for Determination or Gaseous HNO, and Particulate NO3 in the Atmosphere. Atmos. Enuiron. 20, 1193- 1201. Fernandez, F. J. (1989) Atomic Absorption Spectroscopy. In Methods q f A i r Sumpling and Analysis, 3rd Ed., J . P. Lodge, Jr., Ed. Lewis Publishers, Chelsca, MI, pp. 143- 150. Ferris, B. G . Jr., F. E. Speizer, J. D. Spcngler, D. W. Dockery, Y . M. Bishop, J. M. Wolfson, and C. Humble (1979) Effects of Sulfur Oxides and Respirable Particles on Human Health. American Review qf‘ Respiratory Diseuse 120, 767 - 779. Fitz, D., M. Chan, G. Cass, D. Lawson, and L. Ashbaugh (1989) A Mtilii-Component SizeClus,s~fj,ingAerosol and Gas Sumpliv f b r Ambient Air Monitoring. Presented at 82nd Annual Meeting, Anaheim, Ca. Air & Waste Management Association, Pittsburgh, PA. Forrest, J . and L. Newman (1973) Sampling and Analysis of Atmospheric Sulfur Compounds for Isotope Ratio Studies. Atmos. Enuiron. 7, 561. Fujita, E. M. and J. F. Collins (1989) Quulity Assurunce j b r the Southern California Air Qttulity Study. Presented at 82nd Annual Meeting, Anaheim, CA. Air & Waste Management Association, Pittsburgh, PA. Fung, K . K., S. L. Heisler. A. Price, B. V. Nuesca, and P. K. Mueller (1979) Comparison of Ion Chromatography and Automated Wet Chemical Methods for Analysis of Sulfate and Nitrate in Ambient Particulate Filter Samples. In: Ion Chromutogruphic Analysis qf l+wironmental Po//utant.q Vol. 2, E. Sawicki and J . D. Mulik (eds.). Ann Arbor Science Publishers, Inc., Ann Arbor, MI, pp. 203-209. Fung, K . (1988) Artifiris in the Sumpling qf Ambient Orgunic Aerosols. Presented a t 1988 EPAiAPCA Symposium on Measurement of Toxic and Related Air Pollutants, Raleigh, NC. Air Pollution Conlrol Association, Pittsburgh, PA. Fung, K. K . (1990) Particulate Carbon Speciation by MnO, Oxidation. Aerosol Sci. Techno/. 12, 122- 121. Gerber, H. E. (1082) Optical Techniques for the Measurement of Light Absorption by Particulates. In: Purticulute C,’urhon: Atmospheric Lifk Cycle, G. T. Wolff and R. L. Klimisch (eds.). Plenum Press, New York, NY. Gordon, R. J. (1974) Solvent Selection i n Extraction of Airborne Particulate Matter. Atmos. Emiron. 8, 189. Gotoh, T. (1980) Physical Examination of a Method for Determination of Nitrogen Dioxide in the Atmosphere by Using Tricthanolamine Filter Paper. Tuiki Osen Gukkuishi 15, 334 (in Japanese). Grosjean, D. (1975) Solvent Extraction and Organic Carbon Determination in Atmospheric Particulate Matter: The Organic Extraction-Organic Carbon Analyzer (OE-OCA) Tcchnique. A n d . Chem. 47 (6). 797 - 805. Harman, J. N. (1989) Inductively Coupled Plasma Emission Spectroscopy. In: Methods of’ Air Sampling und Analysis, 3rd ed.. J. P. Lodge, Jr. (ed.). Lewis Publishers, Chelsea, MI, pp. 88 - 92. Heintzenbcrg, J. (1982) Measurement of Light Absorption and Elemental Carbon in Atmospheric Aerosol Samples from Remote Locations. In: Purticulrite Curbon, Atniosplwric Lifi, C ~ c l e , G. T. Wolff and R. L. Klimisch (eds.). Plenum Press, New York, NY, pp. 371 -377. Heisler, S. L., R. C . Hcnry, and J. Collins (1980a) The Nutiire o / the Denver Huze in Noucrnher und Decrrnber qf’ 1978. Presented at 73rd Annual Meeting, Montreal, Quebec, Canada. Air Pollution Control Association, Pittsburgh, PA. Heisler, S. L., R. C. Henry, J. G. Watson, and G. M. Hidy (1980b) The 1978 Denuer Winter Huze Study. Document #P5417-1, prepared for Motor Vehicle Manufacturer’s Association of the United States by Environmental Research and Technology, Inc., Westlake Village, CA.

I56

J . G. Wmtsoii iind J . C. C%OM~

Hering, S. V., R. C. Flagan, and S. K. Friedlander (1979a) Design and Evaluation of New Low-Pressure Impactor-I. Eni+on. S1.i. T i ~ c h o l 12, . 667. Hering. S. V., S. K . Friedlander, J . J. Collins, and L. W. Richards (1979b) Design and Evaluation of 21 Low I'rcssure Impactor-11. Bioiron. Sci. Tecltnol. 13, 184- 188. liering, S. V., 0. R. Lawson, I . Allcgrini, A. Febo. C. Perrino, M. Possanzini, L. E. Sickles, II., K. G. Anlauf, A . Wicbe, B. R. Appel, W. John, J . Ondo, S. Wall, R. S. Braman, R. Sutton, G . R . Cass, P. A. Solomon, D. J. Eatough, N. L. Eatough, E. C. Ellis, D. Grosjean, B. B. Hicks, J. D. Womnck, J . Horrocks, K . T. Knapp, T. G. Ellstad, R. J. Paul-, W. J. Mitchell, M. Pleasant. E. Pcake, A . MacLcan, W. R. Picrson, W. Brachaczek, H. I. Schif!". G. I. Mackay. C. W. Spicer. D. H. Stedman, A. M. Winer, H. W. Bierniann, E. C. Tuazon (1988) The Nitric Acid Shootout: Ficld Comparison of Measurement Methods. Atmos. Enoiron. 22 ( I ) , 1519- 1539. Hering, S . V. ( 1989) Inertial and Gravitational Collectors. In: Air Sumpling fnstr~tments,fiw Euuluc~fionof At/no.splier'ic Conmnirzant.~.7th ed., S. V. Hering (ed.). American Conference of Governmental Industrial Hygienists, Cincinnati, OH, pp. 337 - 385. Hering, S. V.. B. R. Appel, W. Cheng, F. Salaymch, S. H. Cadle. P. A. Mulawa. T. A. Cahill. R. A. Eldred, M. Surovik, D. Filz, J . E. Howes, K. T. Knapp, L. Stockburger, B. J. Turpin. J. J. Huntzicker, X.-Q. Zhang, and P. H. McMurry (1990) Comparison of Sampling Methods for Carbonaceous Aerosols in Ambient Air. A m m d Sci. T i ~ h i o l12, . 200 - 21 3. Hering, C. V., J . C. Chow. and S. Chandra (1993) Component Tr.sting,for LI Two-Week Stunpler fbr Fin<]Particle Ions und Guseo~isAcid.r. Presented a t thc American Association for Aerosol Research Twelfth Annual Meeting, Oak Brook, IL, 1 1 - 15 October 1993. Huntzicker, J. J., R. L. Johnson, J . J . Shah. and R. A. Cary (1982) Analysis of Organic and Blcmental Carbon in Ambient Aerosol by a Thermal-Optical Method. In: Particirkufe C'urhon, A I I ~ I O S ~ ILifi I P ~C I' j "i . /~c , G . T. Wolff and R. L. Klimisch (eds.). Plenum Press, New York, NY, pp. 79 - 88. Husar, R. R. (1 974) Atmospheric Particulate Mass Monitoring with a Beta Radiation Dctector. Armos. Enuiron. 8 , 183- 188. Jaklevic, J. M., B. W. Loo, and F. S. Goulding (1977) Photon-Induced X-Ray Fluorcscencc Analysis Using Energy-Dispersive Dctection and Dichotomous Samplcr. In: X-RCIJFluorescence Anrr(i~.yisof Enuiro~iniiwrulSumnpks, 2nd ed., T. G. Dzubay (ed.). Ann Arbor Science Publishcrs. Ann Arbor, MI. Japar, S. M., A. C. Szkarlat, R. A. Gorse, Jr., E. K . €leyerdahl, R. L. Johnson, J . A. Rau, and J . J . Huntzicker (1984) Comparison of Solvent Extraction and Thermal-Optical Carbon Analysis Methods: Application to Diesel Vehicle Exhaust Aerosol. Enairon. Sci. Tr.clinol. 18 (4), 23 I - 234. John, W. and (i. Reischl (1980) A Cyclone for Size-Selective Sampling of Ambient Air. J . Air Poll. Control. Assoc. 30, 872- 876. John, W., S. M. Wall, and J. J. Wesolowski (1983a) Project Sunirnarj,: Vulidntion qf Sampler.?.fbr fnhaletl Particulute Murrer. Document #EPA-602/S4-83-0 10, U.S. Environmental Protection Agency, Environmental Monitoring Systcms Laboratory, Research Triangle Park. NC. John, W., S. flering, G. Reischl. and G. V. Sasaki (1983b) Characteristics of Nuclepore Filters with Large Pore Size - l l Filtration Properties. Aftnos. Enuiron. 17, 373. John, W.. S. M. Wall. J. L. Ondo. and H.-C.Wang (1986) Dry Dqmsition of Acidic Guses und Ptrr'ric.los. Air and Industrial llygiene Laboratory, California Ilepartmcnt of Health Services, Bcrkcley, CA. John, W.. S. M. Wall, and J. L. Ondo (1988) A New Method for Nitric Acid and Nitrate Aerosol Measurement Using the Dichotomous Sampler. Atmos. Enuiron. 22, 1627- 1635. John, W. and H.-C. Wang (1991) Laboratory Testing Method for PM-I0 Samplers: Lowered Effectiveness from Particle Loading. Aerosol Sci. Technol. 14, 93. Johnson, I). A. and D. H. 1:. Atkins (1975) An Airborne System for the Sampling and Analysis of Sulfur Dioxide and Atmospheric Aerosols. Atmos. Enoiron. 9 , 825. Johnson, R . L. and J. J . Huntzicker (1979) Analysis of Volatilizable and Elemental Carbon in Ambient Aerosols, In: Proceedings, Ciirhonaceous Pur[icIe.y in die Atnwsphere, T. Novakov (ed.). Document #LBL-9037. Lawrence Berkeley Laboratory, Berkeley, CA, pp. 10- 13. Johnson, R . L., J . 1. Shah, and J. J. Huntzicker (1980) Analysis of Organic, Elemental, and Carbonate Carbon in Ambient Aerosols, In: Sampling and Analysis q j Toxic Orgunics in f l i e Atmosphere. American Society for Test and Materials, p. 1 1 I .

Measurements on Filters

157

Johnson, R. L., J. J. Shah, R. A. Cary, and J. J. Huntzicker (1981) An Automatcd Thermal-Optical Method for the Analysis of Carbonaceous Aerosol. In: A C S Symposium Series #167: Atmospheric Aerosol, SourcelAir Quality Relationships, E. S. Macias and P. K. Hopke (eds.). Plenum Press, New York, NY, pp. 223-233. Knapp, K. T., J. L. Durham, and T. G. Ellestad (1986) Pollutant Sampler for Measuremcnts of Atmospheric Acidic Dry Deposition. Environ. Sci. Technol. 20,633. Koutrakis, P., J. M. Wolfson, J . L. Slater, M. Brauer, and J. D. Spengler (1988) Evaluation of an Annular Denuder/Filter Pack System to Collect Acidic Aerosols and Gases. Environ. Sci. Techno/. 22, 1463. Koutrakis, P., J. M. Wolfson, K. M. Thompson, J. D. Spengler, G. J. Keeler, and J. L. Slater (1991) Determinution of Aerosol Strong Acidity Losses Due to Interactions qf Collected Particles: Results froni Lahoratory and Field Studies. Presented at the 84th Annual Meeting, Vancouver, British Columbia, Canada. Air & Waste Management Association, Pittsburgh, PA. Koutrakis, P., K . M. Thompson, J. M. Wolfson, J. D. Spengler, G. J. Keeler, and J. L. Slater (1992) Determination of Aerosol Strong Acidity Losses Due to Interactions of Collected Particles: Results from Laboratory and Field Studies. Ainios. Environ. 26A (6), 987-995. Kukreja, V. P. and J. L. Bove (1976) Determination of Free Carbon Collected on High-Volume Glass Fiber Filter. Enuiron. Sci. Technol. 10, 187. Lancaster, N. and N. G. Nickling (1993) Aeolian Sediment Transport. In: Geomorphology uf' the Desert Environment, A. Abrahams and A. J. Parson (eds.). Chapman and Hall, New York, NY, pp. 447-473. Lee, K. W. and M. Ramamurthi (1993) Filter Collection. In: Aerosol Measurement: Principles, Techniques and Applications, K. Willeke and P. A. Baron (eds.). Van Nostrand, Reinhold, New York, NY, pp. 179-205. Lillienfeld. P. and J . Dulchinos (1972) Portablc Instantaneous Mass Monitoring for Coal Mine Dust. A m . Indus. Hyg. Assoc. J . 33, 136. Lillienfeld, P. (1975) Design and Operation of Dust Mcasuring Instrumentation Based on the Beta-Radiation Method. Stuub-Reinhalil. Luff 25, 458. Lillienfeld, P. (1979) G C A '7 Model A P M - 1 Ambient Particulate Monitor-Field Applications and Results. Presented at 72nd Annual Meeting, Cincinnati, OH. Air Pollution Control Association. Pittsburgh, PA. Lin, C. I . , M. Baker, and R. J. Charlson (1973) Absorption Coefficient of Atmospheric Aerosol: A Method for Measurement. Appl. Opt. 12, 1356. Lippmann, M. (1989) Sampling Aerosols by Filtration. In: Air Sampling Instruments f o r Eualuation of Atmospheric Contaminants, 7th ed., S. V. Hering (ed.). American Conference of Govcrnrnental Industrial Hygienists, Cincinnati, OH, pp. 305 - 336. Lodge, J. P., Jr., (ed.) (1989) Methods qf Air Sumpling und Analysis, 3rd ed. Lewis Publishers, Chelsea, MI. Lynch, A. J., N. R. McQuaker, and D. I. Bicun (1980) ICP/AES Analysis and the Composition of Airborne and Soil Materials in the Vicinity of a Lead/Zinc Smelter Complex. J . Air Poll. Control Assoc. 30, 259. Macias, E. S. and R. B. Husar (1976) Atmospheric Particulate Mass Measurement with Beta Attenuation Mass Monitor. Environ. Sci. Technol. 10, 904. Macias. E. S., R. Delumyea, L. C. Chu, H. R. Appleman, C. D. Radcliffe, and L. Staley (1979) The Determination, Speciation, and Behavior of Particulate Carbon. In: Proceedings, Carbonaceous Particles in the Atmosphere, T. Novako (ed.). Document # LBL-9037, Lawrence Berkeley Laboratory, Berkeley, CA, p. 70. Malissa, H. ( I 979) Some Analytical Approaches to the Chemical Characterization of Carbonaceous Particulates. In: Proceedings, Carbonaceous Particles in the Ainmsphere, T. Novakov (ed.). Document #LBL-9037, Lawrence Berkeley Laboratory, Berkeley, Ca, p. 3. Malm, W. (1979) Considerations in the Measurement of Visibility. J . Air Pollution Control Assoc. 29, 1042- 1052. Marple, V. A,, B. Y. H. Liu, and G. A. Kuhlmey (1981) A Uniform Deposit Impactor. J . Aerosol sci. 12, 333. McCarthy, R. and C. E. Moore (1952) Determination of Free Carbon in Atmospheric Dust. Anal. Chem. 24. 41 1.

IS8

J . C. Wutson ot7d J . c'. Chow

McDow, S. R . and J . J . Huntzicker (1990) Vapor Adsorption Artifact in the Sampling of Organic Acrosol : Face Velocity Effects. Atnros. Enuiron. 24A, 2563 -2567. McFarland, A. R., C. A . Ortiz, and R. W. Bcrtch (1978) Particle Collection Characteristics of a Single-Stage Dichotomous Sampler. Euciron. Sci. Teclznol. 12, 679. McFarland, A. R., C. A. Ortiz, and C. E. Rodes (1980) Characterization of Sampling Systems. In: P,ocwrtitrg.s oftlrr Tc~chnicuIBusis fbr A Size Spcciji". Prrrticulutc. Srundurd. Air Pollution Control Association, Pittsburgh, PA. p. 59. McFarland. A. R . and C. A. Ortiz (1984) ~ ' I ~ ~ i r u c t i ~ r i ~qf'~ rSierra-Awdersm tioii P M I o lnlet Morld 246h. Texas A & M Air Quality Laboratory Kcport 4716/02/02/84/ARM, College Station, TX. McQudker, N. R., P. D. Khickner, and N. C. Gok (1979) Calibration of an Inductively Coupled Plasma-Atomic Emission Spectrometer for the Analysis of Environmcntal Matcrials. A t d C'hrtii. 51. 888. Mer7. W. ( 1978) Automatcd Rapid Methods in Organic Elemental Analysis. Mikrochitn Acts. WiOll., 11, pp. 519-534. Meserolc, F. B., K . Schwitzgcbel, B. F. Jones, C. M. Thompson, and F. G. Mesich (1976) Szdfirr Diosidc. lntc~rji~rvncrs in rhe Merisrrr~rnc~nt of Amhirnt Particulate Srrlfutes. Electric Power Research Institute, Palo Alto, CA. Meserole, F. B., 8 . F. Jones. L. A. W. C. Hawn, K. R. Williams, and T. P. Parsons (1979) Nitrogcn Oxide Intwfiwrrcrs in the Meti.siirc~rn~n/ of' A trno.rpkivic Purtic~iiluteNitrutes. Document # EA1031. Electric Powcr Research Institute, Palo Alto, CA. Miller, F. J., D. E. Gardncr, J . A. Graham, R . E. Lee. Jr., W. E. Wilson, and J. D. Bachmann ( 1979) Size Considcrations for Establishing a Standard for lnhalable Particles. .I. Air Pollrrtiorr Control Assoc. 29, 610-615. Muellcr, P. K., R. W. Mosely, and L. 13. Pierce (1971) Carbonate and Non-Carbonate Carbon i n Atmospheric Particulates. In : P r o c d i n g s , S m m d Internutionul Clwn Air Congrr~ss.Academic Prcss. New York, NY. Mueller. P. K.. B. V. Mcndom. J. C. Colliiis, and E. S . Wilgus (1978) Application of Iron Chromatography to the Analysis 01' Anions Extracted from Airborne Particulate Mattcr. In: Ion C'lrrornrrtogrri/)lrii,Anu/j~si.vo / k~nnuironmmtmlPollutnnts, E. Sawicki, J. D. Mulik, and E. Wittgenstein (cds.). Ann Arbor Science Publishcrs, Inc., Ann Arbor, MI, pp. 77-86. Mucllcr, P. K., Ci. R. Hilst, G. M . Nidy, T. F. Lavcry, and J. G. Watson (1981) E P K l Srr/fufc R q i o n u l E.ypcrinirnf : Rr~.ctilt,smid 1tnplicutiorr.s. Document # EA-2077-SY -LD, Electric Powcr Research Institute, Palo Alto, CA. M u c h . 1'. K., K. K . Fung, S. L. Hcisler, D. Grosjean, and G. M. Hidy (1982) Atniosphcric Particulate Carbon Observations in Urban and Rural Areas of the United States. In: Purticir/ur(, Curhon, Atnro.~phr~ric /,if> Cjdi>,G. T. Wolffand R. L. Klimisch (eds.). Plenum Press. Ncw York, N Y , pp. 343 - 370. Mucllcr. P. K., G. M. Ilidy, R . L. Baskctt, K . K . Fung, R. C. Henry, T. F. Lavcry, N . J . Nordt, A. C. Lloyd, .I.W. Thrasher, K. K. Warren, and J . G. Watson (1983) Sulfute Re~ionulE.upcri/iicnt ( S I J R E ) : Report of' Firrtlings. Docunicnt #EA-1901, Electric Power Research Iiistitutc, Palo Alto, C A . Muhlboier, J. L. and R. L. Williams (1982) Fireplaces, Furnaces and Vchicles as Emission Sources 01' Particulate Carbon. I n : Purticulrrtc C'arhorz, Atmospheric Lift. Cycle, G. T. Wolff and R. L. Klimisch (eds.). Pleiiuin Prcss, New York, NY, pp. 185-205. Mulik, .I., R. Puckett, D. Williams, and E. Sawicki (1976) Ion Chromatographic Analysis of Sulfate and Nitratc i n Ambient Aerosols. Knuiron. Sci. Trchrrol. 9 (7), 653 -663. Mulik, J. D., R. Puckett, E. Sawicki, and D. Williams (1977) Ion Chromatography - A New Analytical Technique for the Assay of Sulfate and Nitrate in Ambient Aerosols. In: Methods tu7d Sturidurtlcfbr Eucironnwnriil M e u ~ u r e ~ ~Document ~ ~ ~ n t . #464. National Bureau of Standards. Washington. DC. Novakov. T. ( I98 1) Microchemical Characterization of Aerosols. In: Nature, Aim and MetIrod~ of' Microclrc~n7i.strj~, Proceedings of the 8th Intcriiational Microchemical Symposium, Graz, Austria, 25-30 August 1980, 11. Malissa. M. Grasscrbaucr, and R. Belcher (eds.). SpringerVerlag, Wicn, pp. 141 - 165. Novakov, T. (1982) Soot in the Atmosphere. In: Purticulute Carbon: Atrnospheric Li/k C'yclc). G. T. Wolll'and R . L. Klimisch (eds.). Plenum Press, Ncw York, NY. pp. 19-41.

Measurements on Filters

159

Ohira, T., M. T. Ishuguro, and 0. Koyama (1976) Gas Chromatographic Analysis of Offensive Odors at Plants Handling Fish Entrails and Bones. Japan Soc. Air Pollut. 2, 58-59. Ohtsuka, T., T. Mirakami, M. Kaneko, and S. Sano (1978) Measurement of Degrec of Air Pollution Using Triethanolamine Filter Papers. Kanagawaken Kogai Senta Nernpo 10, I69 (in Japanese). Okita, t. and S. Kanamori (1971) Determination of Trace Concentration of Ammonia in the Atmosphere Using Pyridine-Pyrazolone Reagent. Atmos. Environ. 5, 621, O h , J. G. and R. R. Bohn (1983) A New P M , , Medium Flow Sampler. Presented a t the 76th Annual Meeting, Atlanta, GA. Air Pollution Control Association, Pittsburgh, PA. Olmez, I . (1989) Instrumental Neutron Activation Analysis of Atmospheric Particulate Matter. In: Methods o f Air Sampling and Analysis, 3rd ed., J . P. Lodge, Jr. (ed.). Lewis Publishers, Chelsea, MI, pp. 143- 150. Patashnick, H. and E. G . Rupprecht (1991) Continuous PM,, Measurements Using the Tapered Element Oscillating Microbalance. J . Air Waste Manage. Assoc. 41, 1079. Patterson, R. K. (1 973) Automated Pregl-Dumas Technique for Determining Total Carbon. Hydrogen and Nitrogen in Atmospheric Aerosols. Anal. f h e m . 45, 605- 609. Perry, W. H. (1989) Sequential and Tape Samplers - Unattended Sampling. In: Air Sampling Instruments ,for Evaluation of Atmospheric Contaminants, 7th ed., S. V. Hering (ed.). American Conference of Governmental Industrial Hygienists, Cincinnati, OH, pp. 291 - 303. Pimenta, J . A. and G. R. Wood (1980) Determination of Frcc and Total Carbon in Suspended Air Particulate Matter. Enuiron. Sci. Techno/. 14, 556-561. Pyen, G. S . and M. J. Fishman (1979) Determination of Anions in Pore Waters from Cores. I n : Ion Chromatography Anulysis of’ Environinentu~Pollutaiits, J. D. Mulik and E. Sawicki (eds.). Ann. Arbor Science Publishers, Inc., Ann Arbor. MI, pp. 235-244. Raabe, 0. G., D. A. Braaten, R. L. Axelbaum, S. V. Teague, and T. A. Cahill (1988) Calibration Studies of the DRUM Impactor. J . Aerosol Sci. 19, 183- 195. Ranweiler, L. E. and .I. L. Moyers (1974) Atomic Absorption Procedure for Analysis of Metals in Atmospheric Particulate Matter. Enuiron. Sci. Technol. 8, 152. Rich, W., J. A. Tillotson, and R. C. Chang (1978) Ion Chromatography: An Analytical Perspective. In: Ion Chromatographic Analysis u f Environmental Pollutants, E. Sawicki, J. D. Mulik, and E. Wittgenstein (eds.). Ann Arbor Science Publishers, Inc., Ann Arbor, MI, pp. 17-29. Rogge, W. F., L. M. Hildemann, M. A. Mazurek, G. R. Cass, and B. R. T. Simoneit (1991) Sources of Fine Organic Aerosol. 1. Charbroilers and Meat Cooking Operations. Environ. Sci. Technol. 25, 1 1 12- 1125. Rogers, C. F., J. G. Watson, and C. V. Mathai (1989) Design and Testing of a New Size Classifying Isokinetic Sequential Aerosol Sampler. J . Air Poll. Control Assoc. 39, 1569 - 1576. Rosen, H., A. D. A. Hansen, L. Gundel, and T. Novakov (1978) Identification of the Optically Absorbing Component in Urban Aerosols. Appl. Opt. 17 (24), 3859-3861. Rosen, H., A. D. A. Hansen, R. L. Dod, L. A. Gundel, and T. Novakov (1982) Graphitic Carbon in Urban Environments and the Arctic. In: Particular Carbon, Atmospheric Life Cycle, G . T. Wolff and R. L. Klimisch (eds.). Plenum Press, New York, NY, pp. 273-294. Rubow, K . L. and V. C. Furtado (1989) Air Movers and Samplers. In: Air Sampling Instrument.? ,for Evaluation of Atmospheric Contaminants, 7th ed., S. V. Hering (ed.). American Conference of Governmental Industrial Hygienists, Cincinnati, OH, pp. 241 - 274. Rupprecht, E., M. Meyer, and H. Patashnick (1992) The Tapered Element Oscillating Microhalance as a Too1,for Measuring Ambient Particulate Concentrations in Real Time. Presented at European Aerosol Conference, Oxford, U.K. Shendrikar, A. D. and J. P. Lodge, Jr. (1975) Microdetermination of Ammonia by the Ring Oven Technique and Its Application to Air Pollution Studies. Atmos. Enuiron. 9,431. Small, H., T. S. Stevens, and W. C. Bauman (1975) Novel Ion Exchange Chromatographic Method Using Conductimetric Detection. Anal. fhern. 47, 1801 - 1809. Small, H. (1978) An Introduction to Ion Chromatography. In: Ion Chromatographic Analysis of Environmental Pollutants, E. Sawicki, J. D. Mulik, and E. Wittgenstein (eds.). Ann Arbor Scicnce Publishers, Inc., Ann Arbor, MI, pp. I I -21. Smith, T. B., W. White, J. Anderson, and S. Marsh (1975) Thc. Chemistry, Dispersion and Transport of Air Pollutants Emittedfrom Fossil Fuel Pawer Plants in California. Prepared for California Air Resources Board, Sacramento, CA, by Ted B. Smith & Associates, Pasadena, CA.

160

J . G . M/iitsoti und J . C . Choii,

Solomon, P. A., T. Fall, L. Salmon, G. R. Cass, H. A. Gray, and A. Davison (1989) Chemical Characteristics of P M , Aerosols Collected in the Los Angeles Area. J . Air Poll. C'ontrul A.s.so1.. 39, 154- 163. Spicer. C. W. and P. M. Schumacher (1977) Interference in Sampling Atmospheric Particulate Nitrate. Atriros. Emiron. 11, 873. Stevens, R. K . , W. A. McClenny, T. G. Dzubay, M. A. Mason, and W. J. Courtney (1982) Analytical Methods to Measure thc Carbonaceous Content of Aerosols. In: Purticulute Carbon, Atnio.splieric Lifk C,vcI(~s,G. T. Wolff and R. L. tilimisch (eds.). Plenum Press, New York, NY. pp. 1 1 1 - 129. Stevens, R. K., R. J. Paur, 1. Allegruni. I:. DeSantis, A. Febo, C. Perrino, M. Possanzini, ti. W. Cox, E. E. Estes, H. R. Turnes, and J. I;. Sickles, I I (1985) Measurement of HNO,, SO,, N H , and Particulate Nitrate with an Annular Denuder System. In: Ptoceetlings of the F@h Annuol Ncitiontrl Sytiiimiurn on Prcsent Arllwniw in the Mensuriviient of Air Pollution. EPA/600/9-8 5-029, U.S. Environmental Protection Agency. Research Triangle Park, NC, pp. 55 -71. Stevens, R. K., L. J. Purdue, H. M. Barnes. R. P. Ward. J. 0 . Baugh, J. P. Bell, H. Sauren, J . E. Sickles, 11, and L. L. Hodson (1990) Annular Denuders and Visibility Studies. In: Tronscrcr i o t i s , VisibilitI~lrnd Fin(, Particl(~.s,C. V. Mathai (ed.). Air & Waste Management Association. Pittsburgh, PA, p. 122. Stevens, R. K.. J. Pinto. T. L. Conner. R. Willis, R. A. Rasmussen, Y. Mamane, G. Casuccio, I . Benes, J . Lanicek, P. Subri. J . Novak, and J . Santroch (1993) CZECH Air ToxicSrudy ( C A T S ) Project Surnnirrry. Presented at the X6th Annual Meeting, Denver. CO. Air & Waste Maiiagcmcnt Association, Pittsburgh, PA. Tanner, R. L., J . S. Gaffney, and M. F. Phillips (1982) Dcterrnination of' Organic and Elemental Carbon in Atmospheric Aerosol Samples by Thermal Evolution. Anal. Clirvn. 54, 1627 - 1630. Tanner, R. L., A. W. Gertler, E. M. Fujita, and J. C. Chow, Atmospheric Organic Acids: A Review. Submitted to Attiio.s. Enuiran. Thurston, G. D., J. E. Gorczynski, Jr., P. Jacques, 1. Currie, and D. He (1992) An Automated Sequential Sampling System for Particulate Acid Acrosols: Description, Characterization, and Field Sampling Results. J . E.-cposure A n d . Emiron. Epi~letniology2 (4), 41 5 -428. Tombach, I . H., D. W. Allard, R. L. Drake, and R. C. Lewis (1987) Western Regionul Air Qircrlif!, * rind Air Quirlity Mecrsrrrernents: 19x1 - 1982. Report #EA-4903, Prepared for Electric Power Research Institute by AeroVironment. Inc., Monrovia, CA. Watson. 1. G. ( 1979) C h e n i i d Elwwrit B h n c e Rcwptor Modd Metlrodology ,/Or A.s.se.s.sing tlrc S o i n w s of' Fine atirl Totrrl Ptrrticultrtr Mutter. Ph.D. Dissertation, Oregon Graduate Center, Beaverton, OR. University Microfilms International, Ann Arbor, MI. Watson, J. G., J. C. Chow, J. J. Shah, and T. G. Pace (1983) The Effect of Sampling Inlets in thc I'M," and P M , S to TSP Concentration Ratios. J . Air Poll. Control Assoc. 33, 114- 119. Watson, J. G., J. L. Bowcn. J. C. Chow, C. F. Rogers, M. G. Ruby, M. J. Rood, and R. T. Egami (1989) Method 501 : High Volume Measurement of Size Classified Suspended Particulate Mattcr. 111: Methocls 01 Air Smipling utid Ana/)~sis,3rd ed., J. P. Lodge (ed.). Lewis Publishers. Inc., Chclsca. MI, pp. 427-439. Watson, J. G., N . F. Robinson, J . C. Chow, R . C. Henry, B. M. Kim, T. G. Pace, E. L. Meyer, and Q. Nguyen (1990) The USEPA/DRl Chemical Mass Balance Receptor model, C M B 7.0. Eni:iron. S(?ft\i'are 5, 38 -49. Watson, J. G., J. C. Chow, and T. G. Pace (1991 a ) Chemical Mass Balance. In: Receptor Modeling / o r Air Q u u l i t ~Muncrgrnirrit, P. K . Hopkc (cd.). Elsevicr Science Publishers, Amsterdam. The Netherlands, pp. 83 - 1 16. Watson, J . G., J . C. Chow. C . A. Frazicr, and D. H. Lowenthal (1991 b) Progrrrwl Pltrn fbr on ltiip(vkd Vdlej/Mr~sicrrliP M I , Sourcc Apportinntiicnt Study. DRI Document #8623. I D I , prepared for U S . Eiiviroiiinciital Protection Agency, Region IX, by Desert Research Institute, Reno, NV. Watson, J . G. and J. C.Chow (1993) Ambient Air Sampling. I n : Acrosol M(usurcwwit: Principles, Twliriiyues crntl,4pi~licution~s, ti. Willekc and P. Baron (cds.). Van Nostrand Reinhold, New York, NY, pp. 622-639. Wedding. J . B., A. R . McFarland, and J. E. Ccrmak (1977) Large Particle Collection Characteristics nf Ambient Aerosol Samplers. Emiron. Sci. Twlinol. 4, 387.

Measurements on Filters

161

Wedding, J. B. and M. A. Weigand, and T. C. Carney (1982) A PM,, Cutpoint Inlet for the Dichotomous Sampler. Environ. Sci. T c d m l . 16 (9), 602 -606. Wedding, J. B., M. A. Weigand, M. W. Ligotke, and R. Baumgardner (1983) Wedding Ambient Aerosol Sampling Inlet for an Intermediate Flow Rate (4 cfm) Sampler. Environ. Sci. Tc~chnol. 17 (7), 379-383. Wedding, J. B. (1985) Errors in Sampling Ambient Concentrations Employing Setpoint Temperature Compensated Mass Flow Transducers. Atmus. Environ. 19, 1219. Wedding, .I. B., M. A. Weigand (1985) The Wedding Ambient Aerosol Sampling Inlet (D50 = 10 pm) for the High Volume Sampler. Atmos. Environ. 19, 535. Wedding, J. B., M. A. Weigand, Y. J. Kim, D. L. Swift, and J. P. Lodge (1987) A Critical Flow Device for Accurate PM,, Sampling and Correct Indication of PM,, Dosage to the Thoracic Region of the Respiratory Tract. J . Air Poll. Control. Assoc. 37, 254-258. Weiss, R. E., A . P. Waggoner, R. Charlson, D. L. Thorsell, J. S. Hall, and L. A. Riley (1979) Studies of the Optical, Physical, and Chemical Properties of Light Absorbing Acrosols. In: Proceedings: Conference an Curhonuceous Purticles in the Atinosphrre, T. Novakov (ed.). Document # LBL9037, Lawrence Berkeley Laboratory, Berkeley, CA, p. 257. Willeke, K. (1975) Performance of the Slotted Impactor. A m . fnd. Hyg. Assoc. J . 39, 683. Witz, S. and J . G. Wendt (1981) Artifact Sulfate and Nitrate Formation at Two Sites in the South Coast Air Basin - A Collaborative Study Between the South Coast Air Quality Management District and the California Air Resources Board. Environ. Sci. Trchnol. 15, 79. Witz, S., R. W. Eden, M. W. Wadley, C. Dunwoody, R. P. Papa, and K. J. Torre (1990) Rapid Loss of Particulate Nitrate, Chloride and Ammonium on Quartz Fiber Filters During Storage. J . Air Pollution Control Assoc. 40, 53-61, Wolff, G . T., P. J. Groblicki, S. H. Cadle, and R. J. Countess (1982) Particulate Carbon at Various Locations in the United States. In: Purticulute Curhon: Atmospheric Lift Cycle, G . T. Wolff and R. L. Klimisch (eds.). Plenum Press, New York, NY, pp. 297-315. Wolff, G. T., M. S. Ruthkosky, D. P. Stroup, and P. E. Korsog (1991) A Characterization of the Principal PM Species in Claremont (Summa) and Long Beach (Fall) During SCAQS. Armos. Environ. 25A. 2173-2186. Woods. M . C., F. Chen, and M. B. Ranade (1986) Wind Tunnel Test Reports I 4 through 25. Research Triangle Institute, Research Triangle Park, NC. Zoller, W. H. and G. E. Gordon (1970) Instrumental Neutron Activation Analysis of Atmospheric Pollutaiits Utilizing Ge( Li) X-Ray Detectors. Anal. Chen?. 42, 257.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

7 Organic Gas Sampling Barbara Zielinska and Eric Fujita

7.1 Introduction Organic gases are emitted from many naturally occurring biogenic and geogenic sources, as well as from anthropogenic sources such as petroleum refining, oil and gas production, agricultural burning, industrial processes, and from motor vehicles. Most of these compounds are highly reactive in the atmosphere: some undergo chemical transformations in the troposphere, contributing to ozone and organic aerosol formation; some are transported into the stratosphere and contribute to the destruction of the ozone layer; and many are toxic to living organisms. Organic compounds exhibit a wide range of volatility and are hence distributed in the atmosphere between the gas and particle phases. Compounds having saturated vapor pressure at 25 "C greater than lo-' mm Hg are generally classified as volatile organic compounds (VOC) and are present entirely in the gas phase. Compounds with saturated vapor pressure at 25 "C between 10- and l o p 7mm Hg are generally called semivolatile organic compounds (SOC)and are distributed between the gas and particle mm Hg are phases. Compounds having saturated vapor pressure at 25 "C less than nonvolatile and are predominantly adsorbed on particles. Different sampling techniques are required for the quantitative collection of VOC, SOC, and nonvolatile organics. This chapter describes only sampling techniques employed for the collection of volatile and - to some extent - semivolatile organic compounds. Aerosol sampling is discussed elsewhere in this book. Methods for organic gas sampling include collection of whole air or preconcentration of samples on chemically selective and nonselective adsorbents (Rudolph et al., 1990).Each sampling methodology includes the following steps: (1) selection and preparation of sampling media; ( 2 ) the actual sampling process; (3) the transport and storage of the collected samples. The selection of the optimal sampling method for target compounds (or a class of compounds) depends greatly on the physicochemical nature of these compounds and their expected concentrations in air - sample volumes must be compatible with the sensitivity of the analysis method, and the expected behavior of the targeted compounds during each step of the sampling process must be carefully considered.

7.2 Whole-Air Sampling Sampling of whole air with containers of defined volume has been successfully employed for volatile compounds of low polarity. This method has two main limitations: (1) the sample volume is limited to a few liters which, for low compound

I 64

B. Zielinsku and E. Fujifu

concentrations encountered in ambient air, may be insufficient for analysis purposes; (2) sample stability during storage is sometimes in doubt due to adsorption on (or desorption from) container walls and chemical reactions between compounds. However, the recent development of very sensitive analytical methods for organic compound analysis and more information available on stability of major classes of volatile organic compounds in stainless steel canisters have made this method of sampling very popular.

7.2.1 Sampling Media Selection and Preparation Containers typically used for VOC collection in whole air include flexible plastic bags (such as TeflonTM,Tedlar, Mylar, etc.), glass bulbs, and stainless steel SUMMATMcanisters. Plastic bags are relatively inexpensive and can be purchased in different sizes, typically from 10 to 1001 of volume. Bags are purchased from the manufacturer (e.g., in the US., Cole-Parmer Instrument Co., Niles, IL; SKC, Inc., Eighty Four, PA; Scott Specialty Gases, Inc.) ready to use and require only minimal cleaning. The bags are cleaned by repeatedly (usually three times) filling the bag with ultra-high purity zero air and evacuating it with a small pressure/vacuum pump. The bags are also checked for leaks by leaving them filled with zero air overnight. However, the simplicity and low cost of bags are offset by potential contamination during sampling and storage. Bags have been found to outgas some residual materials used in bag processing, for example, sulfur gases (Hoyt et al., 1993), N,N-dimethylacetamide (Siegl et al., 1993), and other contaminants, depending on the bag’s brand and batch. In addition, some bags are permeable to certain chemicals and have been observed to lose significant amounts of sample when stored for prolonged periods of time or to allow contaminant influx from the outside (Jayanty, 1989). Since the only cleaning procedure available is repeated filling with zero air, the bags may show a memory effect from previous sample collection (Holdren et al., 1989). Glass bulbs (distributed, for example, by such companies as Supelco, Inc., Bellefonte, PA, and ACE Glass, Inc., Vineland, NJ) can be thoroughly cleaned, have an inert inside surface, do not outgas any contaminants, and are not permeable. However, they allow for collection of very limited sample volumes (usually up to 1 l), are fragile and difficult to transport and store. Stainless steel canisters as sample containment vessels offer several advantages over other containers for whole-air sampling. The interior surfaces of the canisters are conditioned by the SUMMATMprocess, a proprietary treatment that passivates the internal surfaces of the canister to minimize surface reactivity. This process allows stable storage for many of the compounds of interest. Passivated stainless steel canisters in a variety of volumetric sizes are commercially available from several manufacturers (e.g., in the U.S., Andersen Samplers, Inc., Atlanta, GA; Nutech, Durham, NC; Scientific Instrumentation Specialists, Moscow ID, etc.).

Organic Gas Sampling

I65

The canisters can be used repeatedly for many years, however, their thorough cleaning prior to sampling is essential. The U.S. EPA recommended method of canister cleaning (EPA Method TO- 14 and the EPA document “Technical Assistance Document for Sampling and Analysis of Ozone Precursors”; U.S. EPA, 1991) involves repeated evacuation and pressurization of canisters with humidified zero air. A process of evacuation to -0.5 mm Hg absolute pressure, followed by pressurization with ultra-high-purity (UHP) humid zero air to 20 psig, repeated three times, is recommended. At the end of the multiple evacuation/pressurization cycle, the canister is filled with humid zero air and, after a period of equilibration (at least overnight), a sample of this humid zero air is withdrawn and analyzed. If the canister meets the project-specific cleanliness criteria, the canister is considered clean and is evacuated to -0.5 mm Hg absolute pressure prior to shipment to the field. An alternative approach to cleaning the canisters uses heat ( - 100 “C) in the evacuation/pressurization process. Although heating has not been proven conclusively to be superior for producing cleaned canisters, according to the authors’ experience and that of others (Rasmussen, 1992; Fung et al., 1994), heating is critical to achieving the desired canister cleanliness. A phenomenon known as offgasing is sometimes encountered with cleaned canisters (Wagoner et al., 1993). Trace levels of various organic compounds may be observed in cleaned canisters, after they have been filled with humid zero air and equilibrated for a period of days. Because of this effect, canisters used for ambient low-level measurements should be segregated from those used for high-level concentrations or higher molecular weight organic compounds.

-

7.2.2 Sampling A variety of sampling systems is employed for whole-air ambient VOC collection. Selection of the sampling device is dictated by the ultimate sampling goals. If only a “grab sample” is needed, no sampling device is necessary for stainless steel canisters. Since canisters are shipped to the field under a vacuum, a grab sample can be obtained by manually opening the canister valve. In most cases whole-air samples are integrated over a specific period of time, and an automated mode of operation is required. In this case ambient air is drawn into a sampling train that consists of an evacuated canister and upstream components that serve to regulate the rates and duration of air sampling. In the simplest design, the differential pressure between the atmosphere and the evacuated canister causes flow into the system. A mass flow controller or a critical orifice placed in-line regulates the flow rate. This system is sometimes called subatmospheric sampling (EPA Method TO-14), since the canister pressure after sampling is below or at atmospheric pressure. The disadvantages of this system are that: (1) the flow rate tends to decrease towards the end of sampling, as the difference between the canister pressure and ambient pressure decreases; (2) this method does not allow for collection

166

B. Zielinska and E. Fujilu

of pressurized samples. It has been shown that, in order to achieve good compound recovery from the canister, the sample pressure should be as high as possible without causing precipitation of liquid water within the canister (Coutant and McClenny, 1991). The most common method of canister sampling is pressurized sampling, i.e., collection of a whole-air sample in a canister with the aid of a pump regulated by a mechanical or electronic flow controller to achieve a final canister pressure above atmospheric pressure. In both the subatmospheric and pressurized systems the flow rate and sample duration must be matched in order to maintain constant flow during the sampling period. A canister sampling system can be obtained commercially (from the canister vendors), or can be custom-built for a specific application. Unattended operation of the sample collection system is a desirable and often necessary capability. In addition, in order to allow collection of certain samples in accordance with a specific schedule (e.g., collecting ozone precursor samples), it is practical to employ a n automated multiple-event canister sampling system. Fig. 1 shows the schematic of

i

Fig. 1. Automated multi-event canister sampling system. (A) Sample manifold inlet, (B) 2 micron filter, (C) flow control device, (D) metal bellows pump, (E) digital timer, (F) latching solenoid valve, (I) 6-L sample canister, (J) by-pass pump.

Organic Gas Sampling

I67

such a system, which does not require an operator to change the canisters between sampling events. This type of system is recommended by the EPA document “Technical Assistance Document for Sampling and Analysis of Ozone Precursors” (US. EPA, 1991).A microprocessor-controlled, fully automated, multi-purpose and multi-event sampler for VOC collection (either by canister or solid adsorbent method) has recently become available commercially (Scientific Instrumentation Specialists, Moscow, ID). Regardless of which system is used for sampling, the cleaning of sampling devices is critical, since a sampler can introduce contaminants or may cause loss of sampled compounds. The sampling system should be constructed of clean, high-quality components, with particular attention paid to pumps, valves, flow controllers, and components having any non-metallic surfaces. Before installation and at periodic intervals during use, samplers should be carefully tested for contaminants or compound loss by analyzing samples of zero air and samples of known concentrations of targeted compounds, collected through these samplers. This procedure is called “certification” and allows the potential contamination characteristics of each sampling system to be assessed. Standard operating procedures have been established for the certification and validation of samplers and can be found in EPA Method TO-14 and in the EPA document “Technical Assistance Document for Sampling and Analysis of Ozone Precursors” (US. EPA, 1991). In the case of bag sampling, a very simple sampling device is commercially available (Thomas Tortorete Co., Fairfield, CA) or one can be easily constructed, for obtaining grab or time-integrated samples. The device consists of an air-tight container (“vacuum box”) with a sampling line extended outside. An evacuated bag is placed inside the box with its inlet attached to the sampling line. As air is evacuated from the box with the aid of an external pump, the bag is filled with outside air. After sampling, the pump is turned off, the bag’s inlet closed, and the bag removed from the box. This sampling system is free from the contaminants introduced by the pump, flow control devices, etc. However, as mentioned above, the bags are not recommended for organic gas sampling due to contaminants from the bag itself.

7.2.3 Storage and Transport The stability of samples upon storage is an important factor to consider in VOC analysis. For the sample to be stable, the compound matrix and concentration of the sample must not change significantly with time. A number of studies (Wagoner et al., 1993; McClenny et al., 1991; Fung et al., 1994; Holdren and Smith, 1986; Oliver et al., 1986) have shown that a wide range of VOC are stable in canisters for at least 30 days. Most of the reported studies were performed in SUMMATMtreated stainless steel canisters at pressure above atmospheric pressure and in the presence of water vapor. Various observations at several laboratories have indicated that the presence of water vapor is essential in maintaining sample integrity; it neutralizes active sites in the sampling/canister system (Pate et al., 1992; Coutant and McClenny, 1991; Kelly et al., 1993).

168

3.Z i e h k u and E. Fujitu

However, some studies indicate that many oxygenated hydrocarbons such as aldehydes, ketones, and alcohols have less desirable storage properties (Pate et al., 1992; Kelly et al., 1993). Generally, organic compounds that are soluble in water do not store well in canisters. Also, hydrocarbons with carbon numbers higher than C , o are not recovered quantitatively from the canisters (Zielinska and Fung, 1992; Zielinska et al., 1993). In addition, the variable quality and SUMMATMtreatment of the canisters, their previous history, age, and storage temperatures, etc., may affect a compound’s stability upon storage. Target analytes for which there is little stability information should be tested prior to sampling for storage stability in the canisters under conditions of use during actual sampling. It has been demonstrated that Tedlar bag samples show significant contamination during storage (Holdren et al., 1989). The previous history of a bag seems to have a major effect on subsequent contamination of samples stored in the bag. In addition, the bag samples should be sheltered from light during storage in order to avoid any photooxidation reactions. However, for some specific target analytes, bags may be the sampling containers of choice; it has been shown, for example, that some sulfur compounds (H,S, COS and CH,SH) are stable for up to 72 hours in Tedlar bags fitted with polypropylene valves (but not with metal valves) but are unstable in SUMMATM-passivatedcanisters (Parmar, 1991). Stainless steel canisters are easy to transport; they are durable and study and do not require special temperature conditions. The plastic bags are more prone to damage during transport if not properly packed. Glass bulbs are the most difficult to transport, which is the main reason they are not used often in field sampling.

7.3 Preconcentration Methods Due to generally low concentrations of organic compounds in ambient air, one of the most widely used methods for sampling of gaseous contaminants is their preconcentration either on a suitable solid adsorbent or, if the contaminant is reactive, in an absorbing solution contained in a bubbler or impinger or coated on some solid porous support. Cryogenic concentration of VOC in an empty tube or a tube filled with glass beads and cooled by liquid oxygen or argon is also employed, especially in connection with gas chromatography (EPA Methods TO-3, TO- 12 and TO-14) (Grosjean and Fung, 1984; Pierotti, 1990).

7.3.1 Preconcentration on Nonselective Solid Adsorbents 7.3.1.1 Sampling Media Selection and Preparation Since the early 1970s the technique of preconcentration of VOC on solid adsorbents has been extensively developed and many applications have been found. A number of solid adsorbents are available commercially. The most widely used adsorbents are listed in Tab. 1, together with some of their physical properties. Porous polymers,

169

Organic Gas Sampling

Tab. 1. The Most Widely Used Sorbents in Organic Gas Sampling Pore Sorbent

Composition"

Silica gel Alumina Florid Charcoals: coconut-based petroleum-based wood-based Carbosieve B Carbosieve S-111 Carbotrap Carbotrap C Carboxen-(564, 563 or 569) Carbosphere Tenax-GC XAD-2 XAD-4 XAD-7 Chromosorb 101 Chromosorb 102 Chromosorb 105 Poropak N Poropak P Poropak Q Poropak R Poropak S Polyurethane foam C - carbon; Mg

-

SiO,

Specific Surface Area (m2/g)

Mean Size (A)

300 - 800

20-40

800 - 1000 800 - 1000

20 18-22

1000 550 100 12

15-40 3000 2000

1000 19 - 30

13 120

290 - 300 750 450 50 300 -400 600 - 700 225 - 300 100-200 630- 840 550 - 700 450 - 600

85 - 90 50 80 3500 90 500 120 150 75 76 76

A1203

Mg silicate Carbon Carbon Carbon Graphitized Graphitized Graphitized Graphitized Graphitized

C C C C C

black black black black black

Graphitized C black Poly(2,6-diphenyl-pphenylene oxide) S-DVB copolymer S-DVB copolymer S-DVB copolymer S-DVB copolymer S-DVB copolymer Polyaromatic type Polyvin ylpyrrolidone S-DVB copolymer EVB-DVB copolymer Polyvinylpyrrolidone Polyvinylpyridine Porous polyurethane foam magnesium; S

-

styrene; DVB - divinylbenzene; EVB

-

ethylvinylbenzene.

such as Tenax-GC (or TA), XAD resins, and polyurethane foam, have found wide application in organic gas sampling (see, for example, EPA Methods TO-1 and TO-13). Tenax-GC (and recently Tenax-TA) is the most popular porous polymer sorbent, mainly because of its high thermal stability (up to 350 "C),and hence low bleed during thermal desorption (Brown and Purnell, 1979), and extremely low affinity to water vapor. The main disadvantages of Tenax-GC are its relatively poor capacity for more volatile compounds (i.eq,those with boiling points [b.p.] < 80 "C) and the possibility of chemical reactions occurring during sampling in the presence of some reactive gases and during thermal desorption (Pellizzari et al., 1984; Walling et al., 1986; Zielinska et al., 1986). Indeed, small amounts of benzaldehyde, acetophenone and cinnamaldehyde were observed to be released from Tenax when it was used for sampling in the presence of substantial amounts of ozone, particularly when the same Tenax cartridge was used repeatedly for sampling (Walling et al., 1986; Zielinska, 1993, unpublished results). Other types of sorbents, such as various types of charcoal, carbon molecular sieves (e.g., Carbosieve and Carboxen offered by Supelco, Inc., or Carbosphere

I70

3.Zirlinsku ond E. Fujitci

offered by Altech Associates, Inc.) and other carbon-based sorbents are also widely used, especially for more volatile, lower molecular weight compounds. However, due to their high surface activity, chemical reactions may occur during storage and desorption of the samples (Rudling et al., 1986). Very low recoveries, for compounds in the ppb range, are found because of partial decomposition of some compounds and strong adsorptive behavior of the charcoal sorbents. The selection of a proper adsorbent for a given application depends mainly on the following factors (Namiesnik, 1988): (a) the volume of an air sample which can be passed through the sorbent without breakthrough of the compounds of interest; (bf stability of target analytes on the sorbent during sampling, storage, and desorption; (c) any background signals due to the sorbent; (d) affinity of the sorbent for water; (e) efficiency of desorption of collected compounds; (f) the enrichment factor. To be analytically useful, the preconcentration proccss must be quantitative. Thus, the evaluation of the sorptive capacity of a given sorbent under defined conditions prior to its use is necessary. The breakthrough volume (i.e., the volume of gas that may be passed through the sorbent without breakthrough of the compound of interest) is a frequently used measure of the capacity of a sorbent bed. The breakthrough volume can be determined either from theoretical calculations, particularly for an active carbon bed, or experimentally (Namiesnik, 1988, and references therein). For some solid adsorbents, this volume can be roughly estimated from data published in the literature, particularly for Tenax (Brown and Purnell, 1979; Krost et al., 1982; U.S. EPA Method TO-1) and Carbotrap (Supelco, 1986, 1988). However, the effects of various parameters on breakthrough volume cannot be easily determined, and these uncertainties are taken into account by applying an empirical correction term to give a “safe sampling volume” (e.g., a 1.5 safety factor is recommended by U S . EPA Method TO-1). Some factors known to have an important effect on the breakthrough volume include (Namiesnik, 1988) temperature, flow rate, relative humidity of the air sampled, concentration of contaminants, physicochemical properties of the sorbent and target analytes, and presence of other components in the mixture. Due to the great effect of these factors (particularly temperature) on breakthrough volumes and some differences found between published and observed retention volumes (Walling et al., 1986), it is recommended that two or more solid adsorbent samples be collected in parallel at different flow rates (Walling, 1984; U.S. EPA Method TO-1) or two sorbent cartridges be used in series (Arey et al., 1987; Atkinson et al., 1988). The latter method is used more often, since it allows the semi-quantitative evaluation of the inadequate retention of a given compound. The desorption behavior of the adsorbents is a very important factor in selecting the proper solid adsorbent for a given application. Compounds trapped on a solid adsorbent must bc released as quantitatively as possible before analysis. Solvent extraction or thermal desorption is used for this purpose. Thermal desorption is preferred over solvent extraction, since it avoids the dilution of an enriched sample with a solvent; it allows the entire amount of a collected sample to be injected at once into a gas chromatographic column, thus providing maximum sensitivity. A disadvantage of the thermal desorption method is that the sample can be analyzed only once.

Organic Gas Sampling

171

The efficiency of thermal desorption depends primarily on the type of the adsorbent and, for a given adsorbent, the volatility of the compound to be desorbed. For example, it has been shown that Tenax-TA releases n-octadecane (C,8H38,b.p. 316 "C) quantitatively when heated at 280 "C over 10 min (Zielinska and Fung, 1992), whereas the efficiency of desorption of n-pentadecane (CI5HJ2,b.p. 270 "C) from Carbotrap at 320 "C is almost zero (Rothweiler et al., 1991). Carbon molecular sieve type adsorbents, such as Carbosieve S-I11 (Supelco, Inc.), are capable of effectively retaining gaseous C, - C, hydrocarbons, however C, (and higher) compounds are desorbed in trace quantities (Levaggi et al., 1992). Recently, multi-bed adsorbent tubes have become popular, since they allow a wider group of gaseous organic compounds to be trapped and released efficiently (Des Tombe et al., 1991, and references therein; Pollack et al., 1991). Some combinations of different adsorbents are available commercially, for example, Carbotrap 300, offered by Supelco, Inc., shown in Fig. 2. The other combinations of solid adsorbents, such as Tenax/Carbotrap/Carbosieve S-I11 (Levaggi et al., 1992), or Tenax/carbon molecular sieves (Soroka et al., 1992) were also successfully employed. All solid adsorbents must be cleaned prior to use. The cleaning procedure depends on the type of adsorbent. Carbon-based adsorbents require only heating under nitrogen flow for several hours at -400 "C (EPA Method TO-2), whereas porous polymers are usually extracted with organic solvent(s) prior to heating. For example, Tenax-GC is cleaned by sequential Soxhlet extraction with methanol and n-pentane (EPA Method TO-1) or in a 6/4 (v/v) acetone/hexane mixture (Atkinson et al., 1988) and then thermally conditioned for four hours by heating at 280 "C under a nitrogen purge. XAD resins and polyurethane foam have low thermal stability and cannot be conditioned thermally. They require complex cleaning procedures (Chuang et al., 1987; Offerman et al., 1990). The cleaned solid adsorbent cartridges are tightly sealed and stored in (preferably) metal containers with activated charcoal on the bottom of the container.

carbotrap c graphitized carbon black

P-

graphitized

carbon black heavlest

(traps C5C8

Carhieve S-Ill molecular sieve (traps light

compounds)

Wool

Fig. 2. Carbotrap 300 tube.

172

B. Zielinskn and E. Fujita

7.3.1.2 Sampling The sampling train usually consists of solid adsorbent cartridge(s) and, downstream of the cartridge, a pump and other component(s) that regulate the rate and duration of sampling. Since collection of a precisely known volume of air is critical to the accuracy of the results, mass flow controllers are preferred over conventional needle valves or critical orifices. If the latter are used, the flow rate should be checked before and after each sample collection. A microprocessor-controlled, multi-event, solid absorbent sampler has recently become available commercially (Scientific Instrumentation Specialists, Moscow, ID).

7.3.1.3 Storage and Transport To preserve sample integrity, the collected samples should be stored in a refrigerator or, preferably, in a freezer and shipped in cooled containers. Since all solid adsorbents have higher sorptive capacity at low temperatures, care should be taken to tightly seal the cartridges and containers used for their storage and shipping. Porous polymers are relatively inert with respect to most of the sampled compounds. Although there is no information available concerning the stability during storage of many compound classes on different polymeric adsorbents, some methods (for example EPA Method TO-1) call for sample analysis within two weeks after sampling. In contrast, such adsorbents as silica gel, aluminium oxide, and activated charcoal, due to their high surface activity, may induce chemical reactions of adsorbed compounds during storage of the sample (Rudling et al., 1986). It has been reported (Rothweiler et al., 1991) that Carbotrap, a very pure graphitized carbon black said to be free from contaminants (Supelco, 1986), causes degradation of more reactive compounds, such as a-pinene and aldehydes. It is not clear, however, if these reactions occur during storage or during the desorption process.

7.3.2 Selective Methods of Compound Preconcentration The classical example of a selective preconcentration method for organic gas sampling is the collection of carbonyl compounds by their derivatization with 2,4-dinitrophenylhydrazine (DNPH).The acid-catalyzed derivatization of carbonyls proceeds by nucleophilic addition of the DNPH to a C = O bond, followed by 1,Zelimination of water to form 2,4-dinitrophenylhydrazone:

IYU 2

Orgunic Gus Sampling

173

The D N PH-hydrazones, formed during sampling, are non-volatile and remain on the sampling medium, which is usually either a reagent-impregnated cartridge or an impinger charged with the reagent solution. The yellow to deep-orange colored DNPH-hydrazones have UV absorption maxima in the 360- 375 nm range and can be analyzed by the high performance liquid chromatography (HPLC) method coupled with UV detection; this method offers high selectivity and sensitivity of analysis. Another example of a selective preconcentration method for organic gas collection is the collection of gaseous organic acids by NaOH- or Na,CO,-coated filters or solid adsorbents. Several recent review articles treat the subject of carbonyl compound (see, e.g., Vairavamurthy et al., 1993; Otson and Fellin, 1988) and organic acid (Tanner et al., 1993)sampling and analysis in detail. This chapter provides only general information concerning sampling techniques used for these compound classes.

7.3.2.1 Sampling Media Selection and Preparation Carbonyl Compounds. Early applications of the DNPH-based method for ambient air carbonyl compound sampling involved the use of impingers charged with D N P H solution in either 2N HCI (Kuwata et al., 1979) or in acetonitrile with the addition of various acids, such as sulfuric acid (Kuntz et al., 1980), perchloric acid (Tejada, 1986), or phosphoric acid (De Bortoli et al., 1986). Although impinger techniques have been used successfully in many studies, and are recommended by the Intersociety Committee (1989a), they are cumbersome and not well suited to large field studies or to sampling at remote locations. The DNPH-coated solid sorbents are not only more convenient to use but also provide higher sensitivity than impingers, since they preconcentrate DNPH derivatives to a higher degree. A number of DNPH-coated solid sorbents have been used as sampling media for gaseous carbonyl compound collection. These include glass beads (Grosjean and Fung, 1982; Fung and Grosjean, 1981), glass-fiber filters (Levin et al., 1985, 1986), silica gel cartridges (Tejada, 1986; U.S. EPA Method TO-ll), Florisil cartridges (Lipari and Swarin, 1985), C I 8cartridges (Kuwata et al., 1983b; Druzik et al., 1990), XAD-2 resins (Anderson et al., 1979), Carbopack-B (Ciccioli et al., 1987), and Chromosorb P (Gromping and Cammann, 1989). Among these, commercially available Sep-Pak silica gel cartridges and Sep-Pak C, cartridges (both from Waters Associates, Milford, MA) have found the widest application. To prepare a sampling medium, an acidified DNPH solution in acetonitrile is passed through a prewashed Sep-Pak cartridge; recently, DNPH-coated ready-to-use silica Sep-Pak cartridges became commercially available (Waters Associates, Milford, MA). However, the Sep-Pak cartridge technique has several problems (Vairavamurthy et al., 1993, and references therein). Although the hydrazones of stable carbonyls, such as formaldehyde, acetaldehyde, propionaldehyde, acetone, etc., maintain their integrity on both silica gel and C I 8 cartridges for over a month under refrigerated storage, olefinic aldehydes such as acrolein and crotonaldehyde degrade partially on the cartridges and form unknown products. However, it has been reported that the acrolein DNPH derivative is not stable in strongly acidic DNPH solutions used

174

B. Zielinskiz cind E. Fujitu

in the impinger method either. In order to preserve its integrity, acrolein DNPH derivative has to be extracted from these solutions immediately after sampling (Freeman, 1992). Most important and still not completely understood is thc effect of ozone on sampling with DNPH-coated solid adsorbents. It has been reported (Arnts and Tejada, 1989) that DNPH-coated silica gel catridges showed a dramatic negative interferencc by ozone in thc determination of formaldehyde. This brings into question the validity of the silica gel sampling technique (recomrnendcd by U S . EPA Method TO-1 I), unless an ozone scrubber is used. In contrast, C I 8 cartridges exhibit no loss of hydrazone-formaldehyde derivative up to 120 ppbv of ozone (Arnts and Tejada, 1989). Arnts and Tejada (1989) suggested that, in the case of silica gel cartridges, a D N P H derivativc which largely forms at the front of the cartridge and is immobilized, is destroyed by ozone. In the case of C I 8 cartridges, the radicals generated by the ozonc attack can be scavenged by thc C I 8 phase, thus limiting further attack on DNPH or hydrazone. Howcver, it has been suggested (Vairavamurthy et al., 1993) that ozone causes the production of artifact carbonyls in reagent-coated C cartridges. In August 1986, during the 10-day Carbonaceous Species Methods Comparison Study (CSMCS) conducted by the California Air Resources Board in Glendora, CA, the efficiency of the C1 cartridge collection method for formaldehyde was compared to more direct methods, such as long-path Fourier transform infrared detector (FTIR), differential optical absorption spectrometer (DOAS), and the Tunable Diode Laser Absorption Spectroscopy (TDLAS). The results compared vcry well (Fung and Wright, 1990; Lawson et al., 1990). Also, during the 1987 Southern California Air Quality Study, formaldehyde measurements were made at the Claremont sampling site by TDLAS and DOAS and by C 1 D N P H cartridges. Measurements of HCHO by C1 D N P H cartridges generally tracked ozone concentrations and were in reasonably good agreement with TDLAS measurements (Fujita and Croes, 1990). These data indicate that the DNPH-coated C , 8 cartridge collection method is reliable in the presence of higher ozone concentrations (the studies were conducted in southern California on days when ozone concentrations ranged from 200 to 300 ppb), at least as far as the formaldehyde measurements are concerned. The artifactual formation of higher carbonyl compounds under the influence of ozone cannot be excluded at the present time. EPA Method TO-11 currently recommends using the silica gel cartridge in conjunction with a KI scrubber to overcomc the ozone artifact. The use of C u O cartridge in front of the sampling cartridge has also been reported (Vairavamurthy et al., 1993).However, the use of ozone scrubbers has not been sufficiently time-tested to ensure its reliability. Gaseous Organic Acids. The preconcentration of gaseous organic acids on filters or cartridges coated with basic materials which convert the acid molecules to their alkali salts is onc of the most widely used methods of collecting these compounds. Particulate acids are removed from the airstream by pre-filters or discriminated in the collection process using diffusion denuder tubcs. Quartz filters impregnated with Na,CO, (Norton, 1985) or cellulose filters impregnated with K,CO, (Andreae et al., 1987) have been used to collect gaseous formic, acetic, and pyruvic acids from the air. KOH-coated and carbonate-coated Sep-Pac C, cartridges have also been

Orgunic Gus Sumpling

I75

employed (Grosjean et al., 1989).NaOH-coated (Winiwarter et al., 1988; Rosenberg et al., 1988; Puxbaum et al., 1988) or carbonate-coated (Norton, 1992) annular denuders have also been used to determine ambient levels of gaseous organic acids. Peroxyacetyl nitrate (PAN) interferes positively with acetate determination (Grosjean and Parmer, 1990; Grosjean et al. 1988), but this bias can be estimated if PAN is simultaneously measured. According to Grosjean and Parmer (1990), conversion of PAN to acetate varied with the collection device, being 11- 17% for carbonate- and KOH-coated filters, 16 - 27% for carbonate cartridges and 100% for K O H cartridges. Clearly, carbonate-impregnated filters are the most suitable choice for field operations in which PAN cannot be routinely measured. Other interferences, such as from ozone (Grosjean, 1988) and the conversion of formaldehyde and acetaldehyde to formate and acetate (Grosjean, 1988; Winiwarter et al., 1988; Keene et al., 1989) seem to be less important.

7.3.2.2 Sampling Development of equipment for selective methods of sample preconcentration is based on the same principles as nonselective methods, briefly discussed in Section 7.3.1.2.

7.3.2.3 Storage and Transport Carbonyl Compounds. As mentioned above, hydrazones of stable carbonyls, such as formaldehyde, acetaldehyde, propionaldehyde, acetone, etc., maintain their integrity on silica gel and C L 8cartridges, as well as in impinger reagent solutions, for over a month under refrigerated storage (Fung et al., 1994; Vairavamurthy et al., 1993, and references therein). However, the olefinic aldehydes such as acrolein and crotonaldehyde degrade partially on the cartridges, either during sampling or storage, and form unknown products. Also, acrolein DNPH derivative is not stable in the strongly acidic DNPH solutions used in the impinger method. In order to preserve its integrity it has to be extracted from this solution immediately after sampling (Freeman, 1992). Gaseous Organic Acids. Since formate and acetate ions in precipitation samples can be consumed by bacteria omnipresent in the atmosphere (Herlihy et al., 1987), it is important to use appropriate techniques to preserve the samples for laboratory analysis. However, it is not clear whether and how rapidly microbial decomposition takes place in the case of sorbed gaseous organic acids, and how effective treatment with a biocide (usually chloroform) is in preserving samples (Tanner et al., 1993).

7.4 Semi-Volatile Organic Compounds Compounds with saturated vapor pressure at 25 "C between l o - ' and mm Hg are generally called semivolatile organic compounds (SOC)and are distributed between the gas and particle phases. This distribution is determined not only by

176

B. Zielinska und E. Fuiitu

the equilibrium vapor pressure of the individual species, but also by the amount and type of particulate matter present (how much adsorption surface is available), and by temperature (Ligocki and Pankow, 1989). For example, Tab. 2 gives the vapor pressures at 25 "C of some representative polycyclic aromatic hydrocarbons (PAH) ranging from naphthalene to benzo(a)pyrenc (BaP). Thc factor of lo7 in the range of vapor pressures is reflected in the fact that, at equilibrium at ambient temperature, naphthalene exists almost entirely in the gas phase, while BaP, other five-ring PAH, and higher-ring PAH are predominantly adsorbed on particles. The intermediate three- and four-ring PAH arc distributcd between the two phases. However, the vapor pressures of these intermediate PAH can be significantly reduced by their adsorption on various types of surfaces. Because of this phenomenon, the amount and type of particulate matter present play an important role, together with tcmpcrature, in the gas-particle partitioning of semivolatile organic compounds. The partitioning of SOC between gas and particle phases has received much attention recently (Ligocki and Pankow, 1989; Cotham and Bidleman, 1992; Lane ct al., 1992; Kaupp and Umlauf, 1992; Pankow, 1992). Most estimates of partition have relied on high-volume (hi-vol) sampling, using a filter to collect particles followed by a solid adsorbent trap, such as polyurethane foam (PUF), Tenax, or XAD-2, to collect the gaseous portion of SOC (cf., Kaupp and Umlauf, 1992, and Foreman and Bidleman, 1990, and references therein). However, the pressure drop behind a hi-vol filter or cascade impactor suggests the possibility of the occurrence of artifacts due to volatilization during the sampling process (Coutant et al., 1988). Such volatilization (sometimes called blow-off) would cause underestimation of the particle-phase concentrations of organics. On the other hand, adsorption of gaseous substances on deposited particles, or on the filter material itself, a process driven by the lowered vapor pressure over the sorbed material, would lead to over estimation of the particle-phase fraction (Bidleman et al., 1986; Ligocki and Pankow, 1989; McDow and Huntzicker, 1990). It has been shown recently

-

Tab. 2. Vapor Pressures at 25 "C for a Series of PAH" PAH Naphthalene Acenaphthylene Acenaphthene Fluorene Phenanthrcne Anthracene Fluoranthrene Pyrene Ben~o[a]anthrdcene Bcnzo[a]pyrcnc Chrysene

Vapor Pressure at 25 "C (torr) 8.0 x 10- * 6.7 x 2.2 x 10-3 6.0 x 10-4 1.2 x 10-4 6.0 x 9.2 x lo-" 4.5 x lo-" 2.1 x 1w7 5.6 x lo-" 6.4 x lo-''

Sonnefeld et al., 1983, except as indicated. Yamasaki et al., 1984.

Organic Gas Sampling

I77

(Kaupp and Umlauf, 1992) that the maximum differences observed between hi-vol filter-solid adsorbent sampling and impactor sampling (the latter believed to be less susceptible to these sampling artifacts) typically do not exceed a factor of two. There is good theoretical and experimental evidence that use of a diffusion denuder technique significantly improves measurements of gas-particle phase partitioning (Coutant et al., 1988,1989,1992; Lane et al., 1988).However, the reliability of existing denuders for investigation of atmospheric partitioning of non-polar SOC needs to be improved, as suggested by contradictions in published field data (cf., Kaupp and Umlauf, 1992, and references therein). A new, improved sampler has recently been introduced (Gundel et al., 1992) which uses a proprietary XAD-6coated tube for vapor collection, followed by filter collection for organic aerosol particles and a sorbent bed to quantitatively retain desorbed (blown-off) organic vapors. Preliminary results from the use of this device look very promising for direct measurements of the phase distribution of semivolatile organic aerosol constituents.

7.5 Passive Sampling Techniques Passive VOC samplers usually utilize solid adsorbents as a sampling medium, however the samples are collected by means of gaseous diffusion rather than by means of a pump. They are easy to use, inexpensive, light, and safe; because of these advantages they have found wide application as personal samplers in indoor air quality studies in residential and workplace environments, in some pilot studies, etc. The most conimon type of passive sampler is a badge with a charcoal disk as the adsorbent (for example 3M passive sampler, available from 3M Centre, St. Paul, MN, or organic vapor monitoring badges from Supelco, Inc., and SKC, Inc.). The passive samplers specific to a particular group of compounds, for example, passive samplers utilizing filters impregnated with DNPH for collection of carbonyl compounds (Levin et a]., 1988), have also been described. The VOC collected with charcoal-based badge-type samplers are usually desorbed with carbon disulfide for analysis (Brown et a]., 1984). Another type of passive sampler is in tube form, with a porous polymer or other solid adsorbent suitable for thermal desorption inside a tube (Coulson et al., 1984). Ideally, the behavior of most passive diffusion-type samplers can be described by Fick’s first law of diffusion, restated as (Shields and Weschler, 1987): m -- D(Ca _ tA

-

Cf)

L

where m = mass of substance that diffuses (pg); t = sampling interval (s);A = crosssectional area of the diffusion path (cm2);D = diffusion coefficient for the substance in air (cm2/s);C, = concentration of substance in air (pg/cm3); Cf = concentration of substance just above the sorbent, assumed to be 0; L = the diffusion path length (cm). Since C, is assumed to be 0, this equation can be rearranged to:

m tCa

-

DA L

178

B. Ziehnda NWOE. Fuiitcr

The term m/(tC,) is commonly referred to as the “uptake rate” or “sampling rate”. In theory, for a given substance and type of sampler, the uptake rate is constant as long as the amount of material collected remains significantly less than the capacity of the sorbent used. Once an uptake rate for a particular substance and sampler has been determined, it can be used to calculate the ambient concentration, C,, of this substance, from measured m (Shields and Weschler, 1987). However, other factors, such as temperature, relative humidity, analyte concentration in air, and non-ideal behavior of sorbents, may affect the uptake rate. In addition, during the sampling period, the passive sampler should be placed in a location with adequate air movement; stagnant air at the face of the sampler will result in the build-up of a static layer of air and in the collection of a nonrepresentative sample. Sampling rates (uptake rates) for most of the commercially available samplers and for a large number of organic vapors commonly encountered in the workplace have been calculated from the diffusion coefficients, or measured experimentally, and are available from the manufacturers. However, it is advisable to calibrate samplers against known atmospheres under a variety of exposure conditions, a procedure that can be time-consuming (Brown et al., 1984). Recently, several types of diffusion samplers were tested under a variety of conditions and over a range of face velocities (Otson and Fellin, 1991). It has been shown that 3M monitorsfitted with draft shields demonstrate only a small velocity effect ( < 10%) over the three face velocities tested (0.01; 0.5; and 1.8 m/s). Most of the passive samplers were developed to monitor elevated concentrations of organic compounds in industrial environments. The use of these samplers in non-industrial settings, such as office or residential buildings, requires long sampling intervals to collect sufficient material for analysis on the order of several days or even months. It has been shown (Shields and Weschler, 1987) that passive samplers used for monitoring ambient concentrations of organic vapors over long sampling periods (between 336 and 1500 h) produced reasonable results (on average 13% reproducibility and 25% accuracy). When only long-time-averaged data on ambient concentrations of organic gases are required, the passive sampling technique may be the method of choice, especially for indoor environments. Siiice pumps and accessories are not required, passive sampling is less costly and easier to implement. It is actually possible to conduct an extensive sampling program through the mail (Sexton et al., 1986) - samplers can be mailed out to the sampling location and then mailed back.

7.6 Summary This chapter reviews the most widely used methods for organic gas sampling which are applicable to many organic compound classes. The most applicable sampling methods for common VOC classes found in ambient air are listed in Tab. 3. This table also includes appropriate reference methods or, in some cases, citation of the appropriate published article@) in which the method is described.

-Go

I . EPA Method TO8 ( U S . EPA, 1988) 2. Hawthorne et al., 1988; 1989 NIOSH Method Manual (NIOSH, 1984)

Canisters; lmpingers with water; Cryogenic traps; Condensation sampling

Porous polymers

Canisters; Porous polymers; Carbotrap; Cryogenic trapping

Charcoal

1. lmpingers with 0.1 N NaOH 2. Filter/PUF

Charcoal

ALCOHOLS

PHENOLS/CRESOLS

ETHERS

NIOSH Methodb Manual (NIOSH, 1984)

EPA Method TO14 (US. EPA, 1988)

Porous polymers; Multibed sorbents

Canisters

HALOGENATED HYDROCARBONS

c,-c4

Andreae et al., 1987

Base-coated C , * Sep-Pak; Denuders; Mist chamber

Base-coated filters

Filter/XAD; Filter/Tenax

ORGANIC ACIDS

CIl~C20

c4

c 2

DNPH-coated C,, Sep-Pak DNPH-coated Si Sep-Pak; Druzik et al., 1990 Impingers; Cryogenic traps

U S . EPA, 1991 U.S. EPA, 1991 Zielinska and Fung, 1992 EPA Method TO1 3 (U.S. EPA, 1988)

Tang et al.. 1993

Reference Method

CARBONYL COMPOUNDS

Carbon molecular sieves Carbotrap; Bags

Other Sampling Methods

Semivolatile PAH

Canisters

Most Applicable Sampling Method

Canisters Canisters Tenax Filter/PUF

-c,

HYDROCARBONS: Methane

Common VOC Class

Tab. 3. Sampling Methods for Common Ambient VOC Classes

See comment for alcohols

The applicability of charcoal collection/CS, elution method for ambient oxygenated compounds is currently being evaluated by AeroVironment, Inc. (1992)

See Tanner et al., 1993, for critical review

Cryogenic method: see Pierotti, 1990

See also Des Tombe et al., 1991, for C,-C4 hydrocarbon sampling and analysis method review

Comments

0

e

Q

2.

G?

Charcoal

ESTERS

Porous polymers

ORGANIC SULFUR GASES

Including PUF. Silica gel is recommended for methanol (NIOSH Method 2000)

Canisters; Bags

FilterIPUF

SEMIVOLATILE PESTICIDES

(b)

FilteriXAD; Filter/Tenax

Charcoal Porous Polymers"

Porous polymers; Impingers with acidic solution Porous polymers; Impingers with acidic solution Porous polymers

Alkyl Nitrates Nitro-aromatics

Acidified C,, Sep-Pak

Silica gel

- C4

EPA Methods T04. TO10 (US. EPA, 1988)

Atlas and SchauMer, 1991

Intersociety Committee, 1989b

EPA Method TO7 (U.S. EPA, 1988) Kuwata et al., 1983a

NIOSH Method Manual (NIOSH, 1984)

Carbotrap; Porous polymers; Cryogenic trapping; Canisters Ascorbic acid solution

Reference Method

Other Sampling Methods

Aromatic Amines

Amines C

ORGANIC NITROGEN COMPOUNDS: Ni trosarnines Porous polymers

Most Applicable Sampling Method

Common VOC Class

Tab. 3. (continued)

See Brunnemanet al., 1980. for ascorbic acid method

See comment for alcohols

Comments

h s

<:

%

9

a

B

P

Organic Gas Sampling

181

In addition to these methods, there are many specific collection and analysis methods that have been developed for particular organic air pollutants. These methods are found in such compendiums as the US. National Institute of Occupational Health and Safety (NIOSH) Analytical Methods Manual (NIOSH, 1984) or the U.S. EPA’s Compendium of Methods,for the Determination of Toxic Organic Compounds in Air (U.S. EPA, 1988) as well as in the peer-reviewed literature. Acknowledgements. The authors thank Dr. John Sagebiel for helpful discussions and Ms. Susan Grobman and Ms. Beverly Finley for their skillful editorial assistance.

7.7 References AeroVironment, Inc. (1992) Document No. A V-SP-1-92-06-066,Proposal submitted to California Air Resources Board, Sacramento, CA, USA, by AeroVironment, Inc., Monrovia, CA, USA. Anderson, G., K. Anderson, C. Nilsson, and J. Levin (1979) Chemosphere 10, 823-827. Andreae, M. O., R. W. Talbot, and S.-M. Li (1987) J . Geophys. Res. 92, 6635-6641. Arey, J., B. Zielinska, R. Atkinson, and A. M. Winer (1987) Atmos. Environ. 21, 1437- 1444. Arnts, R. R., and S. B. Tejada (1989) Environ. Sci. Technol. 23, 1428- 1430. Atkinson, R., J. Arey, A. M. Winer, and B. Zielinska (1988) Final Report under Contract No. A5-185-32, for California Air Resources Board, Sacramento, CA, by Statewide Air Pollution Research Center, University of California, Riverside, CA, USA. Atlas, E., and S. Schauffler (1991) Environ. Sci. Technol. 25, 61 -67. Bidleman, T. F., W. N. Billings, and W. T. Foreman (1986) Environ. Sci. Techno(. 20, I038 - 1043. Brown, R. H., and C. J. Purnell (1979) J . Chromatog. 178, 79-90. Brown, R. H., P. C. Cox, C. J. Purnell, N. G. West, and M. D. Wright (1984) in: Identijkation and Analysis of Organic Pollutants in Air. Lawrence, L. H. (ed.). Boston, MA: Butterworth Publishers, pp. 37 - 50. Brunnemann, K. D., W. Fink, and F. Moser (1980) Oncology 37, 217-222. Chuang, J. C., S. W. Hannan, and N . K. Wilson (1987) Environ. Sci. Technol. 21, 798-804. Ciccioli, P., R. Draisci, A. Cecinato, and A. Liberti (1987) in: Proceedings of the Fourth European Symposium on Physico-Chemical Behaviour qf Atmospheric Pollutants. Angeletti, G., Restelli, G. (eds.). Dordrecht; Reidel Publishing Co., pp. 133- 141. Cotham, W. E., and T. F. Bidleman (1992) Environ. Sci. Technol. 26, 469-478. Coulson, D. M., S. J. Selovcr, E. C. Gunderson, and B. A. Kingsley (1984) in: Identification and Analysis of Organic Pollutants in Air. Lawrence, L. H : (ed.). Boston, MA: Butterworth Publishers, pp. 51 -60. Coutant, R. W., L. Brown, J. C. Chuang, R. M. Riggin, and R. G. Lewis (1988) Atmos. Enuiron. 22, 403-409. Coutant, R. W., P. J . Callahan, M. R. Kuhlman, and R. G. Lewis (1989) Atmos. Environ. 23, 2205 - 221 1. Coutant, R. W., and W. A. McClenny (1991) in: Proceedings of the U . S . E P A / A & W M A International Symposium on Measurement of Toxic and Related Air Pollutants. Pittsburgh, PA : Air & Waste Management Association, p. 382. Coutant, R. W., P. J. Callahan, and J. C. Chuang (1992) Atmos. Environ. 26, 2831-2834. De Bortoli, M., H. Knoppel, E. Pecchio, A. Peil, L. Rogora, H. Schauenburg, H. Schlitt, and H. Vissers (1986) Enuiron. Int. 12, 343-350. Des Tornbe, K., D. K. Verma, L. Stewart, and E. B. Reczek (1991) Am. Ind. Hyg. Assoc. J . 52, 136 - 144.

182

B. Zielinska and E. Fujita

Druzik, C. M., D. Grosjean, A. Van Neste, and S. S. Parmar (1990) Intern. J . Environ. Anal. C h m t . 38, 495-512. Foreman, W. T., and T. F. Bidleman (1990) Atmos. Environ. 24A, 2405-2416. Freeman, R. R. (1992) in: Proceedings o f the Third Annuul West Coast Regional Air & Wuslc Munugement Association Conference On Current Issues in Air Toxics. Edwards. L. (cd.). Pittsburgh, PA: Air & Waste Management Association, pp. 48-53. Fujita, E., and B. Croes (1990) in: Troposplzeric Ozone and the Enoironrutent. Pittsburgh, PA: Air & Waste Management Association. Fung, K., and D. Grosjean ( I 98 I ) A n d . Chem. 53, 168. Fung, K.. and B. Wright (1990) Aerosol Sci. Technol. 12, 44. Fung, K., M. Porter, D. Fitz, and E. Fujita (1994) In: Planning and Managing Air Quality Modding and Measurement Studies: A Perspective Through S J V A Q S I A U S P E X . Solomon, P. A,, Silver. K.A. (eds.). Air and Waste Management Association, Pittsburgh, PA. Grbmping, A , , and K Cammann (1989) Fresenius Z . Anal. Chem. 335. 796-801. Grosjean, D. (1988) Atmos. Enuiron. 22. I637 - 1648. Grosjean, D., and K. Fung (1982) Anal. Chem. 54. 1221 - 1224. Grosjean, D., and K. Fung (1984) J A P C A 34, 537. Grosjean, D.. E. Williams, and A. Van Neste (1 988) Final Report qfAgreiwient A5-177-32, California Air Resources Board, Sacramento, CA, USA, September. Grosjean, D., and S. S. Parmar (1990) Environ. Sci. Techno/. 24, 102. Grosjean, D., A. Van Neste, and S. S. Parmar (1989) J . Liq. Chromutog. 12, 3007-3017. Gundel, L. A,, R. K. Stcvcns, J. M. Daisey, V. Lee, and K. R. R. Mahanama (1992) presented a t 11th Annual Meeting, American Association for Aerosol Research, San Francisco, CA, USA. Hawthorne, S. B., D. J. Miller, R. M. Barkley, and M . S. Kricger (1988) Emiron. Sci. Technol. 22, 1191. Hawthorne, S. B., M. S. Krieger, D. J. Miller, and M. B. Mathiason (1989) Environ. Sci. Ti,c.hnct/. 23, 470. Herlihy, L. J., J. N.Galloway, and A. L. Mills (1987) Atmos. Environ. 21, 2397-2402. Holdren, M., and D. L. Smith (1986) Final Report, prepureil under EPA Contract 68.02.4127, by Battelle Columbus Laboratories, Columbus, OH, USA. Holdrcn, M. W., J . E. Orban, C. W. Spicer, G. W. Keigley, D. P. Margeson, M. C. Matthews. A. J. Pollack, D. L. Smith, F. R. Todt, and G. F. Ward (1989) Evaluation and Improvement of Methods for [he Sampling and Analysis of Selected Toxic Air Contaminants, Draft Final Report, prepared for California Air Resources Board. Sacramento, CA, USA, by Battellc. Columbus, OH, USA. Hoyt, S. D., V. Longacre, and M. Stroupe (1993) in: Sampling und Anulysis of Airborne Pol1utunt.s. Winegar, E. D.. Keith, L. H. (eds.). Boca Raton, FL: Lewis Publishers, Chapter 9. pp. 3- 19. Intcrsociety Committee (198921) in: Methods of Air Sampling and Analysis. 3rd Ed., Lodge, Jr.. J. P. (ed.). Chelsa, MI: Lewis Publishers, pp. 293-295. Intcrsociety Commitee (l989b) in: hfel/70d.S .f' Air Sampling and Avdysis. 3rd Ed., Lodge, Jr.. J . P., (ed.). Chelsea, M I : Lewis Publishers, pp. 649-653. Jayanty, R. K . M. (1989) Atmos. Enuiron. 23, 177-782. Kaupp, H., and C . Umlauf (1992) Atmos. Enuiron. 26A, 2259-2267. Keene, W . C., R. W. Talbot, M. 0. Andrcae, K. Beecher, H. Berresheim, M. Castro, J. C. Farmer, J. N. Galloway, M. R. Hoffman, S.-M. Li, J. R. Maben, J. W. Munger, R. B. Norton, A. A. P. Pszenny, H. Puxbaum, H. Westbcrg, and W. Winiwarter (1989) J . Geophys. Res. 94, 6451 - 641 1. Kelly, T. J.. P. J. Callahan. J. Plell, and G. F. Evans (1993) Kninrliron. Sci. Technol. 27,1146 - I 1 53. Krost, K . J.. E. D. Pellizzari, S. G. Walburn, and S. A. Hubbard (1982) Anal. Chenz. 54,810- 817. Kuntz, R., W. Laumcman, G. Namie, and LA. A. Hull (1980) Anal. Lett. 13, 1409-1415. Kuwata, K.. M. Uebori. and Y. Yamasaki (1979) J . Chromatog. Sci. 17, 264-268. Kuwata, K., E. Akiyama, Y. Yamazaki, H. Yamasaki, and Y. Kuge (1983a) A n d <'hem. 55, 21 99 - 2201. Kuwata,K.,M.Uebori,H.Yamasaki,Y. Kuge,andY.Kiso (1983b)And. Chcm.55,2013-201h. Lane, D. A,, N. D. Johnson. S. C. Barton, G. H. S. Thomas, and W. H. Schroeder (1988) Enviroti. Sci. Technol. 22, 941 -947.

Organic Gus Sampling

I83

Hanley, W. H. Schroeder, and D. T. Ord (1992) Enuiron. Lane, D. A., N. D. Johnson, M. J. .I. Sci. Tpchnol. 26, 126 - 132. Lawson, D. R., H. W. Biermann, E. C. Tuazon, A. M, Winer, G. I. Mackay, H. I. Schiff, G. L. Kok, P. K. Dasgupta, K. Fung (1990) Aerosol Sci. Technol. 12, 64. Levaggi, D. A,, W. Oyung, and R. V. Zerrudo (1992) in: Proceedings of the 1YY2 U S . EPAIA & W M A International Symposium on Measurement of’ Toxic und Related Air Pollutants. Pittsburgh, PA: Air &Waste Management Association, p. 857. Levin, J.-0.. R. Lindahl, and K. Andersson (1988) Enuiron. Technol. Letters 9, 1423- 1430. Levin, J., D. Andersson, R. Lindahl, and C.-A. Nilsson (1985) Anal. Chem. 57, 1032-1035. Levin, J., R. Lindahl, and K. Andersson (1986) Environ. Sci. Technol. 20, 1273-1276. Ligocki, M. P., and J. F. Pankow (1989) Environ. Sci. Technol. 23, 75-83. Lipari, F., and S. J. Swarin (1985) Enuiron. Sci. Technol. 19, 70-74. McClenny, W. A., D. P. Joachim, G. F. Evans, K. D. Oliver, M. W. Holdren, and W. T. Winberry (1991) A & W M A , 1308-1318. McDow, S. R., and J. J. Huntzicker (1990) Atmos. Enuiron. 24A, 2563-2572. Namiesnik, J. (1988) Talantu 35, 567- 587. NIOSH (1 984) Manual qfAnal.yticul Methods: U.S. Department of Health, Education, and Welfare. NIOSH; Washington, D C : GPO. Norton, R. B. (1985) Geopliys. R P ~Lett. . 12, 769-712. Norton, R. B. (1992) J . Geophys. Res. 97, 10,389-10,393. Offerman, F. J., S. A. Loiselle, J. M. Daisey, A. T. Hodgson, and L. A. Gundel (1990) Final Report, preparedunder Contruct N o . A732-106, for California Air Resources Board, Sacramento, CA, USA. Oliver, K. D., J. D. Pleil, and W. A. McClenny (1986) Atmos. Enuiron. 20, 1403. Otson, R., and P. Fellin (1988) Sci. Total Enuiron. 77, 95- 131. Otson, R., and P. Fellin (1991) in: Proceedings o f t h e EPAIA & W M A International Symposium on Measurement of Toxic and Related Air Pollutants. Pittsburgh, PA: Air & Waste Management Association, p. 291. Pankow, J. F. (1992) Atmos. Environ. 26A, 2489-2497. Parmar, S. S. (1991) in: Proceedings of the U.S. E P A / A & W M A International Symposium on Measurement Toxic and Relaled Air Pollutants. Pittsburgh, PA: Air & Waste Management Association, p. 544. Pate, B., R. K. M. Jayanty, M. R. Peterson, and G . F. Evans (1992) J A P C A 42, 460-462. Pierotti, D. (1990) J . Atmos. Chem. 10, 373-382. Pellizzari, E., B. Demian, and K. Krost (1984) Anal. Chem. 56, 793-798. Pollack, A. J., M. W. Holdren, and W. A. McClenny J . A & W M A Note-Book. September 1991, 41, NO. 9, 1213-1216. Puxbaum, H., C. Rosenberg, M. Gregori, C. Lanzerstorfer, E. Ober, and W. Winiwarter (1988) Afmos. Enuiron. 22, 2841 -2850. Rasmussen, R. A. (1992) presented a t International Symposium on Measurement of Toxic and Related Air Pollutants, Durham, NC, USA, cosponsored by U S . Environmental Protection Agency, Research Triangle Park, NC, USA, and Air & Waste Management Association, Pittsburgh, PA, USA. Rosenberg, C., W. Winiwarter, M. Gregori, G. Pech, V. Cascnsky, and H. Puxbaum (1988) Fresenius Z. Anai. Chem. 331, 1-7. Rothweiler, H., P. A. Wager, and C. Schlatter (1991) Atmos. Enuiron. 25B, 231 -235. Rudling, J., E. Bjorkholm, and B.-0. Lundmark (1986) Ann. Occup. Hyg. 30, 319-327. Rudolph, J., K. P. Muller, and R. Koppmann (1990) Anal. Chim. Acta 236, 197-211. Sexton, K., K.-S. Liu, and M. X. Petreas (1986) J A P C A 36, 698. Shields, H. C., and C. J . Weschler (1987) J A P C A 37, 1039-1045. Siegl, W. O., F. F. 0. Richert, T. E. Jensen, D. Schuetzlc, S. J. Swarin, J. F. Loo, A. Prostak, D. Nagy, and A. M. Schlenker (1993) S A E Technical Paper Series, #930142, Detroit, MI, March 1-5. Warrendale, PA: SAE International. Sonnefeld, W. J., W. H. Zoller, and W. E. May (.1983) A n d y . Chem. 55, 275-280. Soroka, J. M., R. Isaacs, G. Ball, R. Singhvi, and T. Pritchett (1992) in: Proceedings qf’the 1992 U.S. E P A l A & W M A International Symposium on Measurement of’ Toxic and Related Air Pollutants. Pittsburgh, PA: Air & Waste Management Association, p. 532.

184

B. Zielinska and E. Fujita

Supelco, Inc. (1986) GC BuNetin 846C, Supelco, Inc., Bellefonte, PA. Supelco, Inc. (1988) Efficiently Monitor Toxic Airhorne Compourds by Thermal Desorption. Bellefonte, PA: Supelco Technologies. Tang, Y.-Z., Q. Tran, and P. Fellin (1993) Anal. Chem. 65, 1932-1935. Tanncr, R. L., A. W. Gertler, E. M. Fujita, and J. C. Chow (1993) Atmospheric Organic Acids A Review. Submitted to Atmos. Enuiron. Tejada, S. B. (1986) Int. J. Environ. Anal. Chem. 26, 167-185. U.S. Environmental Protection Agency (1988) in: Compendium of Methods,for the Determination of’ Toxic Organic Compounds in Ambient Air. EPA/600/4-8910 17, Research Triangle Park, NC, USA: U.S. Environmental Protection Agency, Methods TOI, T02, T03, T04, T07, T08, T 0 9 , T010, TOII, T012, T013, and T014. U S . Environmental Protection Agency (1991) EPA/600-8-91/215, U S . Environmental Protection Agency, Research Triangle Park, NC, USA. Vairavamurthy, A,, J. M. Roberts, and L. Newman (1993) in: Sainpling and Analysis of Airhorne Pollutants. Winegar, E. D., Keith. L. H. (eds.). Boca Raton, FL: Lcwis Publishers, Chapter 10, pp. 149- 174. Wagoner, D. E., S. Rice, and S. T. Bursey (1993) in: Sampling and Analysis of Airborne Pollutants. Winegar, E. D., Keith, L. H. (eds.). Boca Raton. FL: Lewis Publishers, Chapter I , pp. 3- 19. Walling, J. F., (1984) Atmos. Enuiron. 18, 855-859. Walling, J. F., S. E. Bumgarner, D. J. Driscol. C. M. Morris, A. E. Riley, and L. H. Wright (1986) Atmos. Enuiron. 20, 52 - 57. Winiwarter, W., H. Puxbaum, S. Fuzzi, M. C. Faccini, G. Orsi, N. Beltz, K. Enderle, and W. Jaeschke (1988) Tellus 40B, 348- 357. Yamasaki, H., K. Kuwata, and Y. Kuge (1984) J . Chem. Sor. Japan 8, 1324- 1329. Zielinska, B., J. Sagebiel, L. H. Sheetz, G. Harshfield, A. W. Gertler, and W. R. Pierson (1993) in: Proceedings of the U.S. E P A / A & W M A International Symposium on Measurement of Toxic and Relnted Air PoNutants, Pittsburgh, PA, USA: Air & Waste Management Association, p. 123- 128. Zielinska, B., and K. Fung (I992) Composition and Concentrations o f Semi- Volatile Hydrocarbons, Final Report for the California Air Resources Board, Sacramento, CA, USA. Zielinska, B., J. Arey, T. Ramdahl, R. Atkinson, and A. M. Winer (1986) J. Chromatog. 363, 382- 386.

B. Water

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

8 Sampling of Freshwaters for Estimation of all Detectable Elements Ursula M . Cowgill

8.1 Introduction This chapter concerns problems associated with the sampling of water from lakes, rivers and rain as well as difficulties encountered in sampling water of various physical states, e.g., ice, snow, and fog or dew. The first part of this chapter is devoted to the types of contamination found due to the use of sampling tools of a variety of composition and sorption and leaching of inorganic substances from such materials. The second part contains cautionary comments when sampling for particular elements. The last part is devoted to variations in chemical composition in relation to the number of replicates, frequency in sampling and sampling location in various shallow and deep water bodies. Examples from lakes, rivers, groundwater, ice, dew and chemical variations during a 3-h rainstorm are presented. Since very few studies have been devoted to the variability of results in relation to sampling protocols for trace organic compounds, all examples used in this chapter will involve inorganic studies (from Cowgill, 1970, 1976, 1980, 1988a). Good general reviews of sampling for chemical analysis are provided by Kratochvil and Taylor (1982) and Kratochvil et al. (1984). In addition, Keith (1981 a, b, c) has devoted some discussion to the problems of sampling for trace organic compounds.

8.2 Problems Associated with Sampling Usually the purpose of sampling any body of water is to obtain a representative sample. If only general chemistry of major elements is of interest, then frequency of sampling, the size of sample required, and the equipment to be used becomes less important than were trace quantities of some element and its seasonal variation of interest. Many of the data presented here stem from preliminary studies that were carried out to discover what the best method of sampling might be. Samples had to be obtained so that precise measurements of major and minor constituents could be achieved in water, ice, dew, snow, rain, mud, plants, and plankton. Examples from such preliminary studies will be used to illustrate samping pitfalls.

188

U. M . Cowgill

8.2.1 Contamination from Sampling Devices and Laboratory Equipment Serious errors may originate during sampling and storage due to contamination from sampling devices. Inadequate washing of the sampling devices and thus failure to adequately remove pollutants from previous samples will bring about erroneous results. In addition, loss of various elemental quantities or reduction of the original concentrations due to sorption on the sample container or precipitation of elements owing to failure to carry out necessary pretreatment, such as pH adjustments in the case of metals may make all the effort taken to obtain proper samples for nought. Tab. 1 shows the types of pollutants that may be contributed to water samples by materials used in sampling devices, well casings and laboratory equipment. PVC-threaded joints or PVC-cemented joints (polyvinyl chloride) may have contaminants that leach from the plastic pipes into the water they hold. These pollutants may be removed by steam cleaning where the source of the steam is distilled water (Cowgill, 1988b). Five separate steam cleanings will remove contaminants below their detection limits (<0.2 pg/L). A similar procedure will remove pollutants leached into water from polypropylene or polyethylene. Although the samples of teflon or fiberglass-reinforced epoxy material tested, no detectable substances were found, it is still wise to steam clean prior to use. Plastic pipette tips should be soaked in 5 N HCl (ultrapure acid) for two days and then rinsed thoroughly with double distilled water. Stainless steel sampling devices should be steam cleaned and used only for water samples intended for organic analysis. Glass and quartz should not be used if Si or B are to be sought in the collected sample. Tab. 1. Contaminants Orginating from Various Sampling Devices, Well Casings and Laboratory Equipment (from Cowgill, 1988a, b, unpublished data) Material

Contaminants found prior to steam cleaning

PVC-threaded joints PVC-cemented joints

choroform methylene chloride, chloroform, acetone, tetrahydrofuran, methyl ethyl ketone, ethyl acetate, benzene, cyclohexanone, toluene, polyvinylchloride, 3 organic Sn compounds (Boettner et al., 1981) Zn, Fe, Sb, Cu (Robertson, 1968) plasticizers, phthalates, Sb nothing detectable nothing detectable Si Sn, Pb, Sb Cu, Fe, Zn, Cd, Ni Cu, Ni, Zn, P, K Cr, Fe, Ni, Mo B, Si

PVC Pol ypropylene/polyethylene Polytetrafluoroethylene (Teflon) FRE (fiberglass reinforced epoxy) Quartz Soldered pipes Plastic pipette tips 0.45 p n filter cellulose acetate Stainless steel Glass

Sampling of Freshwaters

189

Contamination resulting from materials used in sampling devices, well casings or pipes used to transport water are not minor. In the case of monitoring wells constructed from PVC-cemented joints, pollutants may be found in the groundwater as much as a decade after construction (cf. Cowgill, 1988b). Steam cleaning and aeration (air administered through a trap) are reliable cleaning procedures (cf. Cowgill, 1988b). Similarly, large amounts of water must be transported through pipes before the contaminants are removed to below the detection limit (Cowgill, 1988a).

8.2.2 Sorption and Leaching of Pollutants by Sampling Tool Materials Tabs. 2 and 3 show sorption of halogenated hydrocarbons from various substances as well as the actual loss of hydrocarbons from solution. Organic carbon effects sorption and this effect is material dependent (Barcelona et al. 1985a, b). This depletion is more dependent on the type of substance than on tube diameter (Barcelona et al., 1985b). But, when the rate of flow is constant, losses increase as the tube diameter increases (Barcelona et al., 1985a, b). Sorption is a function of mass as well as of surface area. Clearly, the use of teflon or PVC compromises the value of total organic carbon as a pollutant indicator. This has not been found to be the case for FRE. Thermoplastic materials sorb many priority pollutants efficiently. This effect may be so pronounced that the detection of pollutants may Tab. 2. Maximum Sorption of Dilute (400 pg/L) Halogenated Hydrocarbon Mixture in Water by Various Substances

Sampling

PVC Teflon FRE

Sorption of halogenated hydrocarbons pg/m2

Length of exposure h

622 237 203

1 1 72

Tab. 3. Actual Loss of Hydrocarbons from Dilute (400pg/L) Halogenated Hydrocarbon Mixtures in Water by Various Substances (from Hunkin, cf. Cowgill 1988b) Substance FRE Teflon PVC

Sorption %

Length of exposure, h

20.25 38 98

72 1 1

Investigator

Barcelona et al. (1985a, b) Barcelona et al. (1985a, b) Hunkin (cf. Cowgill, 1988b)

190

U . M . Cowgill

be delayed when synthetic sampling tools are suspended in wells and are in contact with well water for extended periods of time or when such tools are employed for continuous sampling of surface waters. Barcelona et al. (1985a, b) found that teflon caused an adsorption loss of chlorinated hydrocarbons of 21% in less than 1 h while similar losses were observed by Hunkin (cf. Cowgill, 1988b) for FRE in 72 h. Adsorption of metals at low concentrations on container walls is determined by the pH of the water, the time the sample is in contact with the container, the concentration of the metal, the chemical composition of the container, the presence of complexing agents and dissolved organic carbon (Miller, 1982). Miller (1982) also noted that the adsorption and subsequent leaching of Pb was increased by Cr(V1) and six volatile compounds than when Pb and Cr(V1) were alone in the solution. Cr(V1) neither was adsorbed or leached from PVC, polyethylene or polypropylene. Precautions are also necessary when determining Ag concentrations. In the event that adsorption to container walls is thought to have occurred, the sample should be made basic with concentrated NH,OH to which 1 mL cyanogen iodide has been added. The resulting solution is mixed and permitted to stand 1 h. The main purpose of this discussion is to emphasize that the composition of tools employed in field work requires serious attention when samples are collected with the intent of trace analysis in mind.

8.2.3 Replication The purpose of sampling determines the number of replicates needed to characterize a water body. Sampling may be for descriptive purposes alone (i.e., the chemical composition of the water body) or for monitoring (i.e.,the substances to be monitored have been previously identified). Replication in the latter case must be sufficient to obtain acceptable (i.e., small standard deviations) arithmetic means and confidence limits. Initially, sampling for the purpose of monitoring requires the knowledge of chemical composition of the water body, of only for background information. Replication as it is used here should be defined. For example, analysis of three aliquots from one liter will yield different results than if three separate liters are collected and an aliquot from each is analyzed. The first method is nothing more than an instrument check. Such data are nonrandom and biased i n a statistical sense from the standpoint of the water body, but not from the standpoint of evaluating the entire analytical method. The second method represents true replication. Variation encountered among the three aliquots obtained from three separate liters is the experimental sampling error. This type of replication provides the measure of sampling precision. The built-in randomization ensures the validity of this measure of precision. Replication of this type will illuminate gross errors in analytical measurement while the results from three aliquots from 1 liter will not. Analytical results of the three aliquots per liter method will provide smaller coefficient of variation than the analytical results of the aliquot from cach of three liters (McBean and Rovers, 1985). However, the latter approach over time will

Sumpling ~f Freshwaters

191

provide the chemical description of the water body which is the ultimate objective of sampling. Tab. 4 illustrates this point with some results from Linsley Pond (North Branford, Ct, USA; Cowgill, unpublished data 1965- 1980). Tab. 4 shows that the coefficient of variation is always less when three aliquots per liter are analyzed than when one aliquot from each of three one liter samples analyzed. The data also reveal that the mean concentration of substances are neither consistently higher or lower but as the quantity of a particular element declines the variation between the two sampling regimens increases. Statistically significant differences are observed between the two systems. The true value is not revealed by either sampling regimen. For further discussion of this point, see McBean and Rovers (1985). Tab. 4. Chemical Results Obtained from Various Sampling Techniques on Linsley Pond Depth sampled: surface -2.5 m; values for Si, Mg, K in mg/L, for Sr, Rb, P in pg/L (from Cowgill, 1970, 1988a)

Number of subsamples replicated ten times 3 aliquots/L Std. deviation Cod. of variation One aliquot from each of 1 1 samples Std. deviation Coef. of variation xz,P C

Si

Mg

K

76 10.1 13.3

30 8.3 27.7

8.2 5.0 61

23 4.2 18.3 0.001

11.8 3.8 32.2 0.005

1.1 0.8 72.2 0.02

Sr

Rb

P

130 20 15.4

32.3 25 71.4

20.4 5.1 25

8.5 8.0 94.2 0.001

72 30 41.7 0.001

64 20 31.3 0.001

Tab. 5 shows the effect on concentration of various elements by taking larger samples. All samples were gathered from each side of a stationary boat with a Van Dorn bottle fitted with a graduated steel tape (Cowgill 1970, 1988a). For further discussion of limnological sampling equipment, see Lind (1974). It is clear that to avoid differences between samples gathered from either side of the boat, larger samples must be collected. It should be noted that results obtained from 20 L or 30 L samples taken from either side of the boat are, from a statistical viewpoint, not different than those from the subsamples of each of three 1 L samples replicated ten times. In addition, it may be noted that no difference, in a statistical sense, occurs between 20 L and 30 L samples regardless of which side of the boat the samples originate. Smaller samples gathered under the same circumstances have too much variation in analytical results to be acceptable. These examples come from the worst case possible, namely (surface -2.5 m) the euphotic zone which supports many living things. In the case of sampling for monitoring purposes it is assumed that the data exhibit a normal distribution over time. As a rule, the estimate standard deviation is set within a certain percentage of its true value at a particular confidence Ievel. The investigator may specify the coefficient of variation that will be tolerated at a particular confidence limit (cf. Berg, 1982). The concentration of the substance of interest determines the volume of sample to be collected, which must be of a size to avoid the north-south side of the boat dilemma.

192

U . M . Cowgill

Tab. 5. Chcmical Results Obtained by Collecting Samples from the North and South Side of the Boat (for conditions, see Tab. 4)

Sample size

Si

3 2-L samples Std. deviation Coef. of variation 3 5-L samples Std. deviation Coef. of Variation 3 20-L samples Std. deviation Coef. of variation 3 30-L samples Std. deviation Coef. of variation

74

3 2-L samples Std. deviation Coef. of variation 3 5-L samples Std. deviation Coef. of variation 3 20-L samples Std. deviation Coef. of variation 3 30-L samples Std. deviation Coef. of variation

100 14 14 76 12 15.8 29.2 4 13.7 26.5

Mg

K

Sr

Rb

P

10

155 25 16.1 120 21 17.5 75 10 13.3 70 10 14.3

32.3 25 77.4 25 18 72 9.3 6 65.2 8.8 5.0 56.8

100

95

45 30 66.7 32 25 78.1 10 6.7 67 9.2 5 54.4

135 17 12.6 85

South side

10 13.5 62

56.2 10 17.8 22

11

10

5

17.7 25 4 16 29 6 13.11

45.5 12 2.1 17.5 14 2.1

60.2 1.6 0.8 50 1.5 0.5 33.3

15

5 50 8.3

25 25 110

25 22.7 75 10 13.3 80 8 10

North side

3

11.3

30 10 33.3 18 5

27.8 11 2 18.2 15 2.4 16

20 5 25 10.3 5.5 53.4 1.9 0.5 26.3 1.7 0.4 23.5

15

15.8 104 15

14.4 68 9 13.2 72 11 15.3

5 9.4 12 9

12.5 78 9.3 11.9

8.2.4 Frequency of Sampling The purpose of sampling determines the frequency of sampling. If seasonal variation is of interest, then sampling must be at least weekly. Tab. 6 shows some of the variation that did occur in Linsley Pond during a 53-week study (Cowgill, 1970, 1976, 1988a). Although these variations have been described in detail elsewhere but refer to the period of human disturbance, the chemical composition of ice, the chemical composition of the density current, the thawing of the ice, the periods of vegetative growth and decay and finally the algal bloom. Frequency of sampling when the purpose is monitoring, is usually specified by the permit. The rule of thumb is that the length of the record should be at least ten times as long as the period of interest (Berg, 1982). If the longest period of interest is one year, then ten years of data would be required.

Sampling of Freshwaters

193

Tab. 6. Weekly Changes in some Elemental Quantities in Various Strata Observed during a 53-Week Study in Linsley Pond Figures in parentheses denote standard deviation; data are presented in total quantity of the strata noted, values in kg and for Hg in g (from Cowgill, 1970, 1988a) Date

Strata m

1965 Sept. 14

0- 14.5

Sept. 20

0- 14.5

1966 Nov. 1

8-11

Nov. 8

8-11

Feb. 6

0- 14.5

Feb. 13

0- 14.5

Feb. 20

0- 14.5

June 26

0-

2.5

July 3

0-

2.5

Ti

3 (0.2) 26 (2)

Fe

Mg

180 (10) 1100 (50)

Hg

Mn

Br

200 (15) 900 (90) 198 (10) 42 (5)

8.2.5 Equipment, Field, and Sampling Blanks Three types of blanks are necessary. Their subsequent analysis may explain erratic analytical results. Such data may identify unsuspected pollutants associated with distilled water purity, inproper dish washing, latent air contaminants that may have been sorbed by samples during collection or contaminants associated with travel. Groundwater samples often sorb contaminants present in the air during the process of gathering the samples. The types of blanks required for quality assurance are equipment, field, and sampling blanks. Equipment blanks are obtained by soaking all sampling equipment in double distilled (glass) water. This soaking water should be stored in glass vessels for analysis. Prior to sample collection, all equipment employed to collect the sample must be soaked again in double distilled (glass) water. This soaking water is stored in glass vessels and is labeled field blank. The sampling blank is obtained by exposing double distilled (glass) water to the atmosphere for the entire period during which the sample is being gathered. All blanks should be sealed, labeled and stored at 4 "C for analysis.

I94

U . M . Cowgill

8.3 Sample Fractions and Sample Preservation When the object of sampling is to examine the state of the substance being sought, pretreatment methods arc required and are given in the following paragraphs. -

~

-

~

Total metals: The concentration of metals analyzed in an unfiltered sample after it has been digested. Total metals may also be calculated by adding the dissolved and suspended fractions together (American Public Health Association, 1985). Dissolved metals: Those substances (metals) of an unacidified sample that pass through a 0.22 pm membrane filter. Experience has shown that more consistent results are obtained when a 0.22 pm filter is used than when 0.45 pm is employed. Suspended metal: Those substances in an unacidified sample that are retained on the 0.22 pm membrane filter. Acid-extractable metals: Concentration of substances after the unfiltered sample has been treated with hot dilute mineral acid, usually nitric (American Public Health Association, 1985).

8.3.1 Preservation of Samples Once it has been decided which fraction is to be analyzed-dissolved, suspended, total or acid-extractable, it will be known how the sample is to be treated. The best sample containers are made of teflon, quartz, or FRE. Samples that do not require treatment should be stored at 4 "C as soon as they have been collected. Samples for metal analysis should be acidified with concentrated nitric acid to a pH of less than 2. Samples to be analyzed for dissolved metals should be filtered before acidifying. The acid used should always be ultrapure. After preserving the samples they should be stored at 4 "C. As a rule, 1.5 mL HNO, per L of sample is sufficient unless the sample is highly buffered, alkaline, or contains much sediment in which case as much as 5 mL may be necessary. Highly buffered samples should have their pH checked frequently to ascertain that the pH is remaining at less than 2. Samples should be analyzed as soon as possible, This is especially truc for samples containing microgram levels of metals. For samples containing milligram levels of metals stability is possible for six months (American Public Health Association, 1985).

8.3.2 Pretreatment, Storage, and General Precautions All equipment that is in contact with the sample should thoroughly be clcaned and washed with 6 N HCl. It should then be rinsed and autoclaved. This procedure removes all acid effectively. If for some reason the equipment is not amenable to acid washing steam cleaning is equally effective.

Sampling of Freshwuters

195

Membrane filters should be soaked in 5 N HCI and then washed thoroughly with double distilled (glass) water before filtering samples. Great effort should be exerted to avoid contaminating the sample. Only double distilled (glass) water should be used. Only ultrapure acids should be employed. It is wise when dealing with samples containing microgram values of metals to carry out all sample preparation in a laminar flow hood as dust, soot and airborne particles in general contain metals. Tab. 7 contains a list of precautions to be taken with some elements and elemental states, Most of the suggestions under storage concerning the material content of the storage containers are made to avoid excessive absorption of the element under study.

Tab. 7. Some Cautionary Comments on Storage and Preservation for some Elements (American Public Health Association, 1985) Element

Storage and other precautions

Ag B Cr cu

light-absorbing containers B-free containers Avoid use of K,Cr,O, dish washing analyze promptly due to sorption on surface of containers polyethylene containers to avoid contamination Glass bottles, colloidal Fe adhers to plastic surfaces 5 weeks-glass; 2 weeks plastic polyethylene containers analyze promptly on analysis pH 7 KOH/NaOH analyze promptly

F Fe Hg K Mn Ammonia/ organic N Nitrate/ nitrite Na P Pb sulfide sulfite sulfate Si Sr U Zn

polyethylene containers glass containers; no detergent use in dish washing avoid loss of organolead compounds collect, handle with minimum agitation fill bottle to top

Preservation

0.5 mL 6NHCl per 100 mL of sample

HNO, to pH < 2 HNO, to pH < 2 conc. H,SO,/L sample to pH 1.5-2.0 Freeze - 20 "C or 40 mg HgCIJL 4 "C 1 mL conc. HCl Freeze - 20 "C or 40 mgHgCI,/L at 4 "C HNO, to pH < 2 5 mL 0.1 N iodine solution 4 drops 2 N zinc acetate/100 mL sampi 1 mL 2.5% EDTAil00 mL sample

4 'C to curtail bacterial action polyethylene containers polyethylene containers

analyze within 6 h-if not possible prior to analysis after acidifying evaporate to dryness in silica dishes to remove acid

HNO, to pH < 2 HCI/HNO, to pH 2 Zn-free HCI to pH 2

U . M . Cowgill

196

8.4 Sampling of Lakes, Rivers, and Groundwater This section is devoted to problems associated with lakes, rivers, and groundwater. In many bodies of water, the stratum from which a sample is taken affects the cheniical composition of the sample. Data presented in Tabs. 8- 12 illustrate the point.

8.4.1 Stratified Bodies of Water Bodies of water that are deeper than 5 in tend to thermally and chemically stratify. Thus for purposes of monitoring, the stratum and the exact position of sampling must be strictly adhered to for the entire sampling period. Tab. 8 depicts data from Linsley Pond. The examples shown represent the height of the density current (Feb. 20), the spring homothermal period (March 13 + 20) and July and August, the period of eutrophication (anoxic mud surface). In the case of rivers (Tabs. 2 -- 10) chemical stratification may occur unaccompanied by any thermal stratification. The example shown here is the Ohio River at Pittsburgh, PA which is a rapidly flowing river of considerable depth.

8.4.2 Unstratified Bodies of Water It is a rule of thumb that bodies of water that are less than 5 m deep are mixed by wind action. Thus, stratification usually fails to become established in such waters. Tab. 9 shows the kind of data to be expected in unstratified lakes such as Cedar Tab. 8. Variation in Chemical Composition in Relation to Sampling Position in Stratified Lakes Figures in parentheses denote standard deviation; data from 1966; values for Ca in mg/L, for Fe, Mn. P in pg/L (from Cowgill, 1970, 1976, 1988a) Date

Strata

Ca

FC

Mn

m Feb. 20

March 13

July

+ 20

+ August

Linsley Pond (stratified) 0 - 2.5 27 400 (3) (38) 8-11 30 5000 (3) (450) 0-2.5 28 480 (3) (50) 8-11 29 900 (3) (83) 0-2.5 31 300 (4) (35) 8-11 34 2700 (2) (100)

300 (35) 1000 (83) 250 (24) 2000 ( 100) 400 (43) 6000 (550)

P

Suniphng a j Fwshwurcw

197

Tab. 9. Variation in Chemical Composition in Relation to Sampling Position in Unstratified Lakes Figures in parenthescs denote standard deviation; data from 1966; values in pg/L (from Cowgill, 1970, 1976, 1988a) Date April 17

Strata m

Ti

P

0 -2.5

4.1 (0.1) 4.2 (0.8)

32 (3) 34 (3)

2.5 - 5 July 17

0 -2.5 2.5 - 5

Aug. 7

0 -2.5

2.5-5

2.2 (0.4) 2.1 (0.51 2.0 (0.3) 1.9 (0.2)

15

(2) 16 (2) 16 (2) 15 (2)

Lake (North Branford, CT, USA). Rapidly flowing shallow rivers such as the Jordan river in northern Israel fails to show any consistent chemical stratification (Tab. 10). When sampling such waters it is important to be familiar with the morphometry and morphology of the water bodies before establishing a sampling regimen. Tab. 11 depicts the kind of variation that may be encountered in a 33-m deep well that had 15 m of the casing screened. It may be noted that Mo and V appear to stratify at the 29-31 m level but Hg and Be are highly concentrated at the 24-26 m level. The reason for sampling this well was to discover if any chemical interchange was proceeding with nearby lakes. Prior to all sampling, at least ten Tab. 10. Variation in Chemical Composition in Relation to Sampling Position in Shallow and Deep Rivers Figures in parentheses denote standard deviation; values for Fe and K in pg/L, for Ca and Na in mg/L (from Cowgill, 1980, 1988a) Date

Strata m

June - Aug. 1963

0-0.5

1970- 1978 biweekly

Fe

K

Ca

Jordan River, lsracl 934 0.82 (88) (0.1) 2-2.5 950 0.88 (85) (0.1)

51 (5) 52 (5)

Ohio River, Pittsburgh, PA 0 - 0.5 1200 2.1 (200) (0.1) 8- 10 4300 4.2 (400) (0.3)

55 (6) 80 (7)

Na

I98

U. M . Cowgill

Tab. 11. Variation in Chemical Composition in Relation to Sampling Strata in a 33-m Drinking Water Well Containing a 15-m Screen Figures in parentheses denote standard deviation; values in pg/L (from Cowgill, 1988a) Strata m

19-21 24 - 26

29-31

Mo x

15 (2.5) 10

(2) 28 (3.5)

v

10-4

0.8 (0.2) 30 (4) 1115 (97.3)

Hg 1.6 (0.4) 18.3 (1.7) 3.0 (0.5)

Be x lo-’

0.7 (0.2) 18 (3) II (1)

bore volumes were removed. Each bore volume was studied for its pH, conductivity and temperature. When these three variables became stable after freshwater had entered the casing as signified by a drop in temperature, sampling at various depths within the secreened area was carried out. To obtain samples at various depths a 2-m long closed bailer was employed (Gibb et al., 1980). The variation in well water composition is natural, not due to stratification but due to the variation in substrata. These data were collected over a 14 year period and during this time the observed variability was reasonably consistent. No coefficient of variation exceeded 28.6.

8.5 Sampling of Ice, Snow, Rain, Dew, and Fog This section is devoted to the problems associated with the physical state of the sample. The position of sample collection is also important in the gathering of ice. In the Linsley Pond study (Cowgill, 1970, 1976), the pond was ice-covered for seven weeks and during that time the thickness varied between 5 and 25 cm in the open water. It was discovered that the chemical composition varied significantly in a vertical direction but little variation was noted in a horizontal direction. To obtain enough material to make chemical analysis precise, ice was gathered weekly so that on melting, there were 250 - 300 L of water. Tab. 12 shows some elemental quantities in ice. There is very little comparative data on the chemical composition of ice in lakes. The chemical composition of ice varies according to the chemical composition of the surface water from which ice forms, entrapped dust that contributes to the quantity of Fe, Ti and Mo (Cowgill, 1970, 1976), entrapped plankton and the rate of formation. Relative to the concentration in the surface water, Ca, Si, Al, and P concentrate in the ice but Fe and Ti have the same quantity in ice as in surface water. How ice is to be sampled depends largely on the purpose of sampling. If it is desireable to learn the geochemical effect of melting on the receiving water, bulk sampling of ice in a number of places on the lake will provide adequate results. However, if it is of interest to learn how the surface water composition is related to that of the ice in contact with it, then the ice in a series of strata should be sampled and the ice samples’ composition compared with that of the surface water.

199

Sampling of Freshwaters

Tab. 12. Variation in Chemical Composition of Ice in Relation to its Thickness and the Chemical Composition of Surface Water under the Ice Figures in parentheses denote standard deviation; values in pg/L (from Cowgill, 1988a) Depth, cm

0-

5

5-10 10-15 15-20

P

Ca

Si

Al

Fe

Ti

29000 (1 500) 27 500 (2000) 28 000 (2000) 28000 (1000)

30000 (3000) 33000 (3000) 32000 (3000) 25000 (I 500)

800 (80) 600 (65) 800 (80) 500 (55)

600 (55) 500 (60) 550 (60) 430 (40)

18 (2) 22 (2) 18 (2) 12 (1)

(4)

180

600 (60)

18 (2)

(1)

38

(4) 42 (4)

50 (5)

40

Water

0-

1

1800 150)

4800 (300)

(15)

9

In severe winters, heavy snow falls can contribute significantly to the elemental composition of a water body. The chemical composition of snow changes on standing and thus snow should be sampled during a storm. If a long period of time has passed since the last incidence of atmospheric precipitation, the snow will be more concentrated than had there been snow within the last few hours. As the storm proceeds, the chemical composition of the snow becomes more and more dilute as it cleans out the atmosphere until finally it resembles a poor grade of distilled water containing easily detectable amounts of C1, Br, I, S, B, Na, K, Ca, Mg and N. If sampling is carried out shortly after a 0.5-m snowfall, there is a chemical difference between that which fell initially and that which fell recently. This of course refers to snow falling within 100 km of industrialized areas. High altitude snowfalls are much more dilute. Tab. 13 shows the chemical results of a 3-h rainfall collected on the roof of a 12-storey building located in Pittsburgh, PA, USA in March 1978. Rain was collected in teflon rain collectors that were covered with very fine previously washed nylon netting to curtail the entrance of airborne particles into the samples. Collections were made at ten-min intervals. Some of the Ca and Mg in the rain is due to limestone dust and some of it is a contribution from the local cement industry. Sulfate and N are still very much more concentrated in the falling rain than the comparative data of uncontaminated rain. Like snow, rain cleans out the atmosphere reasonably well. The point to be made is that pH and elemental composition must be reported in terms of the time the sample was taken during the rainstorm and the date. The composition and the pH varies with time. Mean pH or elemental composition of a rainstorm is meaningless. Much discussion has occurred on the sampling of precipitation (cf. Asman and Jonker, 1982; Campbell and Scott, 1985; Raynor and Hayes, 1983). Tab. 14 shows the chemical composition of three strata of dew extracted from a grass field. No physical state of water is more difficult to sample. A glass vacuum pump was constructed to collect dew from three arbitrarily selected strata of a

200

U. M . Co~t~gill

Tab. 13. Variation in Chemical Composition ofa 3-h Rainstorm in Pittsburgh, PA in March 1978 The composition of uncoiitaminatcd rain (Gorham, 1955) is added for comparison. Figurcs in parentheses denotc standard deviation; values in mg/L (from Cowgill, 1988a) Time (min)

0-

10

PH

N

CI

1.2

4.7 4.2 (1) 3.0 (0.5) 1.8 (0.5) 0.9 (0.1)

10.1 (2) 8.8 (2) 6.5 (1) 4.6 (1) 4.2 (1)

<0.009

5.1

(1)

10- 20 50- 70

1.5 2.4

120-150

3.9

170-180

5.2

so4

B

29.3 0.18 (4) (0.03) 25.8 0.15 (4) (0.03) 0.12 18.8 (4) (0.02) 12.0 0.08 (0.01) (3) 7.6 0.08 (0.01) (2) uncontaminated rain 1.7

Na

K

Mg

Ca

5.8

0.6 (0.1) 0.5

3.1 (0.6)

3.1 (0.05) 2.4 (0.05)

0.4 (0.1) 0.3 (0.1) 0.3 (0.1)

2.8 (0.5) 2.4 (0.5) 1.8 (0.5) 1.0 (0.2) 0.4 (0.1)

3.1

0.2

0.3

0.2

(1)

5.2 (1) 3.7 (1)

(0.1)

2.7 (0.5) 1.9 (0.5) 1.4 (0.5) 0.3 (0.1)

10 acre (4.05 hectare) grass field. Each stratum yielded about 1 L of dew. Three replicates of each stratum were analyzed. To be able to detect K, C1, SO, and N, it was necessary to concentrate the dew samples over P,O,. As may be noted from Tab. 14 the coefficient of variation is unacceptable, varying from 42% to 54%. The lower stratum undoubtedly is influenced by upward capillary action in the soil. All samples were collected at 5 :00 a.m. The C1, S and N originate most likely from the atmosphere while K comes from the grass. Tab. 14. Chcmical Composition of Three Strata of Dew Extracted from a Grass Field Composed of Plileum pratense L. (timothy) and Bromus infermus L. (Bromegrass) Figures in parentheses denote standard deviation, valucs in pg/L (kom Cowgill 1988a) Date 1968

Strata

July 15

top middle bottom

July 25

top

middle bottom Aug 15

top middle bottom

K

CI

so4 1200 (500) 985 ( 5 10) 1962 (990) 1500 (700) 1783 (800) 2000 (990) 1700 (810) 1400 (750) 975 (500)

N

Sampling

of Freshwaters

20 1

Recently, amorphous silicates have been developed that can absorb up to 28 times their weight in liquids. In addition, these silicates can become fully saturated in 20 s. In August 1990, the same field was again examined. Diking tubes were used to surround ten stations. Prior to use, 100 g of amorphous silicate were soaked in double distilled water for 48-h. The water was then pressed out at 1.6 metric tons/cm2. This was repeated three times and the water which was concentrated over P,O, was checked for its K content. When none could be detected (< 40 pg/L) by atomic absorption spectrometry, the sample of amorphous silicate was presumed to be clean. This 100 g sample was then divided into ten equal parts and used to absorb the dew in the ten sampling stations. After the dew had been absorbed by the amorphous silicate, the sample was removed, leaving a fine layer in contact with the soil, and sealed in plastic bags containing a closure at one end. The samples of dew were pressed out using the technique previously described, concentrated and analyzed. Unfortunately, in duplicate samples collected from the same station, the coefficient of variation showed about the same range as samples collected with the vacuum pump. Until duplicate samples can be gathered that provide concordant results the chemical composition of dew will remain poorly known. Samples of fog were collected using a glass vacuum pump. Samples gathered simultaneously 200 m apart had a 200% spread on elemental quantity such as Ca, K, S and Mg. Again until duplicate samples can be collected that provide concordant results, the chemical composition of fog will remain largely unknown. The one measurement of fog that appears to be reasonably concordant is the pH of duplicate samples which are quite acid (1.9, 2.1).

8.6 References American Public Health Association (1985) Standard Methodr ,fix the Examination yf Water and Wastewater. Washington, D.C., USA: American Public Health Association, 16th Ed. Asman, W. A. H., Jonker, P. J. (1982) in: Deposition of Atmospheric Pollutants. Georgic, H. W., Pankrath, J. (eds.) Amsterdam: D. Reidel, pp. 231 -239. Barcelona, M. J., Helfrich, J. A,, Garske. E. E. (1985a) Anal. Chem. 57, 460-464. Barcelona, M. J., Helferich, J. A,, Garskc, E. E. (1985b) Anal. Chem. 57, 2752. Berg, E. L. (1982) Handbook,fbr Sampling and Sample Preservation. Washington, D.C., USA: U.S. Government Printing Office. Boettncr, E. A,, Ball, G . L., Hollingsworth, Z., Aquino, R. (1981) Report U.S.E.P.A.-60O/l-S1062, U.S. Environmental Protection Agency, U.S. Government Printing Office, Washington, D.C., USA. Campbell, S., Scott, H. (1985) in: Quality Assurancc .for Enuironriiental Analysis. Taylor, J. K., Stanley, T. W. (eds.). Philadelphia: American Society for Testing and Materials, pp. 272 -283. Cowgill, U. M. (1970) Arch. H.vdrobio1. 68. 1-95, Cowgill, U. M. (1976) Arch. Hydrobiol. 78, 279-309. Cowgill, U . M. (1980) Int. Reu. Gesanften Hydrobiol. 65, 379-409. Cowgill, U. M. (1988a) in: Principles of’Enuironmentu1 Sampling. Keith, L. A. (ed.). Washington, D.C., USA: American Chemical Society, pp. 171 - 189. Cowgill, U. M. (1988b) in: Ground-Water Contamination: Field Methods. Collins, A. G., Johnson. A. I . (eds.). Philadelphia: American Society for Testing and Materials, pp. 172- 184.

202

U. M . Cowgill

Gibb, J . P., Schuller, R . M., Griffin, K. A . (1980) Report U.S.E.P. A.-600/9-HO-010, U.S. Environmental Protection Agency, Cincinnati, OH, USA. Gorham, E., (1955) Geochim. Cosmochim. Acta 7, 23 I - 239. Keith, L. A,, (1981 a) Environ. Sci. Technol. 15, 156- 162. Kcith, L. A. (1981 b) Adimnces in the Identification and Analysis ( I / Organic Pollutants in Woter, Ann Arbor: Ann Arbor Press, Vol. 1. pp. 3-419. Keith, L. A. (1981~)Advances in the Identificution and Analysis of Orgunic Pollutcmts in Wuter, Ann Arbor: Ann Arbor Press, Vol. 2, pp. 481 - 1170. Kratochvil, B. G., Taylor, J. K. (1982) N B S Technical Note 1153. National Bureau of Standards, Washington, D.C., USA. Kralochvil, B. G., Wallace, D., Taylor. .I.K . (1984) Anal. Clwm. 56, 113R- I19R. Lind, 0. T. (1974) Handbook of Common Methods in Limnology. St. Louis: C. V. Mosby. McBean, E. A., Rovers, F. A. (1985) Ground Water Monit. Rev. 5, 61 -64. Miller, G. D. (1982) Proc. Nut. Symp., Aqufer Restor. Ground Water Monit. 2nd, 236. Raynor, G. S., Haycs, J. V. (1983) in: Sampling and Analysis ( I / Rain: Campbell, S. A. (ed.). Philadelphia: American Society for Testing and Materials, pp. 50 -60. Robertson, D. E. (1968) Anal. C/iem. A c f a 40, 1067- 1072.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

9 Guidelines for Sampling Freshwater for Eutrophication Management Programs* H. Klapper, W. Rast, und D. Uhlmann

9.1 Introduction Determination of the trophic state and general water quality represents the core of any assessment or classification of lakes and reservoirs. The objectives of eutrophication management programs should be clearly defined, and primary attention should be given to obtaining the information considered essential to the implementation of these objectives. The reliability of eutrophication control programs can be no better than the reliability of the data bases used in developing the programs. Many eutrophication measurement programs normally focus on nutrient inputs from the land and the atmosphere, and how the nutrients and other related materials partition themselves among the biotic and abiotic components of the aquatic environment. The goal of this chapter is to provide general guidelines for obtaining the necessary in-lake information for development of effective eutrophication management programs.

9.2 What to Sample In addition to the primary causative variables of eutrophication (mainly nutrients), water quality parameters which reflect the impacts of eutrophication also have to be measured and assessed. It is necessary to collect analytical data for several reasons, including: Assessment of the “average” condition of a waterbody during a specific time interval (e.g., seasonal or annual); classification of a lake or reservoir; assessment of the load/response relationship of a lake or reservoir; selection of appropriate control measures for external (drainage basin and atmospheric sources) and internal (sediment regeneration and ground waters) growth-limiting nutrients; prediction of changes in trophic state and/or water quality by means of mathematical models.

* This is a modified reprint of an article on “Guidelines for sampling a waterbody” published in “The control of eutrophication of lakes and reservoirs” (Ryding, S.-0. and Rast, W., eds.), Man and the Biosphere Series, UNESCO, Paris, 1989, p. 147- 168, (with permission).

H . Klcipper, W. Rust, and D.Uhharzn

204

I n selecting appropriate analytical procedures for determining the values of water quality parameters related to eutrophication, several factors should be considered, including: (1) (2) (3) (4)

The required rapidity of the analysis; the required sensitivity and detection limits; the constraints on accuracy and precision; the total number of analyses to be done per year.

A watcr sample, whether measured in the laboratory on in the field, should be indicative of the actual condition of the waterbody at the time and point ofcollection. This type of sample can be considered a representative sample of the waterbody for the component or coniponents of interest. Furthermore, the sample should provide a true description of the temporal and spatial variations in eutrophicationrelated water quality for the duration of the sampling program. This latter characteristic describes a valid sample. Satisfactory sampling should be both valid and representative. Thus, a non-representative sample cannot be valid. Because of this requirement, the following factors must be Considered in designing an adequate eutrophication sampling program:

( I ) Validity uf the sumples: Validity of the sampling sites, - validity of the sampling frequency and timing. ( 2 ) Represenruriveness of' the samples: - The necessary sample size, - a network of single samples at random, versus integrated samples, - sample collection (discrete intervals versus continuous hauls), - sample transportation and storage. ~

A detailed discussion of these factors and how to assess them is provided by Mancy and Allen (1982). Essential criteria for assessing the eutrophication-related water quality in a lake or reservoir are outlined in Tab. 1. In addition, the minimum parameters to be measured at the sampling site itself include: (1) (2) (3) (4)

Temperature and dissolved oxygen profiles; PH; specific conductance; Sccchi depth.

Dissolved oxygen may also be determined in the laboratory, after appropriate preservation in the field. Although many samples can be stored for long periods if properly preserved, the shorter the time interval between the collection of a sample and its analysis, the more reliable the results will be. Microbial activity can change the nitrate/nitrite/ammonia balance, as wcll as the concentration of dissolved reactive phosphorus and dissolved total phosphorus, in a water sample. Ideally, therefore, filtration of the water sample at the sampling site should be done to allow the most accurate determination of these parameters. If this is not possible, one usually is able to obtain reasonably accurate values if

Freshwater f o r Eutrophication Management

205

Tab. 1. Primary Parameters for Assessing the Eutrophication Status of a Waterbody

I.

11.

Parameter

Units"

Morphometric conditions: Lake surface area Lake volume (average condition)' Mean and maximum depth Location of inflows and outflows

km2 lo6 m3 m -

Hq'droc&namic conditions: Volume of total inflow (including ground water) and outflow for different months Theoretical mean rcsidence time of the water (renewal time, retention time) Thermal stratification (vertical profiles along longitudinal axis, including the deepest points) Flowthrough conditions (surface overflow or deep release, and possibility of bypass flow)

111

In-lake nutrient conditions: Dissolved reactivc phosphorus; total dissolvcd phosphorus; and total phosphorus Nitrate nitrogen; nitrite nitrogen; ammonia nitrogen; and total nitrogen Silicate (if diatoms constitute a large proportion of phytoplankton population)

IV

In-lake eutrophication response purumeters: Chlorophyll a ; Pheophytin a Transparency (Secchi depth) Hypolimnetic oxygen depletion rate (during period of thermal stratification) Primary productionc Diurnal variation in dissolved oxygen' Dissolved and suspended solids" Major taxonomic groups and dominant species of phytoplankton, zooplankton and bottom fauna' Extent of attached algal and macrophyte growth in litoral zone'

m3/day

Y'

g 0,/m3 . day g C/m3 day; g C/m2 day mg/L mg/L

The terminology and units proposed by the International Organization of Standardization i s recommended for expressing the parameters. ' A bathymetric map and a hypsographic curve also is necessary in many cases. Can provide additional information on the trophic conditions of a waterbody; recommended if resources are adequate or if special situations require more detailed information.

a

the time between sample collection and analysis does not exceed a couple of hours. Transportation of such samples should be in a cooler, or appropriate cold storage container, and samples should be kept in the dark. This precaution is also valid for biological samples. A collection of acceptable sample preservation techniques and analytical methods for measuring the necessary water quality parameters is provided by Golterman (1971), Wetzel and Likens (1979) and American Public Health Association et al. (1 985).

206

H. Klapper, W . Rast, and D. Uhlmann

As a general rule, analytical results also should include information on the statistical characteristics of the data, including its accuracy, precision, etc. It also is important to attempt to provide an estimate of the potential error associated with both the sampling strategy and the analytical results. Ekedahl et al. (1982) discuss the topic of analytical error associated with laboratory analysis.

9.3 Necessary Temporal and Spatial Resolution for Data If sampling techniques are not carefully selected, analytical results may be partially or completely invalid for their intended uses. In particular, samples must be collected and handled so that, upon measurement, the values of the parameters of interest are the same as those in the waterbody at the time of sampling (see discussion of representative samples above). This topic is discussed further by Mancy and Allen (1982) and Wilson (1982).

9.3.1 Where to Sample Selection of sampling sites and frequency of sample collection depend largely on the morphometry and hydrodynamic properties of a waterbody (OECD, 1982). Therefore, it is important to know the position of all major inflows and outflows, and any related “short-circuiting” of water and associated materials entering the waterbody. The short-circuiting of inflow materials can occur when major inflows and outflows of a waterbody are located relatively close to each other. In this situation, inflow waters and the materials carried in them may exit through a nearby outflow tributary before they have the opportunity to mix with the main volume of the waterbody. The possibility of water short-circuiting can be assessed with appropriate measurements of temperature and/or specific conductance, as discussed in the following sections. All major tributaries must be considered, including measurement and/or calculation of both the concentrations and masses of nutrients and other parameters of interest entering the waterbody. Within the waterbody itself, selection of sampling locations should make allowances for the possibility of a heterogeneous distribution of water quality. Standing waters can be thermally or chemically stratified throughout portions of the year. Thus, particulate materials of differing densities from that of the lake water tend to be distributed heterogeneously. This is especially the case for suspended silt in the water column and for blue-green algae with gas vacuoles. Winds can often blow such algae to one side of a lake or reservoir. In principle, winds also can cause lateral heterogeneity in the distribution of dissolved materials. The vertical or horizontal distribution of such materials can be assessed easily with rapid field tests for temperature, dissolved oxygen, electrical conductance and turbidity.

Freshwater f o r Eutrophication Management

207

Spatial consistency in the siting of sampling stations in the waterbody is important, both for assessment of changes in the values of measured water quality parameters over time and for insuring a reference point against which water quality in other parts of the waterbody can be assessed. Thus, as a practical matter, one normally would not change the location of a sampling station for a given parameter or group of parameters once a sampling program has begun. If one desires to sample another location in a waterbody, it is better to add a new sampling station, rather than change the location of an existing station. The exception would be cessation of a sampling program at a given site, coincident with the initiation of a new sampling program. During the stratification period, the location of the thermocline can be determined by vertical temperature profiles at 0.5 m intervals. The point at which the temperature shows the maximum change (decrease) per depth interval corresponds to the location of the thermocline. A temperature change of 1 "C or more per meter of depth can be used as a rule of thumb in temperate lakes and reservoirs. Lee and Jones (1980) suggest that for temperate zone waterbodies with surface temperatures of 20 "C or more, a temperature decrease of 0.5 "C or more per meter of depth signifies a stratified waterbody. Any significant increase in specific conductance with depth might signify salt-related density stratification. The required number of vertical samples depends on the purpose of the measurement. For non-stratified waterbodies, it usually is adequate to take samples approximately 0.5 m below the surface, 0.5 m above the bottom, and at mid-depth. Ideally, for stratified waterbodies, samples should be taken approximately 0.5 m below the surface, middle of the epilimnion or bottom of the euphotic zone, approximately 1.0 m above and below the thermocline and approximately 0.5 m above the bottom. Shallow waterbodies may not be sufficiently deep to sample all five depths. In such cases, one should consider both the relative proportion of the epilimnion and hypolimnion, and the depth of the euphotic zone in selecting the sampling depths. However, shallow lakes and reservoirs may not exhibit a sustained stratification during the growth season, especially in windy areas. In such cases, an integrated sample (e.g., collected with a hose lowered through the water column, thereby obtaining water from each water layer) may provide adequate information on average in-lake nutrient and chlorophyll levels. If resources are inadequate to sample all five depths, the two thermocline samples might be omitted, although some valuable information may be lost (e.g., see the following discussion on metalimnetic chlorophyll maxima). An exception to these general sampling depth guidelines are waterbodies which exhibit maximum chlorophyll levels in the thermocline or the hypolimnion. Several researchers have reported this phenomenon (e.g., Eberly, 1959; Ichimura et al., 1968; Baker and Brook, 1971; Watson et al., 1975; Fee, 1976; Pick et al., 1984; Moll et al., 1984). The occurrence of non-epilimnetic maximum chlorophyll values has been suggested as a relatively common characteristic of Alaskan lakes (personal communication, Paul Woods, U.S. Geological Survey, 1985). The possibility of non-epilimnetic chlorophyll maxima can be assessed by initial reconnaissance sampling of a waterbody for chlorophyll concentrations, optical

208

H. Klapper, W . Rast, and D.Uhlnicmn

density and/or dissolved oxygen concentrations during the growing season. Examination of water column profiles usually will indicate the presence of sub-surface peaks of such parameters. Small sampling intervals will be necessary, since chlorophyll maxima can be concentrated in very small water layers in some cases. It also is mentioned, however, that it is not clear how knowledge of the occurrence of non-epilimnetic chlorophyll maxima should be used in assessing the overall water quality or trophic status of a lake or reservoir. Pick et al. (1984), for example, suggest that the importance of metalimnetic peaks in regard to the total primary productivity of a waterbody may be greatly exaggerated by the measurement of chlorophyll u. Thus, from the perspective of publicly perceptible deterioration of surface water quality conditions, sub-surface chlorophyll peaks appear to be less important than surface water quality in regard to selection of appropriate measures for the eutrophication control. If a waterbody exhibits a high degree of lateral heterogeneity in water quality, the necessary numbcr of discrete water samples to adequately characterize the waterbody may be prohibitive. One can attempt to compensate for this heterogeneity by mixing portions of individual water samples in quantities proportional to the heterogeneity, thereby obtaining a single composite sample for the waterbody. This approach would not be appropriate, however, for lakes or reservoirs which exhibit distinct longitudinal gradients of eutrophication-related water quality. An example would be a reservoir with an elongated shape. In such a case, depth measurements should be taken at the outflow, at the inflow and at selected intermediate positions. The area of greatest depth should always be included into the sampling program. Particular care should be taken when sampling under ice cover. Samples taken from the hypolimnion are necessary to assess hypolimnetic oxygen depletion and phosphorus regeneration from the bottom sediments (i.e., internal nutrient loading) during the stratification period. The latter parameter often is determined with mass balance calculations of phosphorus inputs and outputs and the in-lake phosphorus concentration. Composite samples containing water taken from both the epilimnion and hypolimnion, proportional to the volume of each layer, usually are required to make such calculations. Alternatively, one can obtain discrete samples from these two water layers and make appropriate adjustments i n the calculations to account for their respective volumes. It should be recognized that selective withdrawal of water from the hypolimnion of reservoirs during the period of thermal stratification must be taken into account in calculating the hypolimnetic oxygen depletion rate during this period. Theoretically, since hypolimnetic withdrawal from reservoirs potentially could remove oxygen-depleted waters at a more rapid rate than in natural lakes, the hypolimnetic oxygen conditions may be better in reservoirs than in natural eutrophic lakcs. However, because of selective discharge of oxygen-depleted waters or waters with high concentrations of nutrients and heavy metals, thc potential for degraded water quality downstream from a rcservoir is increased, compared to natural lakes. With regard to spatial location of sampling stations, Lee and Jones (1980) suggest that, if a watcrbody has no significant arms or sub-basins, and is fairly well-mixed horizontally, a single sampling station at the deepest part of the waterbody is usually adequate to characterize the eutrophication-related water quality. However, if the

Freshwarev,forEutrophcution Manugement

209

chlorophyll or nutrient concentrations vary by more than a factor of f 10 along the length of an elongated waterbody, or if the specific conductance varies by greater than 40-60 pS/cm (25 “C), a single sampling station usually will not provide an adequate description of the “average” water quality conditions of the waterbody. In such cases, it will be necessary to establish additional sampling stations along the length of the waterbody. The number of additional stations should be sufficient to take into account the longitudinal variability in water quality. Determination of the necessary number of sampling stations to account for longitudinal gradients of water quality in reservoirs is discussed further in following sections of this chapter. Another approach to account for longitudinal gradients in water quality is to treat the waterbody as a series of serially-connected sub-basins. This approach essentially partitions the waterbody into a series of connected “sub-basins”, each with its own unique water quality characteristics. Each sub-basin can be examined individually for its trophic status and water quality characteristics. Kerekes (1982) previously has demonstrated the utility of this approach in assessing phosphorus load-trophic response relationships in reservoirs. This concept also is discussed by Walker (1985) and Frisk (1981).

9.3.2 When to Sample Eutrophication-related water quality parameters measured in surface water samples, or samples taken in the upper portion of the epilimnion, can be highly variable over time in both natural lakes and man-made impoundments. In contrast, the amplitude of variations in hypolimnetic water quality at diurnal (or even weekly) intervals is much lower. Thus, samples from the lower layers of a stratified waterbody reflect water quality characteristics of a longer period of time, compared to epilimnetic samples. However, because of the possibility of hypolimnetic oxygen depletion, and related nutrient and heavy metal releases from the bottom sediments, the measured water quality in the hypolimnion can be worse than in surface water samples. When diurnal variations in water quality do occur, and are relevant to the sampling program, sampling times have to be chosen carefully to reflect this variation. For example, in waters with high chlorophyll levels, the dissolved oxygen concentration in the surface waters is normally minimum at sunrise and maximum at noon. Thus, if samples are always collected at the same time of day and do not reflect this diurnal variation, one may obtain biased results for the mean value of the dissolved oxygen. To determine diurnal variations in dissolved oxygen, initial investigations over a 24-hour period are needed. Minimum requirements include measurements at pre-dawn, midday and pre-sunset. To obtain adequate data for accurate water quality assessment and trophic classification of a waterbody, the following minimum sampling frequency for the essential criteria (see Tab. 1) is recommended:

210

H. Klrcpper, W. Rust, and D.Uhlmann

(1) Samples should be collected monthly from November to March, and approximately biweekly from April to October (encompassing the period of thermal stratification) in northern temperate climates. The same regime would apply over the corresponding growth and non-growth months in southern temperate climates. (2) In tropical/sub-tropical regions, samples should be collected biweekly from the start of the rainy season until three months after it is over, as well as during the period of thermal stratification. Samples should be collected monthly at other times of the year. (3) In both cases above, sampling also should be done during any overturn periods. In addition, if algal blooms occur between the above-noted sampling intervals, samples should be taken during the bloom periods. As a practical matter, if one has to choose between temporal and spatial resolution, temporal resolution may be the logical choice, since one station often is sufficient to dcscribe the average conditions in a lake or reservoir. The exception would be reservoirs with longitudinal gradients in water quality (discussed further in the following section). In temperate lakes with a water residence time of many years, average lake concentrations are close to spring overturn concentrations. In contrast, in irregularlyflushed lakes and in reservoirs, the spring concentration may be substantially different from the annual average concentrations. In these latter cases, full year measurement cycles usually are required to obtain reliable information (Janus and Vollenweider, 1981). For waterbodies which exhibit substantial year-to-year fluctuations in hydrological conditions, calculations based upon a single year cycle may be inadequate. For example, in warm climates with dry and wet seasons, the variability in both volume-flow and nutrient concentrations is often very high. Thus, as a general rule, a minimum sampling program covering three consecutive years is recommended. The magnitude of the internal phosphorus load (as phosphate release into the euphotic layer) can be approximated with a relatively high frequency of vertical sampling (e.g.,biweekly) in the epilimnetic and hypolimnetic waters. Special attention should be given to the period of thermal stratification (Janus and Vollenweider, 1981). Ryding and Forsberg (1977) and Vollenweider (1976) also provide guidelines for calculating the internal phosphorus load to a waterbody.

9.3.3 Sampling Strategies in Waterbodies with Longitudinal Water Quality Gradients As noted previously, longitudinal gradients of water quality in a waterbody usually prohibit the use of a single sampling station to characterize the “average” water quality in the waterbody, especially reservoirs. Consequently, a relevant concern with such waterbodies is to determine how many sampling stations are needed to characterize the water quality adequately and where they should be located in the waterbody.

Freshwater f o r Eutrophication Management

21 1

Based on a study of DeGray Lake, a reservoir in southern Arkansas (United States), Thornton et al. (1982) provide an approach for addressing this concern. Previous monitoring data indicated this reservoir had considerable longitudinal and vertical variation in water quality. The data showed the water quality gradient was most pronounced at the upper end of the reservoir (nearest the tributary inflow), and progressively decreased as one moved closer and closer to the downstream (dam) end of the reservoir. To address this problem, Thornton et al. (1982) partitioned the reservoir into fifteen transects, averaging five stations/transect. The reservoir was sampled at 0, 2,4,6, and 10 m, and at 5 m intervals thereafter to the bottom, for total phosphorus, turbidity and chlorophyll a in July, 1978, and in January and October, 1979.Analysis, of variance and Duncan’s multiple range test (Cochran, 1963; Snedecor and Cochran, 1967) were used to detect significant differences between the means of the parameter values between the various transects. General linear statistical models were used to characterize the transect means, assuming a normal distribution of the data, homogeneity of variance, etc., usually required for statistical models. If the model showed that the slope of the mean value of a parameter in a transect was not significantly different from zero, then a minimum of one station was needed to characterize the transect area. If a linear function was needed to account for a significant portion of the variance among the transect means, a minimum of two stations was needed to characterize the transect area. If a quadratic function was necessary, a minimum of three stations were needed, etc. Once the analysis of variance was completed, the transects were compared in order to identify areas of similarity and overlap. In the 15-transect DeGray Lake example, Thornton et al. (1982) showed that transects 1 - 5 had similar means for all variables over all dates. Thus, one station was sufficient to characterize the entire area represented by all five transects. The means of transects 6- 13 required a linear model to account for the variance among the transect means. Thus, a minimum of two stations was necessary to characterize the water quality in the area represented by these transects. Transects 14 and 15 were either distinct or required a linear model. Thus, these two transects each needed a separate sampling station. Based on this analysis. DeGray Lake needed a minimum of five sampling stations to characterize its longitudinal water quality gradients. Thus, sampling stations on transects 3 , 10, 12, 14 and 15 (Fig. 1) would be adequate to characterize the water quality of DeGray Lake. In addition to the number of sampling stations, the number of samples to be collected at each station also depends on the date variability of the desired water quality parameter, as well as the desired precision of the parameter estimate. Based on these factors, Thornton et al. (1982) also provided a general formula for random sampling in DeGray Lake, as follows: n = t2s2/d2 where: n = number of samples; t = appropriate value from student’s distribution; s2 = sample variance (based on existing data or preliminary surveys); d = desired precision about the mean value.

(1)

212

H . Klupprr, W . Rust, und D.Uhlrruirrn

1 5 1 4 1 3 1 2 1 1 1 0

9 8 7 TRANSECT

6

5

4

3

2

1

Fig, 1. Reprcsentative sampling station locations for characterizing longitudinal watcr quality gradients in DcGray Lake (from Thornton el al., 1982)

They suggested an initial t value of 30 degrees of freedom to begin the analysis procedure. When applied in an iterative manner, Equation (1) will produce a value for n converging on the appropriate sample size, usually after 3 - 5 iterations. For the period when the reservoir is stratified, a different random sampling design can be used. A general formulation for stratified sampling is:

n

=

(CWiSi)2/(d’/t’)

where: n = total number of samples; wi = weighting factor for stratum ( e g , ratio of volume of epilimnion to total lake volume); si = standard deviation of samples in stratum; t = appropriate value from student’s t distribution. Within each individual stratum (e.g., epilimnion or hypolimnion), the number of necessary samples can be calculated as: n,/n

=

wisi/C (wisi)

(3)

where: ni = number of samples in the stratum i ; n = total number of samples. One point to emphasize here is that, while simple random sampling requires only a single mean and variance (see Equation (I)), stratified sampling requires an estimatc of the mean and variance of each stratum sampled.

Freshwater j b r Eutrophication Managenzent

2 13

9.4 Calculating the Costs of Sample Collection Thornton et al. (1982) provide a methodology for calculating the costs of obtaining necessary samples, based on the type of analyses presented above. Cale and McKown (1986) also present an approach for calculating the costs of collecting the necessary number of samples in a waterbody to achieve a desired data precision. This approach is essentially a comparison of the maximum expected variance of the data with the desired data reliability. Although both approaches overlap to a certain degree, that developed by Cale and McKown is discussed here because of its greater emphasis on calculation of sampling costs. Cale and McKown have approached the problem of desired data precision versus costs so that one can attempt to answer the following types of questions about a water quality sampling program: (1) Will the funds available for a sampling or monitoring program be sufficient to achieve data of a desired precision? (2) If not, what is the precision of the data that can be obtained for the available funds? (3) If the initial funding level is insufficient to obtain the desired data, how much additional funding will be necessary to achieve this precision? Data quality is expressed in terms of a desired statistical confidence level. In scientific/technical work, a 95% confidence level (i.e., p = 0.05) often is used as the standard of data acceptability. However, as Cale and McKown point out, this value is actually an arbitrary criterion which has gained general acceptance simply because of its repeated usage over time. It is both a reflection of the fact that improvements in data quality are less expensive and less difficult at lower significance levels than at higher levels and a standard of data reliability. In fact, in a given situation, data with a confidence level of 90% may be adequate for answering the question being asked. That is, one might conclude that the funds necessary to increase the confidence level of the data from 90 to 95% may not be worth the additional accuracy gained. In order cases, a particularly sensitive issue may require a data confidence level even greater than 95%. Unfortunately, there are no upriori guidelines for deciding unequivocally whether or not incremental improvements in data reliability are worth the additional costs. In most cases, it is as much a socioeconomic and political decision as a technical decision. As noted above, the approach of Cale and McKown for calculating sampling costs is basically a comparison of the maximum expected variance of the data being sought with the desired data reliability. The results of this comparison are then related to the associated costs. This approach consists of the following steps : (1) Based on the desired sampling frequency and the associated costs, determine the number of samples per trip (n)that can be collected with the available funds. (2) Based on literature values or pilot study results, calculate the largest standard error and standard deviation likely to occur for the desired number of samples (HI.

H . Klupper, W . Rust, and D. Uhbnann

214

( 3 ) Based on the results of step (2), calculate the maximum variance likely to occur under the specific conditions of the sampling program. (4) Calculate the half-width of the confidence interval (e.g., 95%) that will result from the variance calculated in step (3). (5) Compare the results of step (4) to the desired data reliability. Based on this comparison and the data quality goals of the sampling/monitoring program, one can decide whether or not it is reasonable to spend the necessary funds to obtain the desired data. The costs of any monitoring activity can be characterized as either “fixed” or “variable”. Fixed costs can be further identified either as being relatively constant and independent of the monitoring program ( F l ) , or as a specific function of the program ( F z ) . Examples of F , costs include literature reviews, data analysis and development and refinement of monitoring programs. F , costs are most directly related to the specific monitoring program and include such items as costs of travel time to the study site, equipment setup and calibration, per diem costs, etc. By contrast, variable costs (0) are a direct function of the intensity of the monitoring effort and include the consumption of chemicals and supplies, collection of field samples and laboratory analyses of the samples. These latter costs are variable in that they are a function of the number of sampling trips, number of samples collected per trip, etc. Given these definitions, the costs (C) of collecting a required number of samples (n)for a given numbers of sampling trips ( M )over the duration of a lake or reservoir monitoring program can be determined as follows:

C

=

F,

+ M ( F z + no)

(4)

where: C = total costs (e.g., monitoring budget); F , = fixed costs independent of monitoring effort (see text); F, = fixed costs dependent on monitoring effort (see text); u = variable costs (see text); M = number of sampling trips; n = number of samples collected per sampling trip within budget limitation. The values of M and n are calculated with the use of standard statistical procedures. The value of n can be calculated as: n

= (s2t:-,,z)/d’

where: n = required number of samples to achieve desired precision of data; s2 = estimate of data variance, based on the preliminary sample; t = Student’s t-test distribution for 1 - a confidence level and no - I degrees of freedom ; no = size of a preliminary sample; d = half-width of the desired data confidence interval.

215

Freshwater for Eutrophication Management

The value of M can be calculated as:

M = (s?z:-a,2))/4

(6)

where: s,2 = variance estimate of population change over time (e.g., chlorophyll concentrati on); Z = appropriate value from a table of cumulative normal distribution, with confidence level of 1 - a ; d, = half-width of the confidence interval for measuring a change in the population. Equations (5) and (6) provide estimates of the values of n and M , based on the desired level of data confidence for the monitoring program. A hypothetical application of this approach is provided with the data in Tab. 2. In this example, it is assumed that a lake manager wishes to initiate a lake restoration program based on reducing the external phosphorus load to LakeX and that he has a total monitoring budget of $30,000. The objective of the control program is to reduce the in-lake chlorophyll concentration from 25 pg/L to 10 pg/L. The manager wishes to sample Lake X biweekly during the growing season (MaySeptember) and monthly during the rest of the year (October- April) for a period of three years. In addition, the manager desires that changes in the chlorophyll concentrations in Lake X as low as 2 pg/L between sampling trips be detected, and that the data obtained have a confidence level of 95 per cent (i.e., p = 0.05). It is also assumed that six days of professional and secretarial assistance are needed per year. Given this background, the basic question to be answered is whether or not the $30,000 monitoring budget is adequate to achieve these monitoring objectives. Using the approach of Cale and McKown (1986), this problem is assessed as follows: (1) Based on the desired sampling schedule, 17 sampling trips can be made each year for the three year monitoring program. Thus, M = (17) (3) = 51. The number of days of necessary professional assistance is six days each year for the three year monitoring program. Based on the cost estimates in Tab. 2, the number of samples Tab.2. Hypothetical Costs of Monitoring Program for LdkeX (modified from Cale and McKown, 1986) Activity

Type

Estimated cost "/day

Samples/day

V

$250 $150 $200 $300 $200

25 50

Field sample collection Laboratory analyses Travel costs Literature research; data analysis; report writing, etc. Secretarial/graphics

U F2

F, F, ~

a

~~

Such items as overhead expenses can be incorporated into this figure or listed as a separate item.

H . Klapper, W . Rust, und D.Uhlmrmn

216

( n ) that can be collected per sampling trip, within the $30,000 monitoring budget (C), is calculated with Equation (4) as follows: $30,000

= =

(18)(300 + 200) + 51[(200) 9,000 + 10,200 + 663n.

+ n(250/25 + 150/50]) (7)

Solving Equation (7) for n, the budget will allow 16.29 samples to be taken during each sampling trip. For this example, the value of n is rounded off to 16 samples per sampling trip. ( 2 ) Based on a review of relevant literature (or pilot study results, if available), the manager determines that the maximum standard error (s.e.) usually associated with measuring the mean chlorophyll concentration in similar waterbodies is 5 &L, for a sample sizc of ten samples. The standard error (M.) = standard deviation (s)/fi. Thus, the maximum standard deviation expected for this monitoring program is calculated as s = (s.e.) = (5)((/10) = 15,81. Since the monitoring budget of $30,000 allows approximately 16 samples/sampling trip (see step 1 above), the maximum expected standard error (s.e.) to be expected should be no larger than (s)/(m) = (15.81)/(4) = 3.95. (3) The manager is concerned with detecting differences in mean chlorophyll concentrations between sampling trips (as contrasted with differences within a single sampling trip). Assuming the data arc independently and normally distributed, the associated variance ( V )of the difference in mean chlorophyll concentrations between any two sampling trips is calculated as:

(fi)

V(2, -

22) =

:s

=

V(2,)

+

V(X2)

=

(s:/nl)

+ (s3n2)

(8)

where X I and X2 are the mean values of the 16 samples collected during any t w o different sampling trips. One can maximize the variance by maximizing each of the standard error terms (i.e. s.e. = s/l:/n) in the summation term on the right side of Equation (8). Since the maximum expected standard error is 3.95 (via step 2), each of the two summation terms will have a maximum value of 3.95. The maximum variance ( V )then can be calculated as: VIXl

~

XZ] =

=

s: = V(X,)

[(.s:/n,)

+

V(X2)

+ ( s z / n 2 ) ]= (3.95)2 + (3.95)2 = 31.2

(9)

It is noted that one could calculate the average variance of thc data in the same manner. However, this would not insure the desired data precision will be achieved. Thus, as a practical matter, this example uses the maximum variance as the criterion for insuring the monitoring data to be obtained are within the desired precision range. (4) Since the manager is concerned with differences in the mean chlorophyll values between sampling trips, the expccted half-width of the confidence level (d) for the maximum variance calculated in step (3) is determined with the use of Equation (6). Knowing that M = 51, :s = 31.2 and Z1-a,2(i.e., the half-width of the 95% confidence level) = Z1-o.05,2= Zo.975 = 1.96, rearrangement of Equation (6) allows one to calculate the value of d as follows: d:

= (s: -

Zqp,,2)/M = [(31.2

-

(1.96)2/51] = 0.536

(10)

Freshwater f o r Eutrophirwtion Management

217

Thus: d, = v0.536 = 50.73 pg/L. Since the calculated half-width of the desired 95% confidence level for the monitoring data is within the required precision of 2 pg/L, the monitoring program, as designed, would allow the manager to obtain data of the desired reliability. In fact, the desired precision range is more than twice the value of the expected precision. Thus, in this case, one can even consider reducing the intensity of the monitoring program and still be assured of achieving the desired level of data realiability. In contrast to the above results, if one had a total monitoring budget of only $20,000 for the same sampling program, the calculated value of d would have between 2.85. This value is nearly 50% greater than the desired precision of t2 pg/L. Based on this result, one would conclude that the desired data precision could not be achieved within the $20,000 monitoring budget. Thus, one would have to decide between either eliminating the monitoring program, accepting the calculated data precision as the best achievable under the financial constraints of the monitoring program, or increase the monitoring budget to the level necessary to obtain data of the desired quality. The necessary funds needed in a specific case to obtain data of a desired precision also can be calculated. The procedure uses the same steps as the above example, but in a different order. Basically, one uses the estimate of the maximum variance to “back-calculate” the specific value of II that would allow the variance not to be exceeded. The total costs are then calculated with Equation (4). Based on the date in Tab. 2, Equation (6) can be rearranged to calculate for the maximum variance (sf) as follows: s: = [(M)(df)/Z -*,21 =

[(51)(2)2/(1.96)’] = 204/3.84

=

53.10

(11) Thus, since s is an estimate of the maximum variance of the change in the mean chlorophyll concentrations in Lake X, it cannot exceed a value of 53.10 if one wishes to insure the desired data precision of f2 pg/L in this example. Equation (9) shows that the total variance is equal to the sum of the individual standard error terms (i.e., s.e. = s Thus, to insure that a total variance of 53.10 is not exceeded, one must insure that the summation of the standard error terms in Equation (9) does not exceed this value. One way to accomplish this is to assume that the two standard eror terms are equal and that their sum is 553.10. Under this assumption, Equation (9) can be written, as follows:

fi).

V

= s:

n

=

= 53.10 = 2(s:/nl)

(12) Since n , = n 2 , one can solve Equation (12) for the value of the maximum standard = 5.15. error (i.e., s.e. = Once the maximum value of an individual standard error term is calculated, the standard error expression in step (2) can be rearranged to solve for the value of n. If one assumes the calculated standard deviation (s) in step (2) is the maximum value to be encountered, (s/s.e.)2 = (15.81/5.15)2 = 9.42

(13)

218

H. Klapper, W . Rast, and D.Uhlmann

Now that the value of n is known, the monitoring budget required to obtain data of the desired precision can be calculated with Equation (4), as follows:

C

+

+

F, M [ F , (no)] = (18)(300 200) 51[(200 9.42(250/25 = 9,000 10,200 6,245 = $25,445 =

+

+

+ +

+

+ 150/50)]

(14)

This calculated value is approximately half-way between the $30,000 budget (which will exceed the desired data precision) and the $20,000 budget (which will not achieve the desired precision), consistent with the monitoring realities of this example. it is emphasized that one can also analyze a multi-parameter monitoring program using this same type of analysis applied in an iterative manner. Initially, one would establish a priority listing of the monitoring parameters of interest, based on the most critical data needs. The first parameter in this list then would be subjected to the analysis illustrated above. Equations (11)- (14) would be used to determine the monetary requirements for obtaining data of the desired precision for the first parameter. The calculated costs for obtaining the first parameter in the priority list would be substracted from the total monitoring budget. The same analysis then would be applied to the second parameter in the priority list, using the revised (reduced) monitoring budget as the appropriate value of C for Equation (4). This procedure can be continued for each succeeding parameter in the priority list until the available budget was expended. If this occurred before all the parameters in the priority list had been examined, one can either reduce the number of parameters to be monitored, reduce the desired data precision, increase the monitoring budget, or some combination of these options. The reader is referred to the report of Cale and McKown (1986) for further details (also see Thornton et al., 1982). Both sources provide detailed examples of the necessary calculations for making the types of assessments outlined above.

9.5 Compilation and Presentation of Data Realistically, one cannot sample a waterbody continuously. Consequently, one should attempt to minimize any bias in the data caused by the sampling frequency. Lee and Jones (1980), for example, suggest that all data should be converted to describe weekly mean values. For the biweekly sampling interval suggested in the previous section, this can be done by calculating the mean value of two consecutive biweekly values. This calculated mean value can be considered representative of the conditions (for the water quality parameter being measured) for the week between the two sampling dates. The same general procedure can be used for the monthly sampling interval, except that the arithmetic mean value obtained from the two consecutive monthly samples would be used for each of the three weeks between the monthly sampling dates. Lee and Jones (1980) suggest that, while not foolproof, this procedure appears to be adequate for determining reliable mean in-lake nutrient and chlorophyll values for the simple, empirical OECD load-response models.

Freshwater ,for Eutrophication Management

219

For in-lake chlorophyll and Secchi depth measurements, it is suggested that both annual and summer mean values be calculated, since the latter period usually represents the period of maximum eutrophication-related water quality degradation (Lee and Jones, 1980). Volume-weighted mean values (i.e., mean values adjusted to reflect the volume of the water column from which they were collected; see further discussion below) would appear to provide the most accurate estimate of the “average” concentration of an in-lake parameter for the waterbody as a whole. However, for the period of thermal stratification, the arithmetic mean in-lake chlorophyll and nutrient concentrations in the surface waters (i.e., approximately the top two meters) usually provide more relevant information on the publicly perceptible symptoms of eutrophication than does a composite sample composed of water from the epilimnion and hypolimnion. As a practical matter, it is obvious that the use of multiple sampling stations for waterbodies with significant longitudinal gradients of water quality will result in the accumulation of a larger set of data points than a waterbody with only one station. Thus, a primary concern is how to calculate one “average” value for a water quality parameter which applies to the whole waterbody, even if the parameter is characterized by markedly different values along the length of the waterbody. There are two possibilities in such cases. To begin with, if it is necessary to partition a waterbody into separate “sub-basins” (including distinct embayments separate from the main body of the lake or reservoir), one can treat the waterbody as a series of connected sub-basins, rather than as one single waterbody (see discussion in previous section). However, this approach may make it difficult to evaluate the effectiveness of potential eutrophication control measures for the waterbody as a whole. An alternative approach is to apply a weighting factor to the data from each of the sub-basins. In this way, an “average” value which takes into account the relative proportion of the sub-basins from which the data are obtained can be calculated. One can calculate the relative volumes of each sub-basin and apply this factor to the mean value of the parameter obtained. The “average” value calculated in this manner will represent an integration of the relative proportions of each of the sub-basins, even though it represents the arithmetic summation of a series of data points, rather than a single measure. One also can attempt to weight the mean values according to the surface areas of the sub-basins, especially if the depth is not significantly different along the length of the waterbody. However, if the depth does change markedly from one end of the waterbody to the other, the use of surface areas as a weighting factor can produce erroneous results. In such cases, the use of volume-weighting can be attempted. A precautionary note is that weighting techniques can produce a single value describing the “average” nutrient concentration (or other water quality parameter) in a waterbody. However, it also must be recognized that this calculated number is an entirely artificial value, since it was derived from mathematical calculations (as contrasted with a measured value). Because it is a derived number (e.g., the total mass of nutrients in the waterbody divided by the total volume of water), one may never actually measure this “average” value in the waterbody. Nevertheless, it is based on arithmetic manipulations of measured data and can be used to describe the relative conditions in a reservoir characterized by longitudinal gradients in water quality.

220

H . Klapper, W . Rast, and D. Uliimann

Another practical observation is that the arithmetic mean may not always be the best value to describe the “average” in-lake condition of a given parameter (e.g., average in-lake chlorophyll or nutrient concentration). This is because the dynamic nature of a waterbody’s metabolism is often characterized by a series of extreme conditions (e.g., high and low values). An algal bloom in a lake or reservoir, for example, can result in high chlorophyll levels for short periods of time, interspersed with low values over longer periods of time. In such cases, the “average” chlorophyll concentration calculated as the arithmetic mean (i.e.% data values/number of observations) may not accurately depict the skewed nature of the data. Heynian et al. (1984), for example, showed that several in-lake eutrophication response parameters (e.g., chlorophyll and phosphorus) showed a non-symmetrical frequency distribution in 25 Swedish lakes. That is, the data did not exhibit the standard “bell-shaped curve” of a normal distribution. This may be the general situation for most lakes and reservoirs. Thus, strictly speaking, the median value should provide a more accurate description of the average situation or conditions in the waterbody. The median value is the value which has an equal number of data points greater and lesser than itself. It can be calculated by plotting all the data for a specific water quality parameter on a probability scale, and selecting the 50% value. Alternatively, one can rank the data in an ascending or descending order and select the middle value. As a practical observation, however, the difference in the “average” condition described by thc arithmetic mean versus the median value often will be insignificant when related to eutrophication assessment and control programs. One can assess whether or not the “average” condition based on the arithmetic mean is significantly different from that based on the median by directly comparing the two values. If one decides (statistically or subjectively) that the numbers differ by too great a margin, it is recommended that the median value be used to describe the average condition. Another method of describing the average value of a given parameter is to use the geometric mean value. This value is calculated in the same manner as the arithmetic mean, except that the data are first converted to common logarithms. The arithmetic mean of the logarithms is calculated. The antilog of the arithmetic mean constitutes the geometric mean value. The geometric mean was used in the OECD ( 1 982) international eutrophication study to calculate theoretical trophic boundary values for several common water quality parameters. Again, as a practical observation, some of the individual investigators involved in the OECD (1982) study indicated that the arithmetic mean value was usually adequate for use in the simple, empirical load-response models developed in the study. Based on these suggestions regarding data compilation, one can readily apply the results obtained to various modes of data analysis (e.g., correlations an regressions).

9.6 References American Public I Iealth Association, American Water Works Association and Water Pollution Control Federation ( 1985) Standard Methods ,fbr the E.xamiriation of’ Water and Wustcwuter, 16th Edition. American Public Health Association, Washington, D.C. Baker, A. L., Brook, A. J. (1Y71) Optical density profiles as an aid to the study of microstratified phytoplankton populations in lakes. Arch. Hydrohiol. 69, 214-233.

Freshwater ,for Eutrophication Munugcwent

22 I

Cale, W. G., McKown. M. P. (1986) A cost analysis technique for research management and design. Enoiron. Management 10, 89 -96. Cochran, W. G. (1963)Sanipling Techniques, 2nd Edition. Wiley&Sons, Inc., New York. 413 p. Eberly, W. R. (1959) The mctalimnctic oxygen maximum in Myers Lake. InocJst. Indianu Lakes Streams 5, 1 - 46. Ekedahl, G., Rondell, B., Wilson, A. L. (1982) Analytical errors. In: Suess, M. S. (ed.), Exurnination q/ Water,[or Pollution Control. A Reference Handbook, Volume I. Pergamon Press, Oxford, United Kingdom, p. 266-315. Fee. E. J. (1976) The vertical and seasonal distribution of chloroohvll in lakes of the Exoerimental Lakes Area, northwestern Ontario: Implications for primary productivity. Limnol. Oceunogr. 21, 767 - 783. Frisk, T. (1981) New modifications of phosphorus models. Aqua Fennica 11, 7- 17. Golterman. H. L. (ed.) (1971) Merhods ,for Chemical Aizulysis of' Fresh Waters. International Biological Programme (IBP). Handbook Series, No. 8, Blackwell Scientific Publications, Oxford, United Kingdom. 188 p. Heyman, U., Ryding, S.-0.. Forsberg, C. (1984) Frequency distributions of water quality variables. Relationships between mean and maximum values. Water Res. 18, 787- 794. Ichimura, S., Nagasawa, S., Tanaka. T. (1968) On the oxygen and chlorophyll maximum found in the metalimnion of a mesotrophic lake. Bot. Mag. Tokyo 81, 1 - 10. Janus, L. L., Vollenwcider, R. A. (1981) The OECD Cooperutive Programme on Eutrophicution. Canadinn contribution. Scientific Series No. 131, Canada Centre for Inland Waters, Burlington, Ontario, Canada. 371 p. Kerekes, J. J. (1982) The application of phosphorus load - trophic response relationships to reservoirs. Cdn. Water Resour. Jour. 7, 349 354. Lee, G. F., Jones, R. A. (1980) Study progrumme,for development of'information.for U S P of OECD eutrophication modeling in boater yuulity management. Manuscript, Quality Control in Reservoir Committee, American Water Works Association. 35 p. Mancy, K. H., Allen, H. E. (1982) Design of measurement systems. In: Suess, M. J. (ed.). E.xamination of Water,for Pollution Control. A Rqfirence Hundbhook, Volume 1 , Pergamon Press, Oxford, United Kingdom, p. 1-22. Moll, R. A,, Brahce, M. Z., Peterson, T. P. (1984) Phytoplankton dynamics within the subsurface chlorophyll maximum of Lake Michigan. Jour. Plank. Res. 6, 751 -766. OECD (Organization for Economic Coopertion and Development) (1 982) Eutrophication of' Wuters. Monitoring, Assessment and Control. Final Report. OECD Cooperative Programme on Monitoring of Inland Waters (Eutrophication Control), Environment Directorate, OECD, Paris, 154 p. Pick, F. R., Nalewajko, C. & Lean, D. R. S. (1984) The origin of a metalimnetic chrysophyte peak. Limnol. Oceanogr. 29, I25 - 134. Ryding, S.-0. & Forsberg. C. (1977) Sediments as a nutrient source in shallow, polluted lakes. In: Golterman. H. L. (ed.), Interactions Between Sediments and Fresh Waters, Junk Publishcrs, The Hague, The Netherlands, p. 227-234. Snedecor, G. W., Cochran, W. G. (1967) Sfutisticul Methods. The Iowa State University Press, Ames, Iowa, USA. 593 p. Thornton, K. W., Kennedy, R. H., Magoun, A. D. & Saul, G. E. (1982) Reservoir water quality sampling design. Water Resour. Bull. 18, 471 -480. Vollcnweider, R. A. (1976) Rotsee, a source, not a sink for phosphorus? A comment to and a plea for nutrient balance studies. Hydrologie 38, 29-34. Walker, W. W. (1985) Empirical mefhods,forpredicting eutrophication in impoundments. Report 3. Model rqfinements. Technical Report E-8 1-9, Environmental & Water Quality refinements. Technical Report E-8 1-9, Environmental & Water Quality Operational Studies, US Army Corps of Engineers, Waterways Experiment Station, Vicksburg, Mississippi, USA. 297 p. Watson, N. H. F., Thomson, K. P. B. & Elder, F. C. (1975) Sub-thermocline biomass conccntration detected by transniissometer in Lake Superior. Verh. Internat. Verein. Limnol. 19, 682 -688. Wetzel, R. G . & Likcns, G . E. (1979) Limnological Anal.y.si.s. W. B. Saunders Co., Philadelphia, Pennsylvania, USA. 357 p. Wilson, A. L. (1982) Design ofsampling programmes. In: Suess, M. J. (ed.), E.xaminution of Water ,for Pollution Control. A Reference Hundbook, Volume I , Pergamon Press, Oxford, United Kingdom, p. 23 - 77. . . I

~

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

10 The Sampling Strategy in the River Elbe Experiences Helmut Guhr and Erich Weber

10.1 Characteristics of the River Elbe Rising in the Sudeten in mountains at an altitude of 1390 m above sea level and draining a catchment area of 51 394 km’, the river Elbe crosses the border between the Czech Republic and the Federal Republic of Germany after a traveling distance of 319.12 km. The length of the river in the former GDR is 566.28 km with a catchment area of 79658 km2 (Fig. 1). This chapter focuses on this section of the river which has a mean gradient of0.299%. The hydrographic system is characterized by few tributaries forming the major part of the catchment area. The tributaries Saale and Have1 cover as much as 60% of the catchment area and affect the water quality of the river Elbe over extended mixing stretches. The hydrological regime of the river Elbe is characterized by high and low water periods throughout the year and is therefore a rain snow-type river (Tab. 1).

10.1.1 Types of Use Due to extreme pollution loads, usage of the river Elbe on the territory of the former GDR was severely limited up to the year 1990. The river Elbe was primarily used for sewage discharge which knowingly exceeded or overtaxed the river’s selfpurification capacity. Despite the knowledge that the water was completely unsuited for drinking water preparation, some service areas had to use this water resource. Elbe water was used for irrigation, both directly and bank-filtered. In addition, industry used the water of the river Elbe for cooling purposes and, in limited amounts, as process water. The river was obviously not used for fishing, but for recreational purposes such as canoeing and other water sports. The river also served as a major freight transport route, mainly used by freight ships of the Czechoslovakian state-owned shipping company. In some areas the banks of the river have been declared nature preserves. The alluvial forests located in the foreshore are in good condition.

10.1.2 Pollution Loads It should be noted initially that measured values of the effluents were not available from any larger sewage treatment plant. The data on the pollution load - in the case of sewage discharge also called the sewage load - shall be considered as

224

H . Gulv mil B. W r h v tBaltic

Sea

The Catchment Area of the river Elbe A, = 148 268 km2

Borderline of catchment area Major sampling point

Austria Fig. 1. Catchment area of the river Elbe and the major sampling points.

estimated values. They refer to the pollution load (waste load) in the untreated waste water of one citizen and are given as population equivalents (PE). According to Imhoff and lmhoff (1985), I PE is equivalent to a content of organic substancc of 60 g/day, expressed as biological oxygen demand (BOD,).

Sampling in the River E l h ~

225

Tab. 1. Selected Hydrological Characteristics of the Rivers with a Catchment Area Larger than 2000 km2 within the Catchment Area of the River Elbe Item Watercourse

Profile

Catchment area

Elbe km

(km2)

Average Average height depth of of preci- runoff pitation (mm) (mm)

Mean Mean Ann. series specific flow rate runoff

*

km2)

(m3/s)

1 Elbe

668

192

6

313

2

667

191

5.1

327

3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

19 20 21 22

51 394 0.0 CR state border 3.43 D D/CR (downstream of source at Kamenice) Level at 53096 55.6 Elbe Dresden 55211 154.6 Level at Elbe Torgau 5541 198.5 Schwarze Mouth Elster Schwarze Level at 4327 Loben Elster 61879 214.1 Level at Elbe Wittenberg Mouth Mulde 7400 259.6 5995 Level at Mulde Bad Duben 69849 274.8 Level at Elbe Aken 24079 290.7 Mouth Saale 23687 Level at Saale Calbe/ Grizehne Level at 94060 296.5 Elbe Barby 94942 326.6 Level at Elbe Magdeburg 97780 388.2 Level at Elbe Tangcrmunde Mouth 24025 438 Havel 23978 Level at Havel Havelberg Level at 123532 454.8 Elbe Wittenberge Mouth 2990 504.1 Elde Level at 2920 Elde MalliB Level at 131950 536.4 Elbe Neu-Darchau Mouth 2253 559.5 Sude 735 Level at Sude Garlitz

665

188

6.2

344

h, Q

-

ha,

-

Q'

-

-

1951185 1931190 1971185 1936190

27.6 583

147

5

657

189

6

780

330

10.8

663

207

6.4

510

164

4.8

682

195

5.9

697

1981/85 Q - 1974/90 (372.0) h, - 1971/85 Q - 1951190 73 64.8 h, - 1971185 Q - 1961/90 446 h, - 1971185 Q - 1936190 115 114 h, - 1971185 Q - 1932190 21.6

hN -

- 1956185 - 1900/90 - 1931191 - 1931/90 h, - 1966185 Q' - 1920190

559

hN

Q h, Q

I87

6

566

195

5.8

567

144

4.8 4.7

I15 114

1981191 1981/90 h N - 1955185 Q - 1900190 hN -

Q

656

191

5.6

608

144

3.9 5.5

64 1

196

6.2

688 11.5 720 4.6

-

hN -

Q h, Q

-

198lj85 1970190

-

1926190

hN -

1955185 1955185

Q

-

H. Guhr und E. Weher

226

The load data of the “Schmilka” border profile show a high initial load caused by sewage discharges in the territory of former Czechoslovakia. The 1992 report (IKSE, 1992a),of the International Commission on the Protection of the river Elbe (ICPE) quotes 28 polluters (> 50000 PE) with a total pollution load of 28.25 million PE on the territory of former Czechoslovakia. Out of a total of 81 polluters from the former GDR, 12.60million PE were discharged into the water bodies in the catchment area of the river Elbe. The major tributaries of the river in the former GDR were polluted by companies of the following branches of industry (see also Tab. 2): Schwarze Elster: Chemical industry, pulp industry and brown coal mining; Mulde: Chemical, pulp and paper industries, mining; Saale: Chemical, textile, potash industries, metal smelting and processing, mining of brown coal and uranium.

-

-

Tab. 2.

Survey of Direct Industrial Polluters and Sewagc Loads for Selected Parameters (1989)

~~

Industry

Chemical and pharmaceutical industries Pulp and paper industry Metal processing industry Sum

Country

Num- COD NH,-N AOX Hg her (Adsorbable organic halogcns) (t/4 (ti4 (kgia) (ti4

Germany (D) Czechoslovakia (CS)

47 358767 22460 13 44510 1245

Germany (D) Czechoslovakia (CS) Germany (D) Czechoslovakia (CS) Germany (D) Czechoslovakia (CS)

16

Sum total

349212 26 6 72950 19 675 2910 I0 18 72 108654 22486 37 120370 1315 119 829024 24801

Cd

(kg/a)

1 184.2 233.0

17053.8 4306.4 2300.0 -

667.0 55.0 0.2

0.8 0.6 0.1 6616.4 6.0 17054.8 10983.4 6.0 2300.0 19354.7 10989.4

-

1851.2 288.0 2139.2

10.1.2.1 Municipal Sewage Discharges In 1989, the total direct municipal sewage discharge (load > 50000 PE) was as follows: -

formerGDR: Czechoslovakia:

81 polluters with 12596000 PE; 28 polluters with 4956000 PE.

The former GDR had almost exclusively sewage treatment plants with only mechanical purification. In 1989, the city of Dresden discharged 1.1 million PE directly into the river without any prior treatment, because the sewage treatment plant at Dresden - Kaditz was out of operation.

Sampling in the River Elbe

227

10.1.2.2 Discharges from Agriculture Point-source discharges were caused by intensive animal breeding. However, the nutrient input resulting from high fertilizer application rates had an even greater impact on the water quality of the river Elbe and its tributaries. Considerable amounts of nitrogen and phosphorus contained in organic fertilizers and in particular in liquid manure, a waste of intensive animal breeding, as well as the excess inorganic fertilizer from arable land entered the water bodies as diffuse sources.

10.1.3 Characteristics of Water Quality The river Elbe is known as one of the most heavily polluted rivers in Europe. This applies to the loading of the river with organic substances, nutrients, salts and, in particular, pollutants in the trace range. On the basis of a standardized six-stage use-oriented classification (Technical Standard, Quality Specifications and Terms of Delivery - TGL, 1981),the analyzed river segment from the border of the Czech Republic to the city of Boitzenburg (566 km) showed the following pattern of water pollution: The measured data used for classification (Tab. 3 ) include: Organic loading: Saprobic index, 02,BOD5, COD, NH:, etc.; salt loading: Ca", M g f i , Na', C1-, SO,-, carbonate hardness, total salt content; - other area-specific constituents: Suspended solids, pH, NO,, F-, phenols, heavy metals, etc. -

The worst water quality occurred in the upper river section and in the central section downstream of the point where the river Mulde flows into the river Elbe. The upper Elbe section was important for potable water supply. The areas are characterized by oxygen contents close to Omg/L, high C O D values and severe Tab.3. Characteristics of the Water Quality of Elbe River According to 1989 Results of Classification Class Organic load

(krn)

(YO)

Salt load

(km)

(YO)

280 273

50.6 49.4

Other areaspecific criteria (km)

(YO)

1

2 3 4 5 6

354 199

64 36

553

Potentials for drinking water preparation

100

with simple treatment technology with comprehensive treatment technology with complex treatment technology unsuitcd unsuited unsuited

228

H . Guhr und E. Weher

loading with pollutants in the trace range. Intensive contamination of the water with mercury also appears in the area downstream of the Mulde mouth and is further increased by the inflow of river Saale. It is manifested by extreme sediment loads and accummulation in organisms. In addition, the river Saale carries high salt and ammonium loads into the river Elbe which, in the past, could not be oxidized because of the low oxygen contents in the water. The high zinc load of the river Elbe is the major reason for the bad classification with regard to the other area-specific constituents. The organic micro-pollutions include pesticides, organochlorine compounds, tin-organic compounds, aliphatic and aromatic hydrocarbons. There remains a severe lack of documentation regarding occurrence, origin and whereabouts of these substances.

10.2 Development of a Monitoring Strategy for the River Elbe - Objectives The water quality on the river Elbe has been analyzed for many decades. It was controlled by regional laboratories which were competent to make these analyses on the basis of binding standard specifications and had been given objectives typical for the particular region. However, until about ten years ago, no coordination occurred with regard to the stream profile. The objectives included: Collection of data showing the impact of sewage discharges and inflows into the river; - monitoring the observed raw water limit values; - acquisition of data required for classifying the water quality; - monitoring the long-term development of water quality. -

The requirements of water management resulting from the types of use, the progress made in analysis methods and the increasing cost of monitoring necessitated a new sampling strategy which has to be coordinated with regard to the longitudinal section of the watercourse. Moreover, another objective was to obtain both high quality and comparable data capable of facilitating balancing, i.e., data suitable for concentration and stream load profiles. Therefore, coordination was required betwen the various laboratories involved regarding measuring points, the range of measured variables, measuring frequency and measuring dates, preparation of samples, methods of analysis to be applied, quality assurance of laboratory work and special investigations. Four Water Management Authorities were responsible for monitoring the river Elbe from the city of Schmilka to Boitzenburg. Based on resolutions adopted on a central level, these four authorities were coordinated by the Magdeburg Water Management Authority which remained in control until 1990. The result was a sampling scheme (Anonymus, 1979) which was agreed upon on an annual basis (Tab. 4). Following German unification and thc establishment of the ICPE, this coordination work was transferred to the “Measuring and Testing Programs”, ICPE working group.

Sampling in the River Elbe

229

Tab, 4. Elbe River Sampling Schedule Quality monitoring stations: Surface water and sediment as of January 1990 Authority in charge

Monitoring station

WWD V Dresden

HPNS Schmilka 1, m, r PNS Pirna I, r HPNS Pillnitz 1, r HPNS Dresden 1, r PNS Scharfenberg 1, r HPNS Zehren 1, r PNS Strehla I, r HPNS Muhlberg 1, r HPNS Schwarze Elster, Gorsdorf

WWD IV Halle

WWD 111 Magdeburg

WWD I S tralsund

PNS Torgau 1, r HPNS Pretsch I, r PNS Wittenberg I, r PNS Coswig r HPNS RoI3lau HPNS Aken I, r HPNS Mulde, Dessau HPNS Breitenhagen 1, r PNS Schonebeck I, r HPNS Magdeburg I, r PNS Hohenwarthe I, r PNS Bittkau 1, r HPNS Tangermunde 1, r PNS Sandau 1, r HPNS Saale, GroD Rosenburg HPNS Have1 Toppel HPNS Wahrenberg I, r HPNS Cumlosen r PNS Domitz r HPNS Boizenburg r HPNS Elde Miindung

Surface water

Sediment

X

X

X X X

X

X

X

X

X

X X

X

X X

X

X X

X

X

X

X

X

X

X

X

X X

X

X X X

X

X X

X

X

X

X

X

X

X

X X

X

X

X

WWD: Water Management Authority, HPNS: Major sampling point, PNS: Sampling point, 1: left-hand river bank, r: right-hand river bank, m: center.

10.2.1 Selection of Sampling Points Various aspects decided about the selection of the sampling points: Sampling history (long series of data), accessibility (e.g., ferry), - collection of impact data of the main polluters, - assignment of reference flow levels.

-

230

H . Guhr and E. Weber

For reasons of laboratory capacity and economy, the sampling points were classified into ordinary sampling points and primary sampling points. The primary sampling points, which were defined as particularly important, included the following: -

-

-

Sampling points located at the state borders, sampling points which are important for drinking water supply, sampling points located at the inflow of larger tributaries, sampling points in the river Elbe downstream of the inflow of larger tributaries and significant polluters, sampling points located at a border line marking the transfer of responsibility from one administrative authority to another (transfer level).

The sampling stations in the river Elbe are shown in Fig. 1. Due to the frequent differences in substance concentrations on the right and left side of the river, both banks are subjected to monitoring. For the inter-state work of the ICPE only six of these sampling stations are included. These are the ones equipped with automatic stations (see below). Since 1990, the remaining stations have been monitored under the sovereignty of the newly established German federal states.

10.2.2 Range of Measured Variables The monitored variables should document the pollution load of the water body comprehensively and classify its quality state. The growing range of monitored variables dictated a measuring program subdivided into two separate programs. First, a basic measuring program which had to be executed at all measuring stations on every measuring date, and second, an extended measuring program the additional had to be criteria of which - compared to the basic measuring program determined at the major sampling points on predetermined dates. The basic measuring program included variables such as flow (carried out by the Hydrological Measuring Service), water temperature, acid consumption, suspended solids, total phosphorus, ortho-phosphate, and spectral absorption coefficient (254 nm) in addition to the variables under classification mentioned above, except for heavy metals and phenols. Moreover, the extended measuring program included the heavy metals commonly determined, colony counts, colony counts of coliform bacteria, TOC, DON, TON, AOX (adsorbable organic halogens), anionic and non-anionic tensides, chlorophyll, phenols and extractable substances. Testing the sediment for heavy metals, loss on ignition and total phosphorus were also included. Due to the lack of instrumentation for analysis, organic trace elements contained in the water could only be determined in the framework of special measurements carried out by cooperation partners before 1990. Basically, the current measuring program of the ICPE includes the criteria described and, additionally, criteria such as the BODzl (relevant for the North Sea), potassium, phytoplankton volumes, arsenic and representatives of organic groups ~

Sampling in the River Elbe

23 1

of substances such as benzene, trichloromethane, hexachlorobenzene, PCB, dimethoate, simazine and others. To date, the international program does not include decisions regarding DON, TON, BOD,, C O D and sediment testing.

10.2.3 Measuring Frequency The Standard Specification on the classification of flowing waters which was in operation until 1990 stipulated 26 tests per year. This level of testing was required because the two subsequent worst values of a year (organic load) and the 90% value of the cumulative frequency (salt components, other criteria) were used for evaluating a water body. Due to reasons of organizing the work, this number of tests was reduced to a maximum of 24 tests per year. For studies of the longitudinal section of the watercourse, flowing time-related sampling is desired. In routine monitoring of a water body, such a requirement cannot be satisfied. In addition, the flow characteristics may change in as little as seven days throughout the 566 km of the river under investigation, i.e., the downstream tributaries may cause a rise in the water level in the lower zones of the flowing stretch of the river Elbe due to precipitation, which might falsify the concentration and load courses. For this reason it was decided that all test laboratories would execute sampling at the same day. The first and third Tuesday of each month were selected. It had also been determined that the extended measuring program was to be carried out at the major sampling points on the first Tuesday of each month. The current measuring program of the ICPE specifies 13 samplings per year. They are taken uniformly on every consecutive fourth Wednesday. The Magdeburg monitoring station has been included in the research work since many years. Tests are carried out on a weekly basis, including a wide range of criteria to qualify flow-dependent seasonal and long-term impacts.

10.2.4 Use of Automatic Monitoring Stations Prior to the year 1990, automatic monitoring stations were used in the former GDR in the cities of Schmilka, Magdeburg, Cumlosen and Boizenburg. The quasicontinuous measurements served operational monitoring of the water body. The standard measuring program included measurements of water level, oxygen content, water temperature, electrical conductivity, pH, turbidity and toxicity (fish test). Data transmission was either done by radio or by cable. When the tested parameters exceeded the limit values or fell below them, automatic samplers took water samples for further laboratory testing. Currently, the automatic monitoring network is being upgraded. The automatic stations in the new federal states which simultaneously are sampling points in the

232

H . Guhr and E. Weber

framework of the ICPE monitoring network are located at Schmilka, Zehren, Magdeburg and at the inflows of the tributaries Schwarze Elster, Mulde and Saale. In addition to the permanent automatic collection of parameters such as the ones mentioned above, weekly composite samples are taken and tested in the laboratory for organic loads, heavy metals and organic trace substances. These analyses form the basis of accurate load determinations. Mobile stations are used for special investigations. Using two cooperating mobile stations, the oxygen hydrographs were measured in the river Elbe upstram of the city of Dresden in order to be able to quantify self-puritication (Maedler, 1988).

10.2.5 Data Flow, Data Processing, and Evaluation Up to the year 1990, the Ministry for Environmental Protection and Water Management was responsible for the monitoring networks of the Meterological Service and water management. Essential parts of the planned and applied monitoring and information system have been utilized for quite some time in investigating particular river sections. This system contains the following partial monitoring networks: Hydrology and water quality, the latter being subdivided into a laboratory for water quality and an automated monitoring network. In addition, there was the monitoring network of the Meteorological Service of the former GDR. The common meteorological monitoring networks included the stations for measuring air pollution as well. Data and information were delivered through the “Water Management Information Flow Line” covering the following functions: Acquisition and collection of measured data, data encoding and transmission, - data decoding and processing, - use of data for reporting and operative decision-making, - data storage and data output for further utilization (e.g., research activities), -

-

Linking the individual steps of data processing is clearly presented in a flow chart by Goltermann et al. (1983) (Fig. 2). Information and data thus collected can be organized in various ways. Based on the origin of the data, the following division is possible: quantitative aualitative

-Water body data

Data and information regarding the catchment area Further distinctions are possible on the basis of the following principles of classification: Original or derived data - current or historic data, etc.

Sampling in the River Elbe Information Input

Processing

233

Storage

Field observations (photographs, survey plots

Archive

Encode

\

h e p m for storage

susp. sediment, samples)

1 hepare for laboratory analyses

sample library

-------)

I

-

1-

I Measurements

turbid.qH, depth,

r--------

1 Interpreted data

1 b

IF

Fig. 2.

I

Archive

,

Data handling procedures (prepared by Goltermann et al., 1983).

According to the structure of the Water Management Authority, the measured data and the results of analysis came from the laboratory and were used by the Water Pollution Control Agency for approving sewage discharge and for monitoring water quality. Comprehensive sets of data which were required for these services were stored as annual files in the archives of the computer center.

10.2.6 Special Tnvestigations Special investigations which, in most cases, were initiated by research and frequently carried out in cooperation with the authorities served to clarify particular problems. Measuring activities of this type focused on collecting data about longitudinal and

234

H . Guhr und E. Weber

transversal mixing of discharged substances, the variation behavior of water pollutions, transport velocities, bioaccumulation of pollutants in fish, dynamics of suspended solids, the content of the alluvial soils of micro-pollutants, etc.

10.3 Experience Made in Implementing the Sampling Strategy - Representativeness of Sampling Points The representativeness of the quality monitoring points should be viewed in the light of measurement objectives (see above). The variables describing both water volume and quality are to a large extent characterized by natural impacts (e.g., meteorological, hydrological and hydrogeological) and by anthropogenic impacts (e.g., utilization by agriculture, industry and trade, communities, water use and sewage discharge, etc.). Wherever these impacts change intensively the river’s behavior or state, measuring stations are required to monitor and record these changes. In addition, there are sampling points which are established due to legal or national or international obligations. The specific impact of the river Elbe on the representativeness of sampling points became clear during the evaluation of the data: The turbulence of flowing which varies considerably in the individual sections is only roughly known and not verified by measured values. Tracer measurements, carried out under the control of the Berlin lnstitute for Water Management, that calculated vertical, lateral and longitudinal dispersion, revealed the lack of knowledge. Plume formation caused by the entering Saale river can be proved by cross-profilc measurements along a mixing stretch of about 100 km. Unsuccessful tracer measurements point out that the theoretically expected rapid mixture at a greater depth does not occur. However, despite an extremely low concentration of dissolved oxygen, an anaerobic state does not occur, partially due to the vertical water exchange. Comparing the calculations of the pollution loads on the basis of the measured values of a profile with right and left bank values with the calculation of the pollution load based on 15 laterally subdivided partial currents in the upper section of the Elbe river. Eidner and Spott (1992) showed that the pollution values obtained for point-type industrial large-scale discharges by the first type of calculation has the potential for significant errors. However, due to the cost it is impossible to carry out routine cross-profile measurements for the mixing stretches of all larger polluters and tributaries with dcnsely located sampling points. In practice, this means that for any further evaluation of data including water quality models these limitations must be considered. If the measuring stations are to demonstrate the impact of the partial catchment areas on the quality of the Elbe water, the structure of the water system should be considered. The spatial distribution of precipitation events largely influences the runoff. Applied to the river Elbe, this means that heavy rainfalls in the low-mountain range, accommodating the springs of the river Elbe and its tributaries, will have a

Sampling in the River Elhe

235

different impact on the mass transfer process by flood waves than the rain falling in the central and lower sections, of the river. The geogenic export of matter from the catchment area is influenced by this structure of the river system. Due to the considerable decrease in anthropogenic discharges, geogenic inputs have regained importance in affecting the water quality. This is also true for the diffuse inputs caused by agriculture and past ecological liabilites causing soil contamination. A good example of the relationship between flow rates of tributaries and the main river and salt concentrations, verified by recent cross-profile measurements (November 1992), is the inflow of the river Saale mentioned above. Upstream of the inflow of the river Saale, the river Elbe belongs to the hydrogen carbonate class based on its principal ion content. The river Saale and its tributaries are severely polluted by chloride, both geogenically and anthropogenically, and changes the ion composition of the Elbe water with decreasing intensity during mixing. This process is clearly dependent on the flow ratio of Elbe and Saale. This means, however, that there cannot be a firm relationship of regression between chloride concentration and Elbe flow downstream of the inflow of river Saale. The same conclusion must be drawn for substances contained in sewage discharges. Therefore, in exceptional cases it will be possible to replace missing concentration values in time series by values calculated from regression functions. If the pollution load values, which are the product of flow and concentration value, are used in these calculations, the apparently good adaptation of the regression function is, in most cases, due to the spurius correlation and, consequently, of no value. The available data on sediment samples are extremely scattered which can be explained to a limited extend by causal reasons. Therefore, it is difficult to include the sediment into the evaluation of representativeness. The concrete positioning of the sampling points on site is very important for its representativeness: It shall be possible to take dipper samples at the same profile in the case of both high water and low water in the same representative way. - Analogously, this applies to the positioning of probes and the location of the suction strainer in the case of automatic sampling points. - In addition, the varying density of the transported suspended solids over the cross-profile should be taken into account when the undissolved substances are to be determined. -

The conditions described above as necessary for ensuring the representativeness of the sampling points are so great in number and make the measurement so costly that compromises must be made in almost every case.

10.3.1 Frequency of Measurements The frequency of measurements should simply follow the temporal variability of the data to be determined if statements about the impact on water quality are intended. Even a static method of assessing a particular water body should be

236

H. Guhr (2nd 8. Weher

evaluated by an error discussion to avoid unfounded conclusions. The well-known factors influencing water quality which statistically must be considered as time series of stochastic character include: -

-

-

-

-

-

-

-

Day/night rhythms of variables which are influenced either biologically or by sewage discharges. Annual cycles of natural variables caused by influences varying with the seasons. Intensive sewage discharges with rhythms determined by production such as processing plants for agricultural products (e.g., sugar factories) or batch-type discharges from chemical plants. The runoff characteristics including high water and dry weather events, the impacts of which vary considerably. Low water levels in large rivers are caused by extended periods without rain. Floodwaters have much higher dynamics and vary considerably in their behavior and affect transport and chemical reactions of the substances in the river. With increasing size of the catchment area, the influences of streamflow and sewage discharge will be balanced. Consequently, measurements must be carried out more frequently for small catchment areas than for larger ones, although the sampling point is often not very important. Finally, it should be pointed out that the potential for failure of individual data of a measurement or the lack of a total set of data on a measuring date requires a higher frequency of measurement than envisaged in the rhythms contained in the time series. Particular biological parameters, such as the saprobic index, are so dependent on the annual climatological cycle so that data permitting the comparison of various sampling points can only be derived from the summer and fall samples. Sampling in the months of March, June, September and December proved successsful. For examination of chlorophyll a, sampling during the vegetation period from May to October provides data that can be compared. For sediment samples outside the main stream, four samples a year should be taken in stagnant water and two samples in flood plains with the high-flood sediment samples to be taken immediately following the lowering of the spring flood in March/April. It is desirable to take weekly water samples from the main profile. These individual samples should be backed by the quasi-continuous sets of data acquired by the automatic monitoring stations, provided the sampling point and the monitoring station do not have fundamental differences in collecting the quality data. Since the quality of Elbe water depends largely on sewage discharges, thc sampling of the sewage discharges should be carried out in a timely relation to the sampling of the flowing water. However, the stronger variability of the sewage composition calls for more frequent sampling. In general, larger sewage discharge should be monitored by automatic monitoring stations.

Practical experience shows that the demands of representativeness and frequency of measurements made at sampling points and automatic monitoring stations can only seldom be satisfied. The most important obstacle is the high cost of invcstrnent

Sampling in the River Elhe

231

and the very high operating expense. For this reason, data evaluation requires specialists who are able to assess the water quality behaviour of the river Elbe in its total complexity.

10.3.2 Data Collection The individual steps of data collection reveal various drawbacks which are responsible for the poor reproducibility of the results within a laboratory and lead to concentration jumps which cannot be explained by the substance balance of a water body between the areas under investigation by various laboratories.

10.3.2.1 Sampling For sampling, the sampling vessels are filled below the water surface, against the current. If a large vessel (e.g., a bucket) is used for sampling, there is a risk that when filling water for specific analyses subsequently into small sample bottles that the substances which can be filtered off may have already settles in the large vessel. It proved to be better to fill each bottle separately in the water body. In the case of small volumes, as required for determining the germ count, large flocs which might get into the bottle accidentally may falsify the result when such a floc or part of it is transferred into the Petri dish by pipetting. Sewage discharged into the river Elbe forms sewage plumes. The tributaries need extended flowing sections until complete mixture is achieved. At severely polluted sections, this plume formation results in a non-linear concentration distribution across the cross-section of the watercourse (Fig. 3).The average substance concentration from the left and right bank multiplied by the flow does not allow to calculate accurate pollution loads. Another falsification occurs if the river bed is not symmetrical, for example, flatter at one side of the bank than at the other. In such cases, the accurate pollution load can only be determined by combined investigations of flow and quality, i.e., the cross-section of the river will be subdivided into lamellas. The concentrations and partial flows which are measured for these segments are used to determine the partial pollution loads. All of the partial pollution loads are added to obtain the total load. However, the result of such an exorbitant procedure is only applicable to the particular hydrological and polluter situation. The stochastics of sewage discharges cause short-term variations of concentrations in the river Elbe (100% and more with regard to the minimal value within 24 hours, as detected by Eidner and Spott, 1992). These variations in concentration are more pronounced the more polluted the river section and the lower its flow rate. A single water sample reflects the current state only. Composite samples (combined single samples from particular time intervals) are better suited to describe the current load status. The total duration depends on the concrete task and the application of automatic sampling devices (e.g., two hours or one week). A spatial composite sample (across the river cross-section, for example) is rarely used at the river Elbe.

H . Guhr cind E. Wehrr

238

- 130 - 120

-110

50

- 100

45

-

- 80

- 40 -g 35 \

-10

5 30 -- 25

-60

g

15 10

\

p5:

- 50

\

-g 20

u

\

m

U

90

-

‘t \

5

‘\\ *--*\--

30

-= DOC 20 --. Fig. 3. Cross-sectional distribu-.----d-.---_’-- - - .-’ tion of concentrations of organic -10 h .

BSB,

o L # 3 1

40

1

I

3.5 9.5 11 21 31

I

--

,

41 51 17 Distance from river bank (m)

,I 0

substanccs at km 44 on river Elbe (Pillnitz).

9 1 104.5

As regards suspended matter and heavy metals and organic components sorbed at the latter, the impact of shipping and increased water levels due to floods on sediment remobilization have to be considered. In periods of no navigation, the concentration of suspended matter and heavy metals measured are much lower than in times of intensive shipping. Cyclical behavior of the suspended matter concentrations are observed in the case of floods (Fig.4), i.e., an increase in concentration with increasing excess water up to a maximum value which lies before the flood crest followed by subsequent decrease to a low content remaining on that level during the decline in the water volume. The cause is to be found in the stirring up of remobilizable sediments. If the corresponding stagnant water areas are emptied, the matter content which can be filtered off decreases and the Secchi depth will improve. Influences of this type can only be recorded by sampling in short-time intervals (e.g., shipping impact on an hourly basis, excess water on a daily basis). Determining the saprobic index for evaluating the biological quality of a water body requires investigation of the macro- and microzoobenthos, i.e., the animals living on the stones in a watercourse are examined. If no organisms are found, toxic effccts from sewage discharged upstream of the point of examination may bc the cause. However, the stones may be covered with deposits (e.g., lime, soot) which are hostile to colonization or the stones themselves brought into the river by man are hostile to colonization (e.g., new copper slag stone). In sampling, these circumstances must be checked to avoid any false interpretation of the results. Since the saprobic values are slightly higher in the six months of winter, the specified dates of sampling must be observed with regard to the stream profile.

Sampling in the River Elhe

% 2

400

600

239

800 1000 Discharge (m’/s)

1200

1400

800 1000 Discharge (m3/s)

1200

1 00

\ cn

-a

g 30-

.-

+ m L + c

:20 Y 0

W L

n

Fig. 4a, b. Cycle of concentration of (a) suspended matter and (b) total copper (high water in winter 1988/89).

-8 1 0 m

c 0

I-

0 400

-

1 600

Sediment sampling is carried out preferrably in the second half of the year since there is a dynamics of sediment loading influenced by the flow characteristics. After floods which clear out the fresh sediments, sediment analyses for heavy metals produced lower values than in August/September when new suspended matter containing heavy metals is deposited. The water levels are low during this time and stagnant water areas with deposits are easier to reach. Sampling should be carried out with grippers or preferrably by hand because the top layer and the fine-grain constituents ( 1 2 0 pm) which are of interest can thus be located by touch.

240

H . Guhr and E. Weher

10.3.2.2 Preparation and Preservation of Samples Treatment requirements for samples were determined in agreement with the Elbe river laboratories. The most important stipulation regards filtration, specifying the type of filter to be used (e.g., membrane or glass-fiber filter), i.e., whether the total content or the dissolved is of interest. For heavy metals, the total content was stipulated because the detection limits of the methods of analysis applied are much too high for determining the dissolved contents. In addition, for practical purposes, the causes of the total loading are of interest. Heavy metal compounds which are sparingly soluble can partly be remobilized in the water body by various mechanisms. Trace analysis requires to specially clean the sampling vessels before use (for example, by steaming out to detect chlorine-organic compounds). Nitric acid is filled into the receiving flask prior to adding the samples for heavy metal analysis (total content). For ortho-phosphate determination, the sample must be filtered on site. Moreover, the BOD determination should also be prepared on site. For reasons of labor organization, it was agreed that the laboratories located at the river Elbe would prepare BOD in the laboratory in the evening of the day of sampling. However, some of the laboratories did not follow this instruction and started work the next morning, thus making the interpretation of the results more difficult. To obtain reliable results, the samples must be transported from the location of sampling to the laboratory and processed without any delay. A certain time of storage is acceptable for conservative substances and in determining total contents if microbial conversions can be excluded. As regards the nitrogen components, changes by nitrification, denitrification, ammonolysis and incorporation into biomass might occur. Relevant works of water analyses (NormenausschuD etc., Deutsche Einheitsverfahren zur Wasser-, Abwasser- und Schlammuntersuchung, 1993; Breitig and von Tiimpling, 1982) provide detailed information about how to preserve the samples on site and about the possible duration of storage after sampling. The composite samples collected in automatic stations can be deep-frozen immediately. In determining the total contents of substances, problems may occur when large flakes are contained in the water sample and only small volumes are required for the course of analysis (e.g., TOC). Prior to analysis, homogenizing with a heavy-duty stirrer is required to smash the flakes.

10.3.2.3 Chemical Analysis Before agreement was achieved about the methods of analysis to be applied along the longitudinal section of Elbe river, it was obvious that different methods permitted in compliance with the standard regulations produced strongly varying results for a tested parameter. This was particularly grave in the case of detecting ammonium. Some laboratories applied the photometric method using Nessler’s reagent, others used the indophenol blue method. In the case of strong organic loading, as is the case in the river Elbe, the first mentioned method of analysis produces exceedingly high values. Therefore, the indophenol blue method was stipulated to be used for ammonia analysis. Likewise stipulations were made for the other parameters for

Sampling in the River Elhe

24 1

the quantitative detection of which several methods are permitted. For example, for determining the chemical oxygen demand, the potassium dichromate method using mercury and silver sulfate were declared as binding. One laboratory did not adhere to this agreement and determined C O D with the semimicromethod. Therefore, the results showed concentration jumps on the section of the river examined by this laboratory as compared with the upstream and downstream sections. For long periods the pulp industry discharge its black liquors into the river Elbe. They contained reductive compounds such as sulfides and sulfites causing “oxygen holes” in the river Elbe downstream of these discharges. BOD could be falsified at such sampling points by chemical autoxidation. In detecting the organically bonded total nitrogen, the latter will be converted, by decomposition, into ammonium. It is the difference between the ammonia nitrogen content of the sample after decomposition and the original ammonia nitrogen content of the sample. Since the original ammonium concentration is determined from the filtered sample, however, this nitrogen component may also be sorbed on the suspended matter, and the total ammonium must be determined separately for the TON-analysis. In the current international measuring program, there is an agreement on methods of sample preparation and analysis to be adhered to.

10.3.2.4 Measuring Errors There is a large number of potential measuring errors. Considering the fundamental differences in data acquisition, a basic distinction between measuring errors made in individual sampling automatic permanent monitoring must be made. Since the cost of determining a measured value is rather high, certain methods have been introduced attempting to prevent measuring errors. The quality of laboratory work is of prime importance because it influences the reliability of the measured values considerably. The laboratories which are subordinated to the Ministries for the Environment try to obtain accreditation by providing the necessary prerequisites, for example, a quality assurance system (QA-system, for short). The aim is to adopt the QA-system called “Good Lab Practices” or “Standard Series 45 000”. To do so, the financial expenditure for the water laboratories in the five new eastern states of Germany is considerably in many cases. To obtain values from analysis which are comparable and of high quality, interlaboratory trials were carried out for the laboratories of the administrative authorities along the river Elbe. However, due to reasons of cost, the scope of these interlaboratory trials was insufficient. It is preferabe to handle analytical quality assurance for the Elbe river laboratories in accordance with the Collaborative Testing for Analytical Quality Assurance adopted by the state of Baden-Wurttemberg. The Institute for Hydraulic Engineering, Water Quality and Waste Management of Stuttgart University is in charge of carrying out these interlaboratory trials, thus ensuring uniform and unbiased evaluation. As a rule, a total offour collaborative tests per year are made including 20 parameters in compliance with “Deutsche

Total COD

-40

-20

0

20

40

03.09.1991 10.09.1991 17.09.1991 24.09.199 1 01.10.1991 08.10.1991 15.10.1991 22.10.194 1 29.10.1991 05.11.1991 12.11.1991 18.11.1991 26.11.1991 03.12.199I 10.12.199 1 17.12.1991 07.01.1992 15.01.1992 22.0 I . 1992 29.01.1992 04.02.1992 12.02.1992 19.02.1992 26.02.1992 04.03.1992 10.03.1992 18.03.1992 25.03.1992 01.04.1992

60

80 2.9.-8.9.1991 9.9.-15.9.1991 16.9.-22.9.1991 23.9.-29.9.1991 30.9.-6.10.1991 14.I0.-20.10.1991 21.10.-27.10.1991 I1.11.-17.11.1991 25.11.-1.12.1991 2.12.-8.12.1991

27.1.-2.2.1992 3.2.-9.2.1992 17.2.-23.2.1992 24.2.-1.3.1992 2.3.-8.3.1992 9.3.-15.3.1992 23.3.-29.3.1992 30.3.-3.4.1992 6.4.- 12.4.1992

08.04.1992 15.04.I992 23.04.1992 29.04.1992 06.05.1992 13.05.1 992 20.05.1992 26.05.1992

20.4.-26.4.1992 27.4.-3.5.1992 4.5.-10.5.1992 11.5.-17.5.1992 18.5.-24.5.1992 25.5.-31.5.1992 1.6.-7.6.1992 8.6.-14.6.1992 15.6.-21.6.1992 22.6.-28.6.1992 29.6.-5.7.1992 6.7.-12.7.1992

03.06.1992 10.06.1992 17.06.1992 24.06.1992 01.07.1992 08.07.1992 15.07.1992 23.07.1992 29.07.1992 05.08.1992

20.7.-26.7.1992 27.7.-31.7.1992 3.8.-9.8.1992

Dates of taking individual samples

Dates of taking weekly composite samples t-

ACOD

-tIndividual samples -9-

Weekly composite samples

Fig. 5. Comparison of COD concentrations of samples taken at different times. Sampling poini: Elbc river, left-hand side km 3 18, at Magdeburg-Westerhuesen (analytical data from State Authority for Environmental Protection, Magdeburg).

Sampling in the River Elhe

243

Einheitsverfahren zur Wasser-, Abwasser- und Schlammuntersuchung” (German uniform procedures for water, sewage and mud testing). The complexity of the inorganic and organic matrix of water and sediment samples from the river Elbe must be assessed as very high and changes drastically in the individual sections of the watercourse. This is why the instrumentation of the individual Elbe laboratories is not consistent from one laboratory to another. However, this makes comparisons of reliability and the error range of the values determined by analyses more difficult. Trend analyses made on a 10-year series of measurements showed a change in the methods of analysis very clearly because both the annual averages and - even more pronounced - the standard deviation varied significantly after the hydrological specificity of the individual year had been eliminated. This must be considered when combining data determined with the help of different methods. As an example from practice, the combination of values derived from field experiments (often colorimetric) with individual determinations (often spectrophotometric) and continuous measurement techniques (using ion selective electrodes) should be mentioned. In addition, the type of sampling, be it as an individual sample or a weekly composite sample produces data differences, as is shown in Fig. 5, which cannot be explained easily. Even a slight difference in the position of the sampling point is very important. Moreover, the type of sample preservation was also noted as a source of systematic error. Based on the origin of analytical error, the following sources of error can be mentioned: The sampling point is not representative (see above). The sampling technique is incorrect. Delayed sample preservation causes reactions in the sample which can no longer be recorded (e.g., turnover with gas exchange). - The temperature of the sample changes by the time of analysis. - Samples are mixed up. - Sample vessels are not suitable. - Laboratory chemicals are too old. - Laboratory personnel works incorrect. - Wrong or defective instrumentation for the quality parameters to be determined. - Mistakes in writing and insufficient statement of unit of measurement in recording the data. -

-

The sources of error listed above are often found in practice, but only if sufficient data checks are made and the errors are very obvious (see also Fig. 2).

10.3.3 Consideration of Sewage Discharges Monitoring of watercourses and sewage discharges should be seen in close connection and should therefore be coordinated because the water management aims to remediate the water bodies. While reasons of data protection currently prevent dis-

244

H . Guhr and E. Weher

closing of the concrete state of emissions, prior to 1990, the potential for monitoring geared to the particular requirements of water pollution control were missing. The sewage outflows were not equipped with flowmeters. Large enterprises often used several points of discharge. Due to batch-type operation, concentration and loading of the sewage varied considerably. The water pollution control administration did ~

I

3

Asellus aquaticus

2 1 -

* o ._

Erpobdella octoculata

W u

n

* 3

Lc

0 2 21g 0 U W

Glossiphonia heteroclita

LL L

3 2-

~

1-

0 0. c

cu

A 4-

-

Fluorides

3-

' #

L 2 -

P

-

1 -

Organically bonded halogens Total content (Absorbable organically-bonded halogens)

250

200 d

\

5

150

m

a

z 100 50

- -

I

0' River kilometer

318

319

320

Fig. 6. Depopulation zone as longitudinal section of the river Elbe. Frequency of macroorganisms, fluoride and AOX concentrations upstream and downstream of the Fahlberg-List company, Magdeburg (July 25, 1989).

I

321

322

Sampling in the River Elbe

245

only consider those parameters for which limit values were specified. The analytical instrumentation required for detecting pollutants in the water body as traces was not available. The results of sewage discharge could be documented in the water body by macrozoobenthos colonization upstream and downstream of the place of discharge. In addition, the frequency of the occurring species was estimated. Downstream of the points of discharge the spaces of colonization were biologically dead due to the toxicity of the wastewater. Depending on the animal species and the dilution of the sewage plume, the depopulation zones varied in expansion (Fig. 6).

10.3.4 Data Processing and Evaluation In general, data processing and evaluation are determined by the following items: -

What is the objective of data evaluation? What water body do the data come from? Which technique was used to collect the data? Can the data be used immediately or only in the future?

With regard to their origin, the water data can be classified as field observations, field measurements and field samples. The latter are used by the laboratories to determine the measured values of the desired quality parameters. All data and information should be checked for errors. Normally, this is done first before the data are saved in files or data banks and subsequently through debugging programs before the data are finally stored in data or file memories. If the measured values are used for topical forecasts or to control processes of ensuring water quality, error checks will be carried out prior to making the forecast or giving the control signal. An important step in data processing is data compression which helps to clearly show essential properties or trends of the water quality processes. In other words, primary data will become secondary data. The data are further processed by using them in reports, maps, tables and diagrams. Trend analyses reveal behaviors over several years or within one year using differing mathematical functions (Fig. 7). To calculate long-term trends of water criteria with inert behavior, the homogeneous time series should be at least five years long. If extreme situations accumulate in the random sample of a time series, longer data series are required. For the operational water and environmental authorities, the individual steps of data processing are laid down in data flowcharts to obtain results which are uniform and comparable with regard to time and space. Among others, binding regulations of this type were laid down in the former GDR for: -

-

The method of analysis to be applied (Selected Methods of Water Analysis, 1982); coordination of sampling over the entire course of Elbe river (ELBE River Testing Program, 1989, Sampling Schedule). electronic data processing including error checks for the preparation of annual and five-year reports on water quality and for water management.

H. Guhr and E. Weher

246

500

1

”i -ub 200

x

JI

x X

100

0

10

10 20 30 40 50 Time series, monthly date, 198511990

I-’

60

-1

I

I i

u/

0

X

20 30 40 50 Time series, monthly date, 1985-1990

10

60

Fig. 7. Trend analysis using smoothing functions. Linear (-) and polynominal (. . . . . .) smoothing curves, for chloride and dissolved oxygen; ( x ) measurement.

At present, annual reports are published for the sections of Elbe river falling into the individual German states. In addition, the International Commission on the Protection of the River Elbe issues an annual report entitled “Elbe water quality report - from source to mouth”. Significant changes in the use of the watercourse or sewage discharges can be tested for their impact on the quality of the river water by means of mathematical simulation models. The processes which are important to simulation are discovered through system analyses which are normally quite costly. For this reason, reliability and usefulness of the measured values should be further investigated. Quite often it is found that additional measured values to be determined by special measurements are required. The expenditure required for data acquisition, system analysis and model preparation by far exceeds the cost of model application.

Sampling in the River E h e

247

10.4 Conclusions Many years of experience in using the Elbe river sampling program have proven that efforts should be focussed on the collection of high-quality data which can be used multivalently.

- For this reason, coordination of testing activities is required if several laboratories are involved in monitoring a water body. Sampling points, dates, preparation and preservation of samples, methods of analysis to be applied and data processingand evaluation must be coordinated. - Quality assurance of laboratory work should be considered a permanent task and should be proved by regular collaborative tests.

-

The coupling of measured values from individual sampling points and automatic monitoring stations, which is both necessary and favorable, should be thoroughly examined for data differences. The objectives of the monitoring programs decide about the spectrum of parameters to be monitored and the frequency of measurement. Compared with the desired and statistically required expenditure, the real measuring expenditure which must be greater in severely polluted and intensively used water bodies than in clean rivers, will be determined by the economic boundary conditions. Composite samples are suited best for determining pollution loads whose decrease has proven the efficiency of sanitation and other management measures. However, a large number of random samples or continuous measurements are required to substantiate the various types of use by demonstrating the probability of exceeding or falling below limit values. In the form of the pollution loads, water quality management includes water quantity management. A representative flow rate monitoring station should be allocated to each quality monitoring station. The limit values which are derived from ecological objectives and legal stipulations should be monitored by specific analyses. To ensure the representativeness of sampling points, the transverse and depth distribution of matter concentrations as well as the impact of the meteorological and hydrological characteristics in the partial catchment area must be known.

10.5 References Anonymus (1 989) Clntersuchungsprogramm Hbe 1989, Probenahmeplan Wasserwirtschaftsdirektion Untere Elbe, Magdeburg, 1989, unpublished. Breitig, G., Legler, Ch., Tiimpling, W. von (eds.) (1982) Ausgewahlte Methoden der Wasseruntersuchung, Bd. 1, chemische, physikalisch-chemische, physikalische und elektrochemische Methoden; Bd. 11, Biologische, mikrobiologische und toxikologische Methoden, VEB Gustav Fischer Verlag Jena. Eidner, R., Spott, D. (1992) Die Ermittlung des Stofftransportes an ungleichmaDig belasteten MeDquerschnitten (Pollution Load Distribution at unevenly Loaded River Cross-Sections), Deutsche Gewiisserkdl. Mitt., 36, H. 1, MCrz 1992.

248

H. Guhr und E. Weber

Goltermann, M. L., Sly, P. G., Thomas, R. L. (1983), A Guide,fbr the collection and intwprc.tution of sediment quality data. UNESCO/Technical Papers in Hydrology, No. 26. Guhr, H., et al., (1991) Rechnergestiitztes Berutungs- und Informcrtionssystem, Elhestrom: Pro:<$nnalyse, In~~rmationssystem, Modellsystem. SchluDbericht zum FIE-Vorhaben 30 F 10 19-4, Forschungsbereich der ehem. Wasserwirtschaftsdirektion Magdeburg (unpublished). Guhr, H., Biittner, O., Dreyer, U., Krebs, D., Spott, D., Suhr, U., Weber, E. (1993) Zusunzmenstellung, A uscwrtung uncl Bewertung des vorhandenen Dutenmateriuls iiher die stuflliche Belustung der Gewassergiite der Mittelelhe nach einheitlichen gemeinsamen Kriterien (Vorstudie). Band I und 11, GKSS-Forschungszentrum Geesthacht GmbH, Geesthacht, GKSS 93/E/18. IKSE (ed.) (l992a), Aktionsprogrumm Elbe - INVENTAR wichtiger Abwassereinleiter im Einzugsgebiet der Elbe im Jahre 1989, Aktionsprogramm Elbe, Internationale Kommission zum Schutz der Elbe (IKSE), Magdeburg. ELBE, Z Y Y O j l Y U I . Internationale Kommission zum Schutz IKSE (ed.) (1992b) Gei~~usserglteherichr der Elbe. Imhoff, K., Imhoff, R. (1985) Taschenhuch der Stadtentwusserung. 26. iiberarbeitete Auflage, R. Oldenbourg-Verlag GmbH, Miinchen. Madler, K. et al. (1988) Erinittlung der Auswirkungen verandorter Ahwussereinleitungen uiid AuJytellung iikologisch hegrlndeter Grenzwerte f u r die ohere Elhe hinsichtlich der orgunisehen wid toxischen Belustung, des Stick,sto~Jhuuskultesund des Selbsrreini~ungsverm~gens. F/ E- Bericht, TU Dresden (unpublished). NormenausschuR Wasserwesen (NAW) im DIN und Fachgruppe Wasscrchemie in der Gesellschaft Deutscher Chemiker (1 993) Deutsche Einheitsverfahren zur Wusser-,Abwasser- und Schlummunrersuchung. 1. his 28. Lieferung. VCH-Verlagsgesellschaft, Weinheim. TGL (1981) Nutzung und Schutz der Gewasser, Klassifizierung der Wasserbeschaffcnheit von FlieBgewCssern (Use and Protection of Water Bodies, Classification of Water Quality of Flowing Waters). Fachbereiclzsstundard TGL 22764. Minist. f. Umweltschutz und Wasserwirtschaft, Berlin (ed.).

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

11 Sampling Treated Wastewaters and Receiving Streams James E. Norris

11.1 Introduction Analytical chemistry is often a major tool of jurisprudence and political action in today’s regulatory climate. Unhappily, those who use the results of the analytical art most are often those who least understand its limitations. This chapter addresses such a fundamental limitation: the validity of a chemical analysis is limited by the validity of the sample chosen for characterization. No amount of statistical transformation of data, application of sophisticated analytical methods, or imposition of onerous quality assurance and quality control measures in the laboratory can transform a bad (nonrepresentative) sample into a good one. Sampling is an attempt to choose and extract a representative portion of a physical system from its surroundings. The subsequent analytical characterization of the sample is intended to define certain properties of the sampled system. This characterization involves extraction of the sample from the physical system of interest, which is the focus of this chapter; preservation and shipment of the field-collected sample, and in-laboratory subsampling of the field sample and subsequent analysis of the selected aliquot. The variables affecting the objectivity of this characterization are profound and varied. This chapter addresses this author’s experiences in the sampling of many diverse water and wastewater streams over a period of 35 years. This chapter is not all-inclusive, but rather addresses certain situations where alternative sampling methods were evaluated and where some conclusions about these alternatives were drawn.

11.2 Sediment Sampling Industrial wastewaters and waters in receiving streams are never pure solutions. Substantial suspended particulates are generally the rule, and rarely are these suspensions stable. When stream velocity and agitation decrease, particulates settle to the bottom. Native bottom particulates and fine clays are also exposed to the components in wastewaters via processes such as absorption, adsorption, and occlusion, which bind these components to bottom materials. Such binding is rarely static; oxidation-reduction and near-surface photo-decomposition affect particle surface composition. Thus, such a stream bottom is characterized by variously intermixed layers of native material and waste sediments. This system is of interest in characterizing the impact of the wastewater on the local environment.

250

J . E. Narris

Sampling of such sedimentary materials is inevitably an exercise in insightful grab sampling. The techniques most used fall into two broad categories: bottom grab (dredge) sampling and core sampling. Abundant variants and devices of each type are available. Bottom grab samplers have the advantage of obtaining a larger sample over a broader expanse of bottom. These samplers are generally easy to use. The greatest disadvantage of such samplers resides in the loss of finely divided particulates, which are carried away by outflowing water from the sampler. Indeed, for many streams, the bottom or near-bottom surface “fines” may be of most interest in stream quality characterization. In such circumstances, a core sumpler represents a suitable alternative. When the closing mechanism or valve is engaged, waters in contact with the fines are entrapped and cannot carry the tines away. A major disadvantage of core samplers lies in the extremely small area of bottom encountered. As a general rule, more core samples of a bottom are required than bottom grab samples to provide sufficient and valid analytical data. The aspect of maintaining the in situ vertical relationship of bottom sediment layers is often academic. Bottom sediment layers are almost always so easily resuspended that sampling confounds and commingles such layers. If, on the contrary, a compacted sedimentary bottom is sampled, or if an underlying clay layer is sampled, vertical integrity of layers can be maintained by using a core sampler having a split-spoon design. For the sorts of stream-bottom characterizations encountered in my experiences, core samplers have almost always been preferred; minimizing the loss of water-suspended fines has been important, and this technique meets that requirement.

11.3 Fish Sampling In the regulatory environment of the 1990s, toxicity-based effluent discharge limits are paramount, and the question of bioaccumulation arises logically. For example, the impact on indigenous fish species in receiving streams of both pesticide manufacturing discharges and pesticide agricultural run-off is often addressed in a fish collection and analysis campaign. The collection of fish is the field of commercial fishermen. This writer’s experience has been that in all but the most limited of studies, the services of a commercial fisherman are an invaluable asset, especially where the study is limited to one or two varieties of fish in the indigenous population. However, where the sampling campaign is conducted by the investigator, some options can affect the success and ease of the collection effort. Among the various techniques for collection of fish, electric shockers and slat boxes present advantages over other alternatives such as hoop nets, gill nets, or trot lines. These alternative techniques are likely to obtain fewer samples per unit time (e.g., trot lines) or are likely to kill the specimen well before retrieval (e.g., gill nets). (As an alternative to repetitive stream collection of specimens, caged fish can be used for biouptake determinations where this use meets the objectives of the study.)

Sampling Treuted Wasfewaterr

25 1

The method of preparation of the specimen is critical if reproducibility is required among multiple laboratories analyzing the fish tissue. If freezing the fish for preservation, shipping, and handling is necessary, then the whole fish should be frozen. A caveat should be noted here: If aluminium foil is used to wrap the fish for storage or transportation, the foil should first be washed with methylene chloride and thoroughly dried. The shiny side of the metal foil should not be in contact with the fish because this side is coated with a slip agent. Where the fillet of the fish is chosen to ascertain bioaccumulation of chemical species, the fillet should be manually removed, cubed, ground, quartered, mixed, and then split as samples for multilaboratory analysis. Contrary to some popular protocols on the subject, use of a high-speed, high-shear blender to attempt homogenization prior to sample splitting is generally unwise. This technique almost always results in physical separation of fatty oils (lipids). If this step is undertaken, then splitting this nonhomogeneous multiphase sample into portions that truly represent the sampled mass will likely be impossible. Because so many organic compounds of regulatory concern are believed to accumulate preferentially in fatty tissue, lipid separation prior to sample splitting clearly should be avoided.

11.4 Sampling of Industrial Wastewater Discharges The sampling of wastewater discharges is the subject of numerous scholarly studies, papers, monographs, and books. Information on techniques and available sampling instrumentation can be found in the U.S. Environmental Protection Agency’s (EPA’s) Handbook for Sampling and Sample Preservation of Water and Wastewater [I]. A more fundamental treatise has been published by the American Society for Testing and Materials [2]. The three major sampling techniques available include grab sampling, composite sampling, and continual sampling. There arc situations where a mixture of techniques may be needed. In practice, grab sampling is almost always manual grab sampling. Composite sampling is usually accomplished by an automatic sampler taking periodic samples and compositing them in a jar or container. A continual sampler withdraws a sample constantly from a steam and accumulates the withdrawn volume for collection a t a later time. In this writer’s experience, the most generally useful sampling technique has been composite sampling with an automatic sampler, which draws a constant sample volume at time intervals proportional to stream flow. In situations where organic compounds are those species of analytical interest, the automatic sampler commonly preferred is that which uses a peristaltic pump, tetrafluoroethylene (Teflon) tubing, and a cooled (4 OC) glass container for collection. Polyvinyl elastomer (Tygon) tubing should be avoided because of leachable phthalate plasticizers. A continuous sampler, in concept, should give the most nearly representative sample of a water or wastewater flow. This statement is especially true for those

252

J . E. Norris

automatic continuous samplers where the pumping rate is proportional to stream flow. Such samplers, however, generally require large collection containers and present a more challenging maintenance problem than the somewhat simpler automatic composite samplers. The automatic composite sampler alluded to is recommended because of its ruggedness, ample but modest size, and ease of maintenance. Where sampling time intervals are proportional to stream flow, the compositing of constant-volume increments gives a sample nearly as representative as that obtained by the continuous sampler with far less sampler upkeep required. The grab sample is the least representative among the sampling choices available and, perhaps unfortunately, the most common technique chosen. If the grab sample is a manual grab sample, then it is also the simplest and easiest to obtain. In some situations, especially those dictated by regulatory protocols, the grab sample is the only choice available. For example, this situation occurs when sampling a wastewater for volatile organic compounds (VOCs) such as halocarbons and volatile aromatic compounds. A single grab sample or a half-dozen grab samples over a 24-hour period give only "snapshots" of the quality of the sampled stream. These samples can be representative at best of the stream condition during the several seconds of sampling. Where a stream is essentially constant-flow, spatially homogeneous in composition at any time and varying in composition only gradually over an extended period of time, a modest series of grab samples may give a fair approximation of stream characteristics. Real-world industrial wastewaters rarely meet the requirements of this idealized model although river flow may approach some constant composition over short intervals of time. Ironically, the officially sanctioned determination in wastewaters of chloroform, dichloroethanes, benzene, and dichlorobenzenes hinges significantly upon the least representative sampling technique chosen, the grab sample. In addition, effluent guideline limitations for such organic compounds consist of monthly averages and daily maxima. These limits imply that the measurements truly represent stream composition. The rationale for the regulatory requirement of grab samples is that VOCs are lost into the airspace of the 4 "C thermostated sample collection bottle if a continuous or automatic compositing sampler is used. Henry's law is usually considered, but one aspect rarely addressed is the dramatic decrease in partial vapor pressure of most VOCs in aqueous solution as the temperature drops from ambient levels to 4 "C (i.e., the Law of Mass Action). This history of the sampled stream is of some importance in grasping the full significance of imputed VOC losses. Most wastewater discharges are warmer than their receiving streams. Indeed, many National Pollutant Discharge Elimination System (NPDES) compliance points measure a composite stream of warm, biologically treated effluent with warmed once-through noncontact cooling water. For example, consider a composite stream at 35 "C (95 O F ) exiting a large header and dropping 6 feet into a concrete-lined ditch basking under a summer sun at 100 "F on its way to a Parshall flume (i.e., an artificial channel or chute for a stream of water calibrated to indicate flow rate from the water level in the chute) and an NPDES outfall sampling point. Are VOC losses at 4 "C in a darkened sample

Sampling Treated Wastewaters

253

collection bottle really significant compared with the VOC losses occurring in the sampled stream? One would have to answer “no.” This situation can be considered from another perspective: When three or six grab samples in 24 hours are used to determine VOC concentrations in a wastewater, does the researcher gain more objectivity of results by minimizing assumed Henry’s law losses, or does the researcher lose more objectivity of results by taking less representative samples? This writer’s experience has been that more objectivity is gained by using automatically composited flow-proportional samples thermostated to 4 “C than by using several grab samples. This choice does not exist for regulatory compliance monitoring because grab samples are specified. However, the regulation does not address the situation of sampling upstream from a compliance point for purposes of control and characterization. Therefore, this situation is where the automatic compositing sampler can be profitably used by the investigator. Thus, the choice of sample type depends upon the intent of the sampling campaign, the regulatory proscriptions, the nature of the sampled stream, and the importance of the results of analyses of the collected sample, particularly for compliance samples.

11.5 Sampling of Surface Waters : Receiving Streams The Clean Water Act addresses the premier position of water quality standards in the regulatory scheme. A given industrial or municipal discharge may meet all defined limitations of the NPDES permit. If, however, water quality of the receiving stream is perceived as being threatened, and established water quality standards are potentially compromised by such discharges, then the discharges must be further abated, notwithstanding the best available technology. As far as the mechanics of sampling are concerned, most of the options discussed in the preceding section are, a priori, applicable to stream or river sampling. When this decision-making process is encountered, however, the peculiarities of stream sampling become manifest to the investigator. Particularly for a river that is a navigable stream, the employment of fixed samplers (e.g., continual or composite) is generally out of the question. This writer’s experience and that of many of my colleagues is that the most practical river sampling program is accomplished from a boat at known sampling locations. The sort of sample collected is almost always a manual grab sample or a series of manual grab samples composited prior to analysis. As far as choice of sampling locations, the EPA Handbook f o r Sampling and Sample Preservation of Water and Wastewater [3] addresses the various techniques for choosing appropriate locations. One technique especially appropriate for sampling rivers for chemical constituents is the spatial gradient technique. By applying this technique, the distance between points on a transect or grid in a grid pattern can be determined. This approach is as good as any, but

254

J E. NorriJ

caveats do exist. The use of this technique must be tempered with a knowledge of many factors such as effluent plume dissipation, mixing zones, segregation of wastewater discharges in the stream, and tidal effects and commercial river traffic patterns. For a valid determination of the impact of an outfall on a receiving stream, an upstream control point must be selected. This control point must be sufficiently upstream to be isolated from the effects of the discharge of effluent. For some streams and rivers, even those having considerable flow and which empty into bays, oceans, or large inland lakes, a measurable tidal effect occurs. This author has observed a river at low annual flow that was influenced so much by an incoming tide 80 river miles downstream that a wastewater discharge plume spread upstream for one-half of a mile. Indeed, for such coastal river zones, laminar flow in the river often has more dense brackish water incoming under an outgoing freshwater flow. For these reasons, the choice of an upstream control point must be carefully made. A control point located nearby has the advantage of convenience of sampling, but one several miles upstream is better insulated from any influence of the discharge. Aerial photogrupphy, including infrared techniques, provides insight into the phenomena of instream segregation of a n effluent plume, variations of mixing zones as a function of stream flow, and impacts of tidal effects. Together with surface observations, these data provide guidance in the choice of downstream sampling locations. Such data are most valuable when a modest library of aerial photographs and surface observations extending over a period of one or two years is availabe. For example, little useful information is gathered from a midriver sampling point one mile downstream from an outfall if the discharge is hugging the near bank of the receiving stream. Such anomalies are easily discerned by aerial photography and careful surface observation from a boat, etc. Against such a backdrop of knowledge of local conditions, good choices of sampling points can be made, and the water quality of the river or stream can be more objectively assessed for whatever purpose the investigator has in mind. Acknowledgements. I acknowledge the support provided by BCM Engineers Inc. as well as the efforts and contributions made by K. Fitzgerald and B. Wilson.

1 1.6 References [ I ] Hundhook,fhr Sonipling and Sample Preservution of Wutrr und WusteMuter; U S . Environmental Protection Agcncy. U.S. Government Prinling Office: Washington, D.C., 1982; EPA-600/ 4-82-029. (Addendum, 1983.) [2] Annuul Book of’ ASTM Stanrlirrds; Amcrican Society for Testing and Materials; Philadelphia. Pennsylvania, 1986; Vol. 11.01, pp, 130- 139; Standard D3370-82. [3] Hondl~ook.forSampling and Sunzple Preservation of’ Wuter und WustPwuter; U.S. Environmental Protection Agency. U.S. Govcrnment Printing Office: Washington, D.C., 1982; pp. I95 -200; EPA-60014-82-029,

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

12 Water and Wastewater Sampling for Environmental Analysis Elie M . Dick

12.1 Introduction The presence of pollutants in our water resources is no longer a concern of just the scientific community. In the last twenty years, the world has experienced a tremendous and continually increasing environmental awareness and understanding of the hazards associated with water pollution. People are demanding stricter pollution prevention and remediation plans. To be effective, these plans require monitoring programs for evaluating the physical, chemical, and biological characteristics of our waters and their impact on health, ecology, and designated uses. A significant part of the monitoring program is the collection of water and wastewater samples for testing and evaluation. Initially, the evaluation process focused on sediments, color, odor, and taste. Today, sampling professionals are interested in a wide variety of parameters including biological oxygen demand (BOD), chemical oxygen demand (COD), solids, conductivity, pH, temperature, dissolved oxygen, oil, grease, volatile organic compounds, and heavy metals. A more detailed list of parameters of interest to sampling professionals is shown in Tab. 1. It must be noted, however, that this list is incomplete and is continually increasing with time reflecting our technical ability to detect and measure pollutants and our level of environmental awareness and regulations. In some countries the list consists of dozens of parameters, while in others it is well over one hundred.

12.2 Why Sample Water? There are three primary objectives for monitoring the quality of water. First, it helps in setting realistic environmental policies. Second, it helps in developing achievable pollution prevention and remediation programs. Finally, it helps enforcement agencies in ensuring compliance with environmental regulations and permits. More specific reasons for conducting water and wastewater sampling for environmental analysis include the following: To determine the current condition of a water resource. To identify the source of a specific pollutant. To evaluable the impact of a pollutant. - To determine the effectiveness of a water pollution prevention program. - To measure the performance of a municipal or industrial treatment plant. - To spot changes and trends in a water resource. -

-

256 Tab. 1.

E. M . D k k

Some Paramcters of interest to Sampling Professionals

Acidity

Fluoride

Alkalinity

Oil and Grease

Ammonia

Organic carbon

Biochemical Oxygen Demand (BOD) Chemical Oxygen Demand (COD)

1

I

Oxygen, dissolved (probe)

pH

Chlorinated hydrocarbons

Solids, dissolved

Chlorine, Total residual

Solids, suspended

Chromium

Specific gravity

Coliform, fecal and total

Temperature

Color

Toxicity

12.3 Elements of the Sampling Plan Sampling objectives vary and so do the parameters to be evaluated. Before embarking on a sampling project, it is important for the sampling investigator to develop a written sampling plan. The plan is an important document. It helps in developing logical, workable strategies and to express them in clear, concise form. It will increase thc chances of putting together a successful program by ensuring that all important questions are asked and answered. This, in turn, ensures that the best possible strategy is developed given the available information and limited resources. Finally, the plan helps the reviewers in evaluating field activities and ensuring that all individuals involved are following the plan. Briefly, the sampling plan should cover the following items.

12.3.1 Sampling Objective It is very important to define the objective before setting up the sampling plan. It must be stated clearly, concisely, and preferably in writing. To set the objective, the sampling professional must have a good understanding of the environmental

Water and Wustewater for Environmental Analysis

251

regulations or permit requirements, if any, that he or she is to operate under. This can be achieved by contacting the appropriate authorities or consulting with experts in the field. Sometimes, the sampling objective is dictated by regulations or government permits. For example, in the U.S.A., federal and state storm water run off regulations prescribe certain sampling programs.

12.3.2 Sampling Location In general, it is assumed that the desire is to always collect samples that are truly representative of the source. Convenience, accessibility, hazards, and safety of personnel are factors that should be taken into consideration, but should not become the overriding ones. If they do, the resulting data can be erroneous and can lead to costly decisions. In some cases, sampling locations are based on permit requirements or enacted regulations. For example, in the U.S.A., the National Pollution Discharge Elimination System (NPDES) permits might specify to an industrial discharger the exact sampling location. In addition to location, the sampling point in the water or wastewater must be specified. At what depth should the sample be collected? Should it be at the bottom, in the middle, or at the surface of the water? Should the sampling be conducted in a stagnant or a turbulent area? The answers to these and other questions depend on the sampling objective. For example, if the objective is to collect a representative sample in a wastewater discharge stream, sampling should be performed where the flow is turbulent and well mixed. On the other hand, in groundwater and volatile organic compounds (VOC) monitoring, sampling should be conducted in very tranquil water.

12.3.3 Sample Types and Collection Techniques The sampling plan should specify the types of samples to be collected, sample volumes, dates, times, frequency, and the conditions under which the samples must be collected. This is extremely important and will be discussed in detail later. Briefly, however, the sampling professional should review any pertinent regulations or permit requirements and should consult the laboratory where the analyses will be performed for sample volume requirements.

12.3.4 Sampling Equipment A wide variety of sampling equipment is commercially available worldwide from a number of manufacturers. Each model is designed to perform a specific function. For example, automatic samplers are available in portable and refrigerated models,

258

E. M . Dick

with basic or advanced programming capabilities, and with or without built-in parameter monitoring capabilities. Some are designed for economy and others for long-term rugged use in very harsh environments. In addition to automatic samplers, other required equipment can include flow meters, analyzers, sample labels and packaging material, chain-of-custody forms, logbook, possibly a notebook computer, and a host of other tools dictated by the objective. For example, an NPDES permit might dictate that certain parameters be measured on-site, such as flow, pH, temperature, dissolved oxygen, and residual chlorine. In this case, the investigator must have a flow meter and analyzers.

12.3.5 Sample Containers and Sample Preservation As discussed in more details later, sample containers and sample preservation are an integral and important part of the sampling plan. Selecting the wrong bottles and preservatives can lead to inaccurate data. Equipment manufacturers and the laboratory where the analyses will be performed can be a major source of help. They can offer the correct containers and the required preservatives.

12.3.6 Sample Labeling and Shipping Once a sample is collected, each container should be properly labeled and packaged. This makes sample tracking accurate and maintains an uninterrupted chain-ofcustody. The containers must be delivered to the laboratory immediately so samples can be analyzed within the prescribed holding time. Holding times can be determined by consulting environmental regulations or by checking with the laboratory.

12.3.7 Types of Analyses The analyses to be performed must be known even before the sampling process begins. These are dictated by the sampling objectives. In the U.S., the analyses might be specified in the National Pollution Discharge Elimination System (NPDES) permit .

12.3.8 Chain-of-Custody Documentation Once sampling is completed, each container must be properly labeled and a complete written record of possession is maintained. Every time a sample or group of samples

Water and Wustewuter for Environmental AnuljJis

259

is transferred, the receiver must sign the chain-of-custody form and record the date and time. Chain-of-custody records are important documents that can be used as legal documents in case of litigation.

12.3.9 Quality Assurance and Quality Control One of the fundamental responsibilities of the sampling investigator is to establish quality assurance (QA) and quality control (QC) procedures that will ensure the reliability and validity of the collected data. These QA and QC procedures should cover all aspects of the sampling program, both in the field and the laboratory.

12.4 Types of Samples There are two basic types of water and wastewater samples: discrete samples or composite sunqples. Each type serves a particular purpose and has its own advantages and disadvantages. The reader is referred to Fig. 1 for a better understanding of the various types of samples. A discrete sample, also known as a grab sample, is an individual sample collected at a set time and deposited in its own individual container. Manual discrete samples are produced simply by dipping a bottle into the source and filling it with the desired volume of water. An analysis of a discrete sample serves as a representative of the source at the time the sample was taken. It does not, necessarily, represent the source at any other time. A discrete sample is appropriate if conditions at the source are essentially constant. If conditions are expected to change, then a series of discrete samples is more appropriate. Obviously, this will result in higher costs for sample collection and analysis. A composite sample consists of two or more smaller samples collected at different times and deposited into the same container. A composite sample can be obtained by mixing two or more discrete samples. Composite samples represent the average characteristics of the source over the period when the sub-samples were collected. Composite samples are useful in determining the average concentration and loading of pollutants during the compositing period. Clearly, the cost for laboratory analysis of a single composite sample is less than a series of discrete samples taken during the same period. It must be emphasized, however, that a composite sample yields less information and may lead to erroneous conclusions. For example, compositing discharge with high and low pH may produce a neutral composite sample. Obviously a series of discrete samples can avoid such misinformation. Depending on the objective, a variety of composite samples can be collected. A time composite sample consists of sub-samples of equal volume collected at equal time intervals and deposited in the same container. This type of sample will produce an average of the conditions at a stream, if flow is relatively constant.

E. M . Dick

260

Bottle Number

Discrete Sampling Each bottle receives one sample from one sample event.

Composite Sampling Each bottle receives a sample from multiple sample events.

1

rj 2

3

4

1 (:I 5

6

7

Discrete Sampling Multiple bottles receive a sample from one sample event.

8

Composite Sampling Multiple bottles receive samples from multiple sample events.

Sample Event Number

Fig. 1.

Types of samples.

If flow varies, a flow^ proportional composite sample will be required. This can be achieved in one of two methods. In the first method, one collects subsamples of equal volume at equal flow volume intervals. For example, collect l00mL samples for every 50,000gallons of flow. In the second method, one collects sub-samples at equal time intervals but makes sample volumes proportional to the flow volumes during the time intervals. For example, if flow volume was 10,000 gallons during the first hour and 50,000 gallons during the second hour, then the sub-sample volume will be 100 mL during the first hour and 500 mL during the second hour. A third type of composite sample is the sequential composite sample. This is obtained by combining a series of composite samples taken during shorter time or flow intervals. For example, each hour one combines 12 sub-samples collected every 5 minutes. At the end of a 24-hour period, we end up with a series of 24 hourly composite samples, which are then combined into a single composite sample. Depending on the objective, a large number of different types of composite samples can be generated by simply varying the cornpositing method.

Wuter und Wastewuter f o r Environmentul Analysis

26 1

12.5 Sampling Programs A sampling program is a set of instructions that prescribe the type or types of samples to be collected, the volume of each sample and sub-sample, and the conditions when these samples are to be taken. The conditions might be based on time, flow, or any other parameter such as pH, temperature, conductivity, or dissolved oxygen. The following are some examples of sampling programs: A program may call for collecting samples at a discharge point of an industrial manufacturing facility. A series of ten discrete 500 mL samples are to be collected each hour starting at 8:OOAM and ending at 5 : O O PM. This program provides ten different snapshots of the effluents from this facility. The above simple program does not necessarily produce an accurate idea of the quality of the effluent wastewater. This is particularly true, if effluent quality fluctuates throughout the day. In this case, the program can be modified by taking a series of ten hourly composite samples between 8 : 00 AM and 5 : 00 PM as follows: each hour, combine in a new bottle four 250 mL sub-samples taken 15 minutes apart. The two programs are similar because they provide ten hourly snapshots of the plant’s effluents. But they differ in that the second program provides an average of the pollutant loading during each hourly period. Fig. 2 provides a comparison of various sampling programs discussed.

.

Flow\\ Rate

Time

---+

Sample Volume Equal Time Intervals-EqualSample Volumes

Sample Volume Equal Time Intervals-FlowProportioned Sample Volumes

Sample Volume Unequal Time Intervals-Equal Sample Volumes

Fig. 2. Comparison of various sampling programs.

262

E. M . Dick

A third sampling program may call for collecting 500mL samples every time 100,000 gallons of wastewater are discharged. This program will produce more representative samples, if the volume of the discharge varies during the day. As such, it can help in estimating the total pollutant load in the effluent. As a last example, a program may dictate that a discrete sample be collected every time the pH goes below 5 or above 8. Such a program can help enforcement agencies in catching illegal dischargers or permit violators. It can help an industrial discharger in determining the cause of the change in pH. A major benefit of parameter or event sampling is the reduction in the number of unnecessary samples and the associated costs for analysis. Whatever the sampling program might be, the ultimate objective is to get as precise and as accurate a picture of the conditions at the monitoring point as possible. This is extremely important because the data obtained from the program can be used for setting environmental policies and regulations, for enforcement and compliance monitoring, for process control and evaluation, for revenue generation, or for program planning purposes.

12.6 Sampling Equipment In the early days, sampling objectives were simple and expectations were low. With time, however, our knowledge base and performance standards increased. This created a need for more sophisticated sampling programs. It is now recognized that, simply filling a bottle with water or Wastewater and sending it to a laboratory for analysis is not acceptable. This approach can produce insufficient or erroneous data that can lead to costly decisions. The rising need for more advanced sampling programs in the 1970s led to the development of automatic water and wastewater samplers. In the 1980s, manual sampling lost ground to automatic samplers at a very rapid rate. Sampling frequency, sample volume measurement, and compositing of samples make manual sampling labor intensive and highly unreliable. Conditional sampling based on parameters such as flow, pH, temperature, conductivity or dissolved oxygen, rainfall, or any other event are very common today. These are not only difficult to perform manually, but are almost impossible to conduct without sacrifing accuracy. With today’s low cost automatic samplers, the sampling professional can save on the labor required for manual collection and substantially increase the accuracy of the results. It is not surprising that most of the sampling in the United States and the industrialized world is conducted with automatic samplers. Automatic samplers can be classified as portable or refrigerated. Portable samplers are designed to be carried from site to site, for use in tight locations such as manholes, or at remote monitoring applications where power is not readily available. Refrigerated samplers, on the other hand, are designed for permanent installation where AC power is available.

Water and Wastewater for Environmental Anulysis

263

Fig. 3. This portable sampler can be carried from site to site and can be used in manholes.

Typically, they are installed inside buildings, but are being used increasingly in outdoor monitoring applications. Fig. 3 and 4 show some portable and refrigerated samplers.

Fig. 4. Refrigerated samplcrs are designcd for permanent applications.

2 64

E. M . Dick

-

Controller Pump

Sam le BOttfL

Sample Intake or

Fig. 5. Various components of an automatic sampler.

An automatic sampler consists of six basic components: a power source, an electronic controller, a sample intake, a sample transport line, a sample storage system, and a sample delivery or gathering system. Refer to Fig. 5 for a better understanding of the various components of an automatic sampler. Each component performs a certain function and can affect the accuracy and precision of the sampling program.

12.6.1 Power Source The power source supplies the power needed to operate the sampler. Depending on the monitoring site, the power source is AC power or a DC battery pack designed specifically for the sampler. Portable samplers are usually operated with a battery, while refrigerated units depend on AC power.

12.6.2 Electronic Controller The electronic controller is probably the most important part of the sampler. It is the brain that controls all aspects of the sampling operation. It can be programmed,

Water and Wastewater f o r Environmental Analysis

265

like a personal computer, to run almost any sampling program. The program controls the date and time of sampling, the conditions and frequency of sampling, the volume of each sample and sub-sample, and the bottle where each sample should be deposited. In addition, the controller can act as a data logger by storing the sampling program, the sampling data, and other parameters such as pH, temperature, conductivity, and dissolved oxygen. Today’s controllers are sophisticated devices that have helped advance the water and wastewater monitoring field. They can interface with other instruments, such as open channel flow meters, liquid level meters, rain gauges, and a wide range of physical and chemical parameter analyzers. They are also capable of downloading or uploading sampling programs and data to a personal computer or a printer. Enforcement agencies and industries find hard copies very useful, since they can provide proof of environmental compliance or violation of regulations. All these features increase the level of accuracy and precision in sampling and at the same time simplify data analysis and report generation. With the increased interest in storm water runoff monitoring, particularly in the United Sates, samplers were introduced to help customers meet the new regulations. With built-in software, the new samplers can automatically take discrete samples during the first 30 minutes of a storm event and composite samples afterwards. The first flush sample serves to identify pollutants that run off immediately following the rainfall. The second flow-weighted composite sample is intended for estimating pollutant loading into our water resources.

12.6.3 Sample Intake The sample intake, or strainer, is the part immersed in the water or wastewater. At this point the liquid enters the sampler and then travels through the sample transport line to storage for later retrieval and analysis. To ensure that a representative sample is collected, the sampling professional should select a sample intake large enough to allow the water to enter freely without clogging, but small enough to prevent settling of solids. Furthermore, the intake should be chemically compatible with the liquid in the stream and physically resistant to possible damage from flowing debris. Fig. 6 shows three examples of sample intakes. One is constructed of stainless steel and is used for priority pollutants. The second is made of polypropylene for general use. The third, with a small diameter, is used for low flow applications.

12.6.4 Sample Transport Line The sample transport line is usually a plastic tubing. Its function is to transport the liquid from the sample intake to the storage system. The inside diameter of the transport tubing should be large enough to prevent plugging, but small enough to

266

E. M. Dick

Fig. 6. Types of samplc intakes.

maintain sufficient flow velocity to prevent settling of solids. Kinks, twists, and sharp bends should be totally eliminated, as it might lead to plugging, low flow velocities, or high pressure drop. Length of the transport line should be as short as possible, in ordcr to minimize pressure drop and reduce the potential for cross-contamination from the previous sampling fill. To prevent deterioration, the tubing should be chemically compatible with the liquid in the stream to be sampled. It is also very important that the tubing be made of a material that will not alter the sample, chemically or physically, by leaching, absorption, or desorption. To eliminate cross-contamination between samples, the automatic sampler should be capable of rinsing the transport lines prior to collecting a new sample. Rinsing cleans the lines from any pollutants left in the tubing or that might have adhered to the tubing walls from previous samples. This is particularly important when coliecting a series of discrete samples.

12.6.5 Sample Storage Generally, the sample storage system consists of one or more bottles or bags to store the samples until they are shipped to the laboratory for analysis. A wide range of bottles is offered by sampler manufacturers to satisfy a variety of needs. Depending on the sampling program, the bottles can hold from a few milliliters to

Water and Wastewatcv f a r Environmental Analysis

267

several liters of the sample liquid. The number and size of bottles depend on the type and number of samples needed and the laboratory analyses to be performed. This can be established by checking the regulations or permit requirements to determine sampling frequency and the parameters that must be monitored. In addition, one must consult with the laboratory to determine sample volume requirements needed to perform all the required analyses. Sample volume requirements may vary from laboratory to laboratory. To maintain the integrity of the sample, care must be exercised in selecting the right sample bottles. Generally, bottles are available in glass, polypropylene, polyethylene, and Teflon. Each material offers advantages and disadvantages. The sampling professional should be concerned with chemical compatibility, leaching, absorption, desorption, and legal requirements. In the United States, the Environmental Protection Agency (EPA) has set strict guidelines for the use of sample containers. Tab. 2 lists some parameters and the required containers per U.S. EPA. For more details, the reader is referred to the U.S. EPA 40 Code of Federal Regulations (CFR) Part 136, July I , 1990. Tab. 2. Selected Parameters and Required Containers (Source: U.S. EPA 40 CFR, Part 136) -____--

____

Container Type

I 1

1

Acidity

Plastic, Glass

Alkalinity

Plastic, Glass

Ammonia

Plastic, Glass

Biochemical Oxygen Demand (BOD)

Plastic, Glass

Chemical Oxygen Demand (COD)

Plastic, Glass

Chlorinated hydrocarbons

Glass, Teflon-lined cap

Chlorine, Total residual

I

Plastic, Glass

Chromium

Plastic, Glass

Coliform, fecal and total

Plastic, Glass

Color

1

Plastic, Glass

Cyanide

Plastic, Glass

Fluoride

Plastic

Nitrate

Plastic, Glass

Oil and Grease

Glass

Organic carbon

Plastic, Glass

Oxygen, dissolved (probe)

I

Glass

Phenols

Glass

Silica

Plastic

Temperature

Plastic, Glass

Turbidity

Plastic, Glass

I I

I

268

E. M . Dick

Tab. 3. Selected Parameters and Preservation Techniques (Sourcc: U S . EPA 40 CFR, Part 136) ~ __ . _ Parameter

Preservative

Acidity

Cool to 40 c

Alkalinity

Cool to 40 c

Biochemical Oxygen Demand (BOD)

1 1

1 I 1

I

I

Cool to 4" C

Chemical Oxygen Demand (COD)

Cool to 4' C, H2SO4 to pH < 2

Chlorinated hydrocarbons

Cool to 40 c

Chlorine, Total residual

None

Coliform, fecal and total

I

Cool to40C

Color

1

Cool to 4 o c

I I

Cyanide

Cool to 4' C, NaOH to pH>12

Fluoride

None

Hydrogen ion

None

Oil and Grease

Cool to 4' C, H2SO4 or HCL to pH < 2

Organic carbon Phenols silica Turbidity

1 1 1

cool to 40 C, ~ 2 ~ or0 HCL 4 to PH < 2 cool to 40 c, ~ 2 ~ to0 p~4 c 2

Cool t o 4 o c ~~

I I

1

Cool to 40 c

To preserve the samples, the bottles should tit into a refrigcrator or into a holding base designcd to accept ice. Ideally, the sample should be quickly cooled down to a temperature of 4 "C. Tab. 3 provides a list of typical requirements for some parameters and required preservation techniques per U.S. EPA. For more details, the reader is referred to U.S. EPA 40 Code of Federal Regulations (CFR) Part 136, October 1984.

12.6.6 Sample Delivery System Thc sample delivery or gathering system provides the motive force to transfer the sample from the source to the bottle for storage. Automatic samplers offer three basic types for sample gathering. They are mec'lrunical, forced flow, and mc'tion lift. As the discussion below shows, each type has advantages and disadvantages that can effect the precision and accuracy of the sampling project: M d m z i c a l dcfivcry systems are usually built in place at the sampling site. Typically, thcy consist of scoops or cups mounted on cables or paddle wheels that

Water and Wustewater for Environmental Analysis

269

are lowered mechanically into the water to collect a sample. The main advantage of mechanical gathering systems is their high sampling lift ability, which can be as much as 200 ft. (60 m) or more. Their main drawback, however, is the large number of moving parts and exposed mechanisms that tend to easily foul and require excessive maintenance. As a result, periodic inspection is required to prevent failure and inaccurate monitoring results. In ,/arced flow delivery or gathering systems, a positive force is applied to the liquid to transfer the sample from the source, through the transport line, and finally into the sample containers. The force is usually supplied by pneumatic pressure or a submersible pump. Both of these forced flow methods have the advantage of being able to extract a sample at very low depths and lift it great distances. Like mechanical gathering systems, they are susceptible to periodic failures that can affect the quality of the sampling program. The most widely used method to extract water and wastewater samples is suction lift. In this method, vacuum is applied to transfer the liquid to the storage container. Generally, this is accomplished by using a positive displacement such as a peristaltic pump or a vacuum pump with a metering chamber. These pumps are extremely versatile, because they have minimal effect on the flow stream and can purge the transport lines by reversing pumping direction. Their main disadvantage, however, is their low lift capability, which is limited to around 29 ft. (9 rn).

12.7 Pumps and Representative Samples Pumps play an important role in delivering representative samples. The following discussion provides a brief overview of the pumps most widely used in automatic sampling. Each pump has advantages and disadvantages.

12.7.1 Peristaltic Pumps Peristaltic pumps are by far the most common type of sample delivery system used on automatic samplers. Their principle of operation is simple. A rotating roller squeezes a tubing creating a vacuum that draws the sample directly from the source into the sample bottle. Please see Fig. 7 for principle of operation. In early samplers, sample volume was determined based on time. In today’s more advanced samplers, volume is accurately measured using a liquid presence detector and a pump revolution counter. Peristaltic pumps offer three main advantages: First, the sample touches only the sample tubing which can be automatically purged or rinsed, decontaminated, or simply replaced. There are no metering chambers that can cause cross-contamination between samples. Second, peristaltic pumps produce a relatively low intake line velocity, This eliminates the potential for scouring high levels of solids at the sample intake point.

6.M . Dick

270

I-

n

Suction Line

!:t-Strainer

Fig. 7. Schematic of a peristaltic pump

Third, peristaltic pumps have an extremely simple design. This reduces the need for maintenance and almost eliminates the potential for failure. Peristaltic pumps are not without their disadvantages. They are limited to a sample lift of approximately 26 ft. (8 m). The vacuum created during operation tends to pull out some of the volatile organic compounds (VOC) in the sample.

12.7.2 Vacuum Pumps Another type of sample delivery system uses a combination of a vacuum pump and a special metering chamber. This system is widely used on automatic samplers sold in Europe. In vacuum samplers, vacuum is applied to the metering chamber which forces the liquid sample to be drawn through the sample intake into the chamber. A conductive sensor in the metering chamber senses the level of the liquid, cuts off the vacuum, and stops further liquid delivery to the chamber. At this point, the liquid in the chamber is drained to the sampling bottle below. In the 1970s vacuum samplers were popular in the North American market, because they provided accurate sample volumes. This advantage, however, was lost following the development of liquid presence detectors and pump revolution counters for use on peristaltic pumps. The main disadvantages of vacuum samplers are: Their high intake velocity tends to scour solids at the sampling point, thus, overstating solids in samples. The vacuum created by these samplers tends to pull out dissolved gases and VOC’s in samples. The sample line and the metering chamber cannot be properly rinsed. This creates the potential for cross-contamination and can result in non-representative samples. Vacuum samplers have an intricate system of moving parts and sensors making them unreliable in rough environments.

Wuter and Wustewuter for Environmental Anulysis

27 1

These factors made vacuum samplers unacceptable to enforcement agencies and industrial dischargers. Ultimately, this led to the demise of vacuum samplers in the North American market.

12.7.3 Bladder Pumps Bladder pumps have been used for collecting groundwater samples for more than two decades. However, it was not until 1993, that bladder pumps were incorporated into automatic wastewater samplers. These pumps use a bladder inside a pipe and a source of compressed gas, usually air. See Fig. 8 for the theory of operation of bladder pumps.

7

Expanded

/ bladder

Contracted bladder

\ Water

Fig. 8. The expansion and contraction of the bladder allows a bladder pump to deliver the sample liquid to the container.

212

E. M Drrh

To operate the pump, it is submerged in water and compressed gas is cycled on and off to squeeze and expand the bladder. This action on the bladder draws the water through the line and into the bottle. Bladder pumps are ideal for sampling: ( I ) Their design does not allow air to touch the sample preserving its integrity and preventing it from deterioration. (2) They are extremely gentle on the sample, making them the ultimate choice for VOC sampling. (3) They have a high sample lift capability of more than 250 ft. (76 m). Bladder pumps, however, are susceptible to damage in waters with high levels of solids. This limits their application unless they are fitted with a screen.

12.8 Advancements in Sampling 12.8.1 Volatile Organic Sampling One of the most significant advancements in sampling technology is the introduction of an automatic sampler for collecting samples containing volatile organic compounds (VOC). The new portable sampler uses a bladder pump to collect samples from a water or wastewater stream. The pump is operated by a small built-in battery-powered compressor. To eliminate cross-contamination between samples, the sampler rinses the sample lines and the bottles several times before collecting each sample. During sampling, the liquid is gently injected into the bottle through a needle in a 360" stream that eliminates air bubbles in the bottle. After filling, a Teflon valve automatically closes and seals the sample with no air exposure. The date, time, and bottle number are stored in memory for future reference. Stored data can be accessed using a personal computer to generate accurate reports and chain-of-custody documentation. Several tests were conducted to evaluate the performance of the new instrument. Comparison tests were made with manual grab sampling, peristaltic samplers, vacuum samplers, and with prepared standard solutions with known VOCs. Results indicate that the new sampler provides the closest results to the standard solutions. Fig. 9 shows a sampler capable of automatically collecting a VOC sample according to the U.S. EPA protocols.

12.8.2 Sample Volume Accuracy Another innovation is the non-contacting liquid presence detector. The new detector improves sample volume accuracy, and hence, the accuracy of composite sampling. Volume inaccuracy results from changes in the level of the liquid to be sampled.

Water and Wastewater for Environmental Analysis

273

Fig. 9. This new automatic sampler is capable of collecting a voc sample according to thc U.S.EPA protocols (photo courtesy Isco, Inc.).

The new detector uses a thin piezo film installed in a plastic housing wrapped around the sample intake tubing. During pumping, the detector senses the approach of the liquid inside the tubing and sends a signal to the electronic controller. With this information and the pump revolution detection scheme, the controller determines the exact number of pump revolutions needed to deliver the exact sample volume required. Unlike contacting detectors, the new piezo film liquid presence detector never contacts the water or wastewater to be sampled. Contact with the sample causes fouling and can contaminate the sample. Fouling means periodic inspection, excessive maintenance, false starts, and worst of all non-representative samples. Another feature of the new detector is that it does not rely on the characteristics of the liquid to be sampled, such as conductivity, pH, or other factors.

12.8.3 Refrigeration While most samplers are of the portable type, refrigerated samplers have undergone significant improvements. Several models are now available that are very rugged, can withstand the harsh outdoor environments, and are built to maintain the right temperature for sample preservation in cold o r hot outdoor applications. Fig. 10 shows a refrigerated sampler designed for outdoor use.

214

8.M . Dick

Fig. 10. A refrigerated sampler pennanently installed outdoors throughout the ycar.

12.9 Preserving Sample Integrity Every effort must be made to ensure that the collected sample represents the source at the time of collection. Often, this is a very difficult task that requires execution of a carefully devised plan.

12.9.1 Sampling Equipment The first step in this direction is to select the correct sampler. Research has established that the collection process itself can alter the integrity of the sample. Vacuum, in vacuum samplers, can pull out all the volatile organic compounds in the sample. Turbulence created by some centrifugal pumps can aerate the sample and change its pH. Metering chambers can cause cross-contamination between samples. Some tubings change the sample by leaching, absorption, or dcsorption.

Wuter und Wustervuterfor Environrnentul Anu1ysi.y

275

12.9.2 Sampling Containers The second step is to select the right sample bottles, which can be either glass or plastic. It is important to select bottles, lids, and linings that are chemically compatible with the sample liquid and would not degrade the sample such as by leaching or absorption. They should be cleaned professionally following approved procedures, based on the analyte of interest. A list of some of the required containers are shown in Tab. 2.

12.9.3 Sample Handling Once collected, every effort must be made to minimize handling of the sample. Samples, for example, should not be transferred from one bottle to another. This can cause oil and grease to stick to the walls of the containers, can cause aeration of samples, degassing of VOCs, or even loss of the sample. Obviously, this would lead to inaccurate results.

12.9.4 Sample Preservation Samples can start changing physically, chemically, and biologically almost instantaneously. Therefore, every effort should be made to preserve the sample from deterioration. Ideally the sample should be completely analyzed immediately after collection. Since this is not always possible, samples should be preserved according to the regulations or by consulting with the laboratory to be used for analysis. Sample preservation consists of refrigeration, pH adjustment, and chemical fixation. Refrigeration is ideal because it does not affect sample composition and does not interfere with any analytical methods. Quick chilling down to 4 "C suppresses microbiological activity and reduces the potential for volatilization of dissolved gases and organic substances. This can be achieved by using samplers with specially designed refrigerators, or by using ice in portable models. It is important to note that small refrigerators sold at retail outlets are not designed to withstand the harsh sampling environment and may not maintain the sample a t the required temperature of 4 "C. Sample cooling must be continued during shipment and until delivery to the laboratory for analysis. See Tab. 3 for a list of preservation techniques per U.S. EPA.

12.9.5 Sample Holding Time Even if preserved, a sample cannot be expected to remain stable for extended periods: Therefore, samples must be analyzed before any significant changes start

E. M.Dick

216 Tab. 4.

I

Selected Parameters and Maximal Holding Times (Source: U.S. EPA 40 CFR, Part 136) -~

7

Acidity

14 days

Alkalinity

14 days

Biochemical Oxygen Demand (BOD)

I I

Chlorine, Total residual

14 days

Fluoride

28 days

Hydrogen ion

Immediately

Organic carbon

28days 28 days

Silica

28 days

Turbidity

I

Ammonia ~~

[Chromium

I I

I

48 hours

I

28davs

I

28 hours

Nitrate

48 hours

Nitrogen, Kjeldahl and organic

28 days

Owpen, dissolved (probe)

Immediately

PH

I

28 days

Mercury

Temperature

I

28 days

I

Phenols

I

I

48 hours

Cyanide

Oil and Grease

I

1

6 hours

I

Color

7 days until extractionImmediately

Coliform, fecal and total

I

I

28 days

Chemical Oxygen Demand (COD)

I Chlorinated hydrocarbons

48 hours

I

Immediately

I

Immediately

to occur. The U.S. Environmental Protection Agency has published a detailed list showing maximum holding times for a large number of parameters. Tab. 4 provides a list of these parameters and maximum holding times.

12.9.6 On-Site Analysis Some parameters, simply, cannot be preserved. The mobt common are temperature, pH, dissolved oxygen, and chlorine residual. Thcse parameters must be measured in the field at the beginning of the sampling program. For even higher accuracy, parameter actuator loggers must be used. These instruments continuously monitor various parameters in water and wastewater

W a t u and Wastewater f o r Environmental AnalyAi.,

Fig. 11. A field pH meter designed to activate ii sampler (photo courtesy Isco, Inc.).

Fig. 12. A multi-parameter field probe designed to measure and log pH, temperature, dissolved oxygen, and conductivity (photo courtesy Neotronics (USA), Inc.).

277

278

E. M . Dick

streams, log the data into memory, and activate the automatic sampler to take a sample bascd on a preset program. Figs. 1 1 and 12 show some instruments designed for rugged field use. They have the capability to interface with personal computers or field printers to generate useful reports and chain-of-custody documentation.

12.10 Conclusion Water and wastewater sampling is a dynamic field. Its importance cannot be over-emphasized, for it is the foundation of most pollution prevention programs and regulations. Therefore, continuously improving our ability to collect truly representative samples should be a never-ending goal.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

13 Sampling of Groundwater for General Quality Monitoring Volker Schenk

13.1 Introduction In general, the objectives for sampling groundwater from wells and springs are quality monitoring for purposes like drinking water, irrigation and industrial use. Another aspect, especially important for groundwater, is to find the origin of pollutions and changes in natural water quality. This is also to make predictions for the future. For optimum information on groundwater quality the whole catchment area should be controlled on pollutants, to insure that dangerous increases or even pollutions are detected early enough to give time for sanitation and to find the source. For this purpose, an investigation program should cover the regions close to wells in production, to places where pollutants might be emitted, and all over the area from which groundwater might reach the wells. In addition to the regional aspects investigations should include the different layers of an aquifer because there is no mixing of groundwater flow to a mean quality. Besides this aspect there are pollutants with a higher or lower specific weight than that of water which need a more differentiated investigation. Any sampling should be done due to an investigation program, which is devised according to the objectives. These are the major factors to determine sampling sites and equipment, frequency and duration of sampling, subsequent treatment of samples, and analytical requirements. During sampling, preparation and preservation no changes should happen to all parameters under determination because faults in this state of investigation cannot be rectified any time later. All facts have to be collected in a manner which makes it possible to read them even under different aspects. Scientific work on correlations between geology, hydrology and water compounds makes it possible to read a lot of facts from analysis saving exploration costs.

13.2 Sampling of Groundwater 13.2.1 Requirements for Sampling Sites Sampling of groundwater is completely different to that of surface waters because one can reach groundwater only at a few sites. To make a groundwater observation program cost-effective, most of the existing observation wells have to be used for

2x0

V . Sclienk

sampling. This rcquircs dctailed investigations to differentiate whether a site is good enough in general, or if only a limited number of parameters can be analyzed or if the well must be replaced. Detailed work on the origin of falsifications shows, that most of them can be limited to acceptable values. Natural changes between samplings in time and from one site to another are greater than those caused by the falsification due to the well. Groundwater is not a homogeneous body with identical quality at all points. Sampling sites like wells and boreholes may alter the quality in several ways. Therefore sampling should be carried out on suitable sites and in a manner that does not alter the composition of the water. It is possible to prevent these influences by a special construction of the site and by using special sampling techniques, or by taking them into consideration. To d o this, some informations about the site are required. The first general requirement is a clear identification of the site. Of special importance are the coordinates, because only these allow the correct location on any map. Also requested are facts on the filter and its position, which is very important for a location of that sample within the aquifer. For it is really impossible to have a representative lot of the whole waterbody, one has to know the part of the aquifer being sampled. In addition, informations are needed on the groundwater level, the type and size of the sampling site, on discharge, time of pumping and other items of interest for interpretation. Special requirements to sampling from wells: Under normal conditions, water is taken from a well more or less continuously and therefore new water will be available. Time of contact to the materials of the filter, wellscreen or pipes is short and so is the contact with the free atmosphere. All this gives little opportunity for falsifications of the main components in that water. The aspect of short contacting is also valid to alterations caused by the pumps, which are installed. Sampling should be done at a special tap connected as close as possible to the pipe coming out of the well and it has to be carried out without bubbles to prcvent oxidation and changes in gas content. If sampling conditions are too poor with the pumps available, special sampling pumps should be installed. Nevertheless sufficient pumping should be done before. Sampling time is determined by monitoring pH value, conductivity and water temperature. If these values keep constant for a reasonable time according to discharge, one can be sure to have a groundwater sample not influenced by the well. It is important to ensure that no surface water can get into the well. Special precautions must be taken during the construction of the well and during sampling. -

Special requirements for sampling from groundwater observation wells: If there is no well available at the required place, special groundwater observation wclls have to be installed. This is usually done with a pipe of 100 or only 50 mm in diameter. Intervals between samplings may be very long. During this time the contact of water and well material might cause alterations of the water within the well or borehole and close to it. These changes in composition are due to various factors. Especially exhaust of gases, contact with the atmosphere, microbial activity -

Grounhwter for Quality Monitoring

28 1

and contact with the material of the well have to be mentioned. Therefore, a material neutral to most compounds such as PVC has to be used. Zinc, for instance, does not only falsify the heavy metal contents, it also can reduce other compounds such as nitrate. Special investigations may require different materials. Another point of importance is the clear division of several overlaying aquifers and other natural stratifications within one aquifer. The dividing layers must be reinstated within the borehole using special clay material or even special concrete. Filters and wellscreens have to be of the same high quality as the rest of the pipes, and even the surrounding gravel has to be a material (pure quartz) which does not alter the water quality. If this is not ensured, information must be given to the reader of the analysis. The diameter of observation wells is an important cost factor of the investigation program. Therefore it should be chosen according to the following aspects:

If the groundwater to be observed is near the surface and if a high sampling frequency is requested, the pipe should have a minimum diameter of 100 to 150 mm. - If the groundwater level is very low and sampling frequency is only once or twice a year, groundwater observation wells of small diameters (50 to 60 mm) should be chosen. The low costs of this well will offset the higher costs of each sampling, due to the small discharge of sampling pumps.

-

Where samples are required for predetermined depths in the aquifer, the specific section of the borehole or well may be isolated using packers. For truly representative samples it is better to have separate pipes to each depth in the borehole. If contaminated and noncontaminated sites have to be sampled in one program, different equipment has to be provided. This may be restricted to different hoses if the rest of the equipment cannot be contaminated because of its special construction. After every use everything has to be cleaned and dried carefully. Special requirements for sampling from springs: Apart from wells of various diameters, springs are the most common sampling sites. Usually they do not require any type of construction work. If the spring is used for the supply of drinking water, care must be taken to ensure that material of the catchment construction does not alter the water quality. Any backflow should be avoided and sampling has to be done as close to the natural outlet as possible. A facility should be provided for flow measurements. -

13.2.2 Sampling Equipment for Groundwater General requirements: The main requirement for any sampling equipment, including that used for groundwater, is that it must not alter chemical composition of the sample during contact and handling. Another requirement is ease of cleaning to avoid contamination from previous samples and, last not least, the equipment has to be -

282

V. Schoik

durable enough for use in the field. It should be simple to handle and spare parts should be available easily. Three types of samplers are in common use: bailers, suction pumps, submersible pumps.

-

-

The main representatives of these groups will be discussed. Special requirements for bailers and other equipment for spot samples: All the equipment in this group is simple but it can be used only where small volumes of water are required and where it is certified that new water will be sampled. In use are beakers and opcn-ended cylinders made of metal, plastic or glass which can be connected to a rope. Samplers of the types used for stratified sampling in rivers and lakes are also common. A special type of bailer is a plastic hose of several meters in length with a bottom valve, holding 5 to 10 liters. Connected to a motor winch it may even be used to pump out enough water to deliver a representative sample from small-diameter wells of a depth of 100 meters or more. -

Special requirements for suction pumps: All equipment belonging to this group is limited to a groundwater level of less than 8 m below surface. At levels lower than 3 m, a bottom valve is useful. The pumps can be operated manually or engine-driven. Usually, a small petrol engine is best because it can be used even at sites without electricity, the exhaust, however, may contaminate the sample. In this case it proved to be better to have an electric puinp and a portable power generator positioned in an adequate distance. Centrifugal pumps are rugged enough for field use but sand and mud cause abrasion within the pump. Spare parts must be available. ~

Special requirements for submersible pumps : Submersible pumps, especially those designed for sampling from observation wells, function in various ways. The most common type is a centrifugal pump with one or more impellers coupled closely to an electric engine. This type has proved good under many conditions and is available for various depths and discharge rates. Use is limited by the well diameter, which should not be smaller than 100 mm, and by the maximum yield of the well. A special construction can be used even in 50 mm wells. When using a submersible pump for sampling under field conditions it has proved to be best to hang it on a flexible pipe. An additional safety rope is recommended. -

13.3 Activities at the Sampling Site 13.3.1 Determinations and Preservations Several parameters of water quality are due to changes in time. The ones undergoing rapid change cannot be measured in the laboratory, they have to be determined while sampling in the field, especially

Groundcvater ,firr Quality Monitoring -

283

temperature, all gases like oxygen and free carbon dioxide, pH value.

Parameters like heavy metals need stabilization by addition of acid to avoid precipitation of metal hydroxides. Other preservations should be done according to requests of the analyzing laboratory.

13.3.2 Transport and Storage To transport a water sample from the sampling site to a laboratory it must be filled into a container. Its selection under the view of size, material, potential for cleaning and tight sealing is very important. Therefore one has to take into consideration: The chemical compounds to be analyzed may range in concentrations from submicrograms to several grams per liter. - The capacity/size of containers must be chosen according to the requests of the analyzing laboratory. - There must neither be any chemical reaction between water compounds and the material of the container nor any absorption on walls or stoppers of that container. - There must be no contamination of the sample as a result of leaching from the container or from previous samples.

-

Sample bottles should be transported with care to prevent breakage and change in chemical compounds, especially if microbial activities are expected. Hence, the containers should be carried in a dark insulated box with shock-absorbent material. The period of transport and storage should be as short as possible. For storage, samples should be cooled to 4 "C, some parameters like BOD request freezing to - 18 "C.

13.4 Sampling Programs and Contents of Analyses Sampling of groundwater is important to assist groundwater resource management where quality monitoring takes part to explore, e.g., increases of salinity, influences of polluted surface waters, or the impact of changes in landuse on groundwater quality. To do so, effective sampling programs are needed which might be very costy. One way to lower investigation costs is to split them. A resonable compromise, practised in Germany, is: -

Close to a water production plant - up to about 2 km - all observation wells should be built as very good sampling sites with a diameter of at least 125 mm. Normally, they are operated and payed by the water-works. The distance

-

-

between sampling sites must be close enough to identify severe changes in quality . Downstream of places with the risk of pollution, sampling sites have to be installed, which must consider special pollutants. These wells have to be payed for and operated by the polluter, if known. But up to now, supervision of such places is not systematic, although it becomes more and more common to have preventive investigations not only at deposits of hazardous wastes. Most pollution does not come from industries and deposits, it comes in low concentrations from many places like small industrial areas and sewage systems. Where no producer can be identified, an institution should set and run a sufficient number of quality observation wells. In case of severe pollution the producer might be found and ordered to pay the costs of wells and investigations and to take care of sanitation. All rcgional investigations, including those to identify the natural background, should be done by an institution. For sampling, observation wells with a 50 mm diameter can be used. Only if a high frequency is required, a larger well diameter might be useful to make sampling easier and faster. Experience shows, that there is no scheme for optimal distances between wells. Hence, one should plan such a net of observation wells in steps, using the experience of a first wide net to make it more dense later.

Compared to the installation cost of observation wells, costs for sampling and analyzing in an adequate frequency will soon become much higher. They must be reduced, which can be done in a way that proved suitable already. Close to drinking water wells sampling should be done monthly. If there is a very dense net of observation wells, the frequency for each well might be reduced until there is at least one monthly sample for a certain sector upstream of a well. For the rest of the catchment area this frequency may be reduced step by step down to twice a year at observation wells in very important or representative positions, and to a frequency of several years at all sampling sites available. Experience shows for general groundwater quality survey, that it is most important to have quality information over many years. In this case it does not matter if there are 12 or only 2 analyses per year because it is better to have a view on the developement of quality at 6 sites with 2 values per year instead of 1 line with more values, which requires thc same effort. From the rest of the observation wells one should try to get analyses in intervals of 3 to 5 years. Close to an identified or possible source of pollution, the frequency should be very high, at least 12 samples in the first year. A reduction step by step down to twice a year is acceptable if no pollution is found. The content of an analysis is another point of increasing costs. Experience shows, that one has to analyze all parameters needed for an ion balance, and some more to make the analysis comparable and useful for interpretation. A comparison of analyses due to these main contents will in most cases give information on severe changes and pollution if the interpreter has a sufficient knowledge of groundwater chemistry in the respective region. In this way some analysis might become more expensive than looking just for a few pollutants only, but in general investigation costs will be lowered.

Groundwater for Quulitj Monitoring

285

13.5 Interpretation Scientific interpretation of groundwater quality is based on the fact, that normal groundwater quality is influenced by numerous anthropogenic and geogenic factors. If one of these factors is changed by pollution others will change too, due to chemical reactions. For example, many polluting substances can drip out of a sewage system. Sewage means organic substances that get oxidized, producing carbon dioxide which can be measured directly or as an increase of hydrogen bicarbonate. Under the influence of organics oxygen is used up and nitrate or even sulfate are reduced. This means, water with 5 mg/L NO, found within a region with a background of 50 mg/L indicates reducing factors. Investigations on other pollutants due to the source of this oxygen demand must be conducted. A scientific interpretation of analyses can not substitute detailed investigations of all parameters, but it can reduce the efforts of investigation programs to a minimum. Pesticides, for example, could be found only in groundwater highly influenced by agriculture and under a soil with a low retardation capacity. Hence, detailed investigations on pesticides should be started only at preselected sites with both indications. In general, interpreation needs a very good knowledge on the hydrochemical background. Documentation should be provided which shows the geologic compounds of soil, aquifer and the resulting groundwater quality. In addition, the anthropogenic background, which results from a large number of small pollutions should be evaluated.

13.6 Conclusions Costs for groundwater quality monitoring may be very high. One way to reduce them is setting up an investigation program due to the objectives using all sampling sites available. Doing so, one has to take into consideration that most of these sites are not especially designed for groundwater observations. Specifications on proper sampling sites and sampling procedures are not only for correct samplings at new sites, they shall help to find out falsifications caused by the existing sites or by former sampling. All these informations must be used in a more scientific interpretation of analyses. This is another way to reduce the efforts of investigation programs as well as the reduction of contents of analysis and frequency of samplings. Experience shows, that even a limited analysis is sufficient to answer various questions. Examples are given based on programs practised in Germany.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

14 Groundwater Sampling for Metals Robert W. Puls

14.1 Introduction The collection of groundwater samples for metals, including metalloids such as arsenic and selenium, is primarily complicated by the fact that many of the target metal contaminants are also part of the immobile geologic matrix through which groundwater flows. Historically, filtration was used to distinguish between the naturally-present formation-bound metals and those present in “dissolved” form. Research has demonstrated, however, that filtration may often result in sampling artifacts which may result in non-conservative estimates of contaminant-loading [I, 21. In addition, recent research has indicated that metal contaminants may move in association with colloidal particles or as particles themselves of colloidal dimensions [3,4, 5,6]. These particles, generally less than 10 pm in size, may through size exclusion and charge repulsive effects actually move at or slightly exceed the average groundwater flow velocity in some subsurface systems. Several researchers [7, 8,9] have developed techniques to sample groundwater which include this potentially-mobile fraction of the contaminant loading in the aquifer while excluding extraneous or artifactual particles associated with the immobile matrix or well construction activities. Additional benefits observed with these techniques include: the ability to increase the resolution of groundwater samling in three-dimensional space, thereby providing better accuracy in contaminant delineation and remediation assessment; less disturbance to the subsurface sampling environment which minimizes aeration, degassing, and other chemical changes to the sample; and the generation of significantly less purge water which must be disposed of as hazardous waste. Prior techniques generally created more disturbance to the sampling zone potentially impacting sample quality and at best provided a volume-averaged number based on monitoring well dimensions, screen length and sampling technique. Where the latter was variable, so were the numbers. It is the purpose of this chapter to familiarize the reader with these newer techniques, their value in terms of sample quality and consistency, and the conditions under which they should be considered for use. The latter depends on sampling purpose or objectives and data quality requirements. While the focus will be on metals samples and sampling in routine monitoring wells, some discussions will also have relevance for sampling organic contaminants as well.

14.2 Sampling Objectives Groundwater samples are collected for various regulatory purposes including leak detection from landfills, site assessment or plume delineation at a hazardous waste

28X

R . W . Yirls

sites, compliance monitoring, and remedial performance. Questions which should be asked prior to the development of a site sampling program or the development of regulatory mandates include: accuracy and sensitivity requirements for target elements, spatial and temporal resolution needs, speciation (including dissolved and colloid-associated) of contaminants, and site-specific hydrogeochemical constraints. Attention to and resolution of these basic questions can then determine what constitutes a “representative” sample for sampling program needs. In general, representative refers to accurate sampling of the formation pore water, however, the intended location or volume to be sampled is dependent upon sampling objectives. Sampling efforts should be compatible with and further elucidate the hydrogeochemical characteristics of the site. This is essential in being able to accurately predict the ultimate fate and transport of contaminants at a given site. O u r interpretation of the data (concentration levels) from the sampling event should be done in the context of the hydrogeochemical setting and sampling point design (e.g., spatial resolution). Consistency in sampling protocols together with comprehensive documentation is essential for any sampling program. Sampling protocols should include specific performance measures, however, some flexibility is important for dealing with complex and variable subsurface settings. This chapter will have reference to all of the abovementioned sampling objectives, however, is primarily focussed on site assessment and remedial performance groundwater sampling programs. In these programs, accuracy and resolution in three-dimensional space is a primary objective. Likewise, accurate information on the physical and chemical characteristics of the subsurface system and its variability is essential for accurate interpretation of the data with regards to contaminant fate and transport.

14.3 Sampling Point Design The location of the sampling point ideally should be based on accurate field-screening to provide information on hydrogeochemical heterogeneities as well as plume delineation at hazardous waste sites. This can be accomplished with a variety of different tools and equipment ranging from hand-operated augers to large drilling rigs. Detailed information on groundwater flow velocity, direction, and horizontal and vertical variability are essential baseline data requirements. Detailed soil and geologic data are required prior to and during the installation of sampling points. This includes historical data as well as detailed soil and geologic logs. Additional details for well or sampling points logs should be collected, particularly in the screened interval, where the level of detail should be centimeters, not meters. The use of borehole geophysical techniques are also recommended in conjunction with detailed written descriptions. Collection of solids for analysis should also be routinely performed, where possible, while acquiring water samples to augment the data base on the physical-chemical varibility of the system. Better initial plume delineation

Groundwutcr Sampling ,fbr Metuls

289

will result in fewer permanent well installations and greater sampling efficiency. Geostatistical programs have been successfully used to assist in initial sampling point designs and for subsequent additions of sampling points based on accumulated data. Installation of sampling points should be done in a phased approach, based on results, interpretation of initial sampling point data, and changing groundwater monitoring objectives. A conceptual model of the site should gradually be developed and evolve as more and more data are collected and interpreted in an iterative manner. Decisions concerning monitoring well materials and dimensions are governed by the nature of the site, the contaminants of interest, and sampling objectives. There are a number of references which address the compatibility of well construction materials with site conditions and contaminants [lo, 11, 12, 13, 141. In general, there appears to be a trend toward smaller diameter monitoring wells ( 5 cm) and shorter screened intervals ( < 2 in). This latter choice is primarily due to the realization that many contaminant plumes may be relatively thin due to hydrologic and geochemical controls on plume movement. Some reasons for using longer screened intervals include sampling in unconfined aquifers with widely fluctuating water tables and to provide monitoring of contaminants less dense than water.

14.4 Monitoring Well Development The installation of a monitoring well necessarily disrupts the formation to be sampled. These disruptions can have significant impacts on the samples obtained from the well, potentially biasing results and causing the aqueous geochemistry of the formation at that point to be misrepresented. It is imperative to minimize these impacts at avery stage of the process, from drilling through installation and well development. Monitoring wells should be developed to establish an effective filter pack around the well screen and reestablish natural flow and water quality in the immediate vicinity of the monitoring well. Well development should be performed as soon as possible following well installation. It is necessary to remove the loose particulates present in the well from the construction activities and to dislodge layers or smearing by clays adjacent to the well screen that can result in decreased or altered groundwater flow. Proper well development is essential in providing low-turbidity representative groundwater quality samples. Methods that force water or air into the formation should be avoided. Some methods of well development along with their advantages and disadvantages are described in several articles and reports [lo, 15,161. Excessive turbulence during any of these techniques could damage the sand pack, resulting in channeling of waters that might confound the interpretation of sampling results. Probably the most widely accepted development technique is simply to pump the well using a pump intake that is raised and lowered (without excessive surging) throughout the entire length of the screened interval. The optimum pumping rate will vary with well-specific factors, such as hydrologic environment,

290

R. W. Pills

screen length, and casing diameter. Turbidity during development will be initially quitc high. As the water becomes clear, it would be useful to measure turbidity over time to determine when a stable value has been achieved and development can be discontinued. Many references state that the goal of monitoring well construction in general, and well development specifically, is to obtain turbidityfree samples. While it is true that artificially induced turbidity is undesirable, it must also be realized that naturally occurring or contaminant induced particles might be mobile in aquifers under certain hydrochemical conditions and capable of transporting contaminants. When these particles are naturally present, neither the development waters nor subsequent samples taken from the well will necessarily achieve a low turbidity condition. Samples should not be taken immediately following well development. A sufficient time should elapse to allow the reequilibration of hydrochemical conditions in the vicinity of the well screen.

14.5 Colloidal Transport The hydrogeochemical significance of colloidal-size particles in subsurface systems has only been realized during the past several years. This realization has resulted from both field and laboratory studies that show faster contaminant migration over greater distances and at higher concentrations than model predictions would allow. Such models typically account for interaction between the mobile aqueous and immobile solid phases and do not allow for a mobile, reactive solid phase. If such a phase is present in sufficient mass, possess high sorption reactivity, large surface area, and remain stable in suspension. it can serve as an important mechanism for contaminant transport. Colloids are particles that are sufficiently small that the surface free energy of thc particle dominates the bulk free energy. Typically, in groundwater, this includes particles with diameters between 1 and 1000 nm [4, 171. The most commonly observed mobile particles include: secondary clay minerals; hydrous iron, aluminium, and manganese oxides; dissolved and particulate organic materials. Estimates of suspended particles in several groundwater systems have exceeded 20 mg/L [4, 5, 181 and in one landfill leachate study exceeded 100 mg/L [ 191. Several mechanisms can account for the presence of suspended stable particles in groundwater, including: ( I ) dissolution or deaggregation of the soil or geologic matrix due to changes in pH, redox condsitions, or surface tension; (2) the relatively slow, natural release of particles duc to matrix dissolution and weathering of clays; ( 3 ) physical disruption of the mineral matrix by large alterations in flow conditions due to contaminant injection, groundwater withdrawal, or large rainfall infiltrations;

Groundwater Sampling f o r Metuls

29 I

(4) supersaturation of the system with respect to an inorganic species that results in the formation of an inorganic colloid; ( 5 ) macromolecules and micelles formed from humic acid molecules or surfactants; ( 6 ) the formation of emulsions or microemulsions as the result of disposal of mixed waste solvents; (7) the release and movement of viruses and bacteria. Gschwend et al. [20] at a coal ash disposal site found that suspended mobile colloids were correlated with waters enriched in CO, from infiltrating rainfall. Ryan and Gschwend [5] demonstrated that the dissolution of iron oxide cementing agents at a site in New Jersey was responsible for the mobilization of colloidal clay particles. Dunnivant et al. [21] noted an increased mobility of cadmium caused by the presence of dissolved organic carbon. A number of studies have demonstrated the reactivity of these ubiquitous groundwater particles [22 - 251. Organic carbon colloids were found to be the major factor controlling the distribution of plutonium between the solid and aqueous phases by Nelson et al. [26]. Buddemeier and Rego [27] found the activity of Mn, Co, Sb, Cs, Ce and Eu to be primarily associated with colloidal particles in samples from the Nevada Test Site. In laboratory column studies, Sandhu and Mills [24] determined that more than 90% of the chromium and arsenic present were sorbed to colloidal iron and manganese oxides. These reactive particles have also been shown to be mobile under a variety of conditions in both field studies and laboratory column experiments [3,28 - 341. Size has been shown to be a significant parameter in colloid mobility and transport. Reynolds [32] studied the transport of carboxylated polystyrene beads in laboratory sand columns and found column effluent recoveries to be lower for the larger diameter particles (45% at 910 nm) and higher for the smaller particles (70% at 100 and 280 nm). In three different field studies, Gschwend and others [3, 5,201 showed that where colloidal-facilitated transport of contaminants was observed the size range was typically between 0.1 and 2 pm. Puls and Powell [34] demonstrated a strong correlation between particle size and transport through porous media, with spherical iron oxide particles in the size range of 0.1 - 0.2 pm being most preferred. Size has also been shown to affect transport time, with larger particles generally having faster transport than smaller particles [33, 351 and, in some cases, faster than a conservative dissolved tracer such as chloride, bromide or tritiated water [34]. This is due to the principle of size exclusion, known from the column chromatography literature. The larger particles are physically excluded from passage through the smaller pore spaces in the porous media due to their size, resulting in a reduced path length relative to the dissolved solutes.

14.6 Well Turbidity Turbidity is a term used to describe the presence of suspended material in a water sample. In general, this material is “foreign” or non-representative material present as a result of disturbance of the sampling zone from well construction, development,

R. W . Puls

292

and excessive or turbulent means of well purging. The effects from these disturbances are cumulative. However, there may be varying amounts of naturally-suspended particles in pore waters depending upon particle size, groundwater flow velocity, water chemistry, impacts to the subsurface from waste disposal, and physical characteristics of the geologic formation itself as discussed above. The measurement units of turbidity are called Nephelometric Turbidity Units (NTU), previously Jackson Candle Units, and are based on kaolin as a standard particle reference material. A turbidity criteria of 5 NTUs (maximum permissible following development) has been proposed for well development [36] and criteria are under consideration as well for groundwater sampling by the U S . EPA. Several studies have demonstrated, however, that natural levels of turbidity may be significantly higher than 5 NTUs in fine textured formations or in “disturbed’ systems as a direct or indirect result of waste disposal [6, 19, 371. While it is important to properly develop wells to minimize turbidity, it must be recognized that in some cases naturally high levels may exist, depending on geology and water chemistry, particularly at hazardous waste sites where often severe changes in geochemistry can occur due to the presence of large quantities of contaminants. It is imperative, however, to use sampling methods which minimize turbidity to exclude “foreign“ materials from the samples (cause false positives), and minimize turbulence downhole which can lead to aeration or loss of dissolved gases, stripping of volatiles, changes in chemical equilibria, and potentially introduce fresh (non-representative) reactive surfaces into the samples from abrasion. Typically, excessive turbidity results from the use of bailers, even when operated carefully [38]. The use of pumps at low flow rates located within the screened interval will generally result in initially high turbidity followed by an exponential drop with time to steady-state values (Fig. 1). The initially high turbidity

Sp. Cond.(mSxlO)

0

5

10

15

0

DO(mg/txlO)

A

Turbidity(NTU/ 10)

20

Volume (1)

Fig. 1. Equilibration or water quality indicator parameters during well purging (MW1 %Elizabeth City, Grundlos Redi-Flo 2 submersible pump, portable sampling mode, 0.26 L/min flow rate).

Groundwater

Sumpling for M e t &

293

values are due to actual insertion of the sampling device and the extent of turbidity is related to the size of the device and the care with which it is lowered into place [38 - 401. The overriding concern in the development of sampling protocols should be minimization of all disturbance to the sampling point. The ideal method of sample collection would be one that simply transfers the water to be analyzed to the analytical device with no changes imparted to the aqueous sample, leaving all solid, liquid, and gaseous equilibria intact. While this is obviously impractical and indeed impossible, it should still be the goal or conceptual framework driving the development of sampling methodologies within current technological and practical constraints.

14.7 Sampling Preparation In addition to having a properly approved Quality Assurance Project Plan (which may include some of the elements below), there are certain basic elements or activities which should be part of any sampling program and include the following items at a minimum: -

-

-

Stated sampling objectives, including sampling frequency; preparation checklist including required supplies and equipment; health and safety plan; sampling log form to document all field sampling activities; sample preservation and handling procedures and guidelines; decontamination procedures; log form to document calibration of field equipment, including all in situ or field analyses; chain of custody control and records management; pre-approved sampling procedures in the form of standard operating procedures.

Calibration of field equipment is necessary in terms of overall data quality. Documentation of all sampling events is important for future reference for sampling the same well, as an aid in the interpretation of the resultant data, and to verify that acceptable and approved practices were followed. In the field, it is important to determine water levels and check for immiscible layers prior to purging and sampling the well. Equipment is readily available for making both of these determinations. Water level data are important for site hydrologic interpretations, but are also useful baseline data for setting well purging flow rates. The flow rate used should result in very little drawndown (<0.1 m) during well purging. The presence of an immiscible layer is a complication, particularly if the sampling point (screened interval) is below this layer. Electrical sensors are preferred for both of these determinations as they impart less disturbance to the underlying water column. If an immiscible layer is present above the desired sampling point, the installation of a stilling tube with a membrane capable of excluding the non-aqueous phase but able to be ruptured by the sampling pump is suggested.

294

R . W . Pu1.Y

14.8 Purging and Sampling It is generally accepted that the water in the casing is non-representative of the formation water and needs to be purged prior to the collection of groundwater sam*ples.However, the water in the screened interval may indeed be representative of the formation, depending upon well construction and site hydrogeology [8,9,41]. Wells are purged to some extent for the following reasons: the presence of the air interface at the top of the water column resulting in an oxygen concentration gradient with depth, loss ofvolatiles up the water column, leaching from or sorption to the casing or filter pack, chemical changes due to clay seals or backfill, and surface infiltration. Purging, whether using portable or dedicated systems, should be done using a pump located within the screened interval near the desired sampling loca t’ion. Placement of the pump too close to the bottom of the well will cause increased entrainment of solids which have collected in the well over time. These particles are present as a result of well development, prior purging and sampling events, and natural colloidal transport and deposition. Therefore placement of the pump in the middle or toward the top of the screened interval is suggested. Placement of the pump at the top of the water column for sampling is only recommended in unconfined aquifers, screened across the water table, when this is the desired sampling point. The following general recommendations are suggested : Use of low purge and sampling rates; isolation of the sampling zone; - monitoring of water quality indicator parameters during purging; - minimizing tubing length and maximizing tubing thickness.

-

14.8.1 Low Flow Purging Low flow rate purging refers to the velocity of the water entering the sampling device intake and imparted to the formation pore water in the immediate vicinity of thc well screen. It does not refcr to the surface flow rate, which could be throttled back using restrictors or valves. The objective again is to sample in as passive a manner as is practical and possible, taking into account your overall sampling objectives. Typically, flow rates on the order of 0.1 -0.3 L/min are used, however, this is dependent on site-specifichydrogeology. Some extremely coarse-textured formations have been successfully sampled in this manner at flow rates to 1 L/min. During purging, the drawdown in the well should be minimal, so that migration of water from above the well screen is prevented. The effectiveness of using low flow purging is intimately linked with proper screen location, length, and well construction and development techniques. The reestablishment of natural flow lines in both the vertical and horizontal directions are important for correct interpretation of the data. For high resolution sampling needs, screens less than 1 m should be used. Most of thc need

Groundwater Sampling , f . r Metals

295

40 -

B'

35 --

5 - 3

25

30 --

Et

E 20

li0

=

/

0

.

Portable Sampling

<>

DO(mg/L)

Turbidity(NTU/lO)

--

s!2 > 2 15 _2

0

DO(rng/L) O

0

l o --

Turbidity(NTU)

5 --

o n 0

5

10

15

I

20

Volume (1)

Fig. 2. Comparison of water quality indicator parameter equilibration during well purging between portable and dedicated pump installations in the same well and at the same flow rate (MW13Elizabeth City, Grundfos Redi-Flo 2 submersible pump, 0.26 L/min flow rate).

for purging has been found to be due to passing the sampling device through the overlying casing water [9] which causes mixing of these stagnant waters and the dynamic waters within the screened interval. Additionally, there is disturbance to suspended sediment collected in the bottom of the casing and the displacement of water out into the formation immediately adjacent to the well screen [39,42]. These disturbances and impacts can be avoided using dedicated sampling equipment, which precludes the need to insert the sampling device prior to purging and sampling. A downhole camera has been used to demonstrate this at several sites [38,40]. Fig. 2 shows the typical pattern of parameter equilibration using both portable and dedicated systems in the same well for dissolved oxygen and turbidity, the most sensitive water quality indicator parameters. Note that the initial values of dissolved oxygen and turbidity are substantially higher for the portable sampling setup. As the pump or flow rate is increased, turbidity will increase together with particle size [6,39]. It remains a current topic of research how fast the flow rate can be relative to the pore water velocity, before detachment of non-representative particles is initiated. This can be evaluated by changes in turbidity (or light scattering) over time and changes in the size and mineralogy of the particles which are sampled. Particles larger than several microns are not expected to be mobile [43,44].

14.8.2 Isolation of the Sampling Zone Isolation of the sampling zone may be accomplished using low-flow purging techniques. If the pump intakes are directed outward toward the well screen and in close proximity, water may be drawn in directly from the formation with little mixing of casing water or disturbance to the sampling zone. However, if the wells

296

R. W. Puls

are not constructed and developed properly, zones other than those intended may be sampled. This may also result where the geologic heterogeneities are sufficiently different within the screened interval that preferential flow lines are accessed. This is another reason to use smaller screened intervals, especially where high spatial resolution is a sampling objective. Research has demonstrated that where wells are properly constructed and developed, passive sampling techniques may be used and provide vertically discrete data over very short distances. There is either no purging or only that of the sampling tubes [9]. It was apparent from this study that the water moving through the monitoring wells was the same as or representative of the immediate formation water. Inflatable packers may also be employed, particularly in fractured rock or in long screened intervals in monitoring wells. However, the same rules apply with respect to disturbance in monitoring wells, namely the larger the sampling device the greater the downhole disturbance during emplacement. Packer assemblies or strings are usually quite large and create significant initial disturbance in standard cased monitoring wells.

14.8.3 Water Quality Indicator Parameters It is recommended that the use of water quality indicator parameters be used to determine how much water needs to be purged prior to sample collection on a well-specific basis. The equilibration of parameters such as pH, specific conductance, dissolved oxygen, oxidation-reduction potential, temperature and turbidity have been used to determine when formation water is accessed during purging. In general, the order of equilibration is pH, temperature, and specific conductance, followed by oxidation-reduction potential, dissolved oxygen and turbidity. Criteria for equilibration should be based on purge rate and equipment specifications. Several instruments are available which utilize in-line flow cells to continuously measure the above parameters. In-line devices are recommended over batch-type instruments to minimize sample handling, changes which might occur due to sample transfer or air contact, and because the measurements are required over relatively short time intervals (about 1- 5 min) for adequate stabilization evaluation. Fig. 3 shows typical data for a monitoring well purged at a rate of 0.26 L/min using a portable submersible pump. The initial turbidity reading was greater than 200 NTUs. Also shown are contaminant (Cr[VI]) concentrations during purging. Fig. 4 shows similar data using the same pump at approximately the same flow rate, but in a dedicated manner. In this case the initial turbidity was 23 NTUs. In the portable mode, stabilization occurred after about 15 L, while in the dedicated mode of sampling, stabilization occurred after only about 6 L were purged. It is important to establish well-specific stabilization criteria and then consistently follow the same methods thereafter, particularly with respect to flow rate and sampling device. Generally the time or purge volume required for parameter equilibration is independent of well depth or well volumes. Dependent variables are well diameter, sampling device, hydrogeochemistry, pump flow rate, and whether the

Groundwater Sampling for Metals

297

7-r'

=

PH

0

Redox (Vxl 0) Sp.Cond.(mSxlO)

0

5

10

15

0

DO(mg/LxlO)

A

Turbidity(NTU/lO)

4

Cr(mg/L)

20

Volume (1)

Fig. 3. Equilibration of water quality indicator parameters and contaminant concentration (chromate [Cr]) during well purging using portable sampling mode (MW 17-Elizabeth City, Grundfos Redi-Flo 2 submersible pump, 0.26 L/min flow rate).

devices are used in a portable or dedicated manner. If the sampling device is already in place (i.e., dedicated sampling systems), then the time required for equilibration is much shorter (Fig. 2). Other advantages of dedicated equipment include less purge water for waste disposal, much less decontamination of equipment, less time spent in preparation of sampling as well as time in the field, and more consistency

25

T

.

t

0

PH Redox(Vxl0)

Sp.Cond.(mSxlO) A

04 0

0

DO(mg/LxlO)

A

Turbldity(NTU)

4

TCE(mg/L)

I

5

10

15

20

Volume (1)

Fig. 4. Equilibration of water quality indicator parameters and contaminant concentration (trichloroethylene [TCE]) during well purging using dedicated pump installation (MW 17-Elizabeth City, Grundfos Redi-Flo 2 submersible pump, 0.23 L/min flow rate).

298

R. W . Pirls

in the sampling approach which probably will translate into less variability in sampling results. The use of dedicated equipment is strongly recommended at wells which will undergo routine sampling over time. If parameter equilibration criteria are too restrictive, then equilibration may never occur as water from different and adjoining zones in the formation are accessed. It should also be noted that turbidity is a very conservative parameter in terms of equilibration and its effects on contaminant concentrations are generally not significant below about 10 NTUs. Turbidity is always the last parameter to stabilize and almost always bcfore total contaminant levels have stabilized. Turbidity levels less than 10 NTUs would generally not cause significant increases in contaminant concentrations even for highly reactive contaminants and adsorptive surfaces. For example, maximum adsorption concentrations determined for arsenic on highly reactive minerals are about 0.01 gig [34,45]. A turbidity value of 10 NTUs corresponds roughly to about 4mg/L [39], or a maximum of 0.04mg/L of liter arsenic associated with the solid phase.

14.8.4 Sampling Materials Numerous articlcs and reports have been published which address the issue of sampling materials and sampling devices [46 - 481. In general, teflon-lined polyethylene tubing is appropriate for most situations. Tubing thickness should be maximized and the length minimized to decrease the time and tubing surface area which the sample contacts. Thin, excessively long tubing will make it more likely for gas diffusion [49] and temperature changes to occur. A number of different sampling devices are available including low-speed submersible pumps, bladder pumps, and gas-driven pumps. All are capable of the low flow rates, but vary as to what depth they may be used. Bailers are only recommended for sampling the water table or immiscible layers floating on the water table. Passage of a bailer through the stagnant water zone present in most wells will corrupt the sample quality and generally causes extensive disruption downhole leading to sampling artihcts and variability in sample quality. The effectiveness of bailers as sampling devices is extremely operator-dependent. Several studies have shown that loss of volatiles is common with bailers [48, 50, 511.

14.9 Filtration and Analysis Filtration is often recommended to separate formation matrix metals from dissolved contaminants [52, 531. This separation is purely artificial however, with 0.4 or 0.45 pm pore size selected as the operationally defined separation between particulate (solid) and dissolved. In contrast, the ASTM Committee on Water specifies dissolved more correctly as “that matter, exclusive of gases, which is dispersed in water to

Groundwater Sumpling for Metulr

299

give a single homogeneous liquid phase”. Further, in ASTM Designation D 3370, they recommend collection of water samples from wells without separation of particulate matter and indeed specify the inclusion of colloidal constituents in representative proportions. If the objective of groundwater sample collection is an estimate of truly dissolved concentrations, which would be important for geochemical modeling purposes, the inclusion of colloidal material less than 0.45 pm will result in incorrect values. Conversely, if the purpose of sampling is to estimate “mobile” contaminant species in solution, significant underestimations of mobility may result, due to colloidal facilitated transport by particles which are filtered out by 0.45 pm filtration. Kim et al. [54] found the majority of the concentrations of rare earth elements to be associated with colloidal species that had passed a 0.45 pm filter. Wagemann and Brunskill [55] found more than a two-fold difference in total iron and aluminium values between 0.05 and 0.45 pm filters of the same type. Some aluminium compounds were observed to pass through a 0.45 pm filter yet be retained on a 0.10 pm filter by Hem and Roberson [56]. Kennedy et al. [57] found errors of an order of magnitude or more in the determination of dissolved concentrations of aluminium, iron, manganese and titanium using 0.45 pm filtration as an operational definition for “dissolved”. During sample collection in anoxic or suboxic systems, iron oxidation and precipitation may occur prior to filtration and result in the removal from solution of previously dissolved species due to instantaneous sorption by the precipitate [18]. Puls et al. [39] found that some in-line filters will absorb contaminants (e.g., chronium) and leach other elements (e.g., potassium) even when flushed according to manufacturer’s recommendations. Filter loading and clogging of pores with fine particles may also occur, introducing filtration errors due to reductions in effective pore size. Sheldon and Sutcliffe [58] found that virtually all filters remove particles smaller than the stated pore size. Using light scattering techniques, Johnson and Wangersky [59] demonstrated that a high proportion of materials dispersed at sizes smaller than the filter pore size will interact with the filter surface. These interactions are dependent upon pore size, particle concentration, colloid surface chemistry, electrolyte concentration and composition, nature and concentration of adsorbents, chemical properties of the filter surface, and the frequency of collisions of dispersed particles with the filter surfaces. If an estimate of dissolved concentrations is a sampling objective, the use of 0.1 pm is suggested using in-line filtration in the field to minimize artifacts. High capacity in-line filters are available as barrel filters and membrane filters. Filter clogging is a problem with the latter where particle concentrations are high. For estimates of mobile contaminants (i.e., dissolved and colloidal-associated), no filtration is recommended using the sampling methodologies described above.

14.10 Summary While specific sampling objectives will dictate to some extent the actual sampling protocol at a specific site, certain consistent recommendations can be made with respect to sampling methodologies in general. The use of low-flow or “passive”

300

R. W . Puls

techniques are recommended whereby chemical and physical alterations of the sample are minimized. Submersible pumps are suggested for this purpose. To determine adequate well purging, the use of in-line continuous monitoring of water quality indicator parameters is suggested. Disclaimer. Although the research described in this article has been funded wholly or in part by the United States Environmental Protection Agency, it has not been subjected to the Agency’s peer and administrative review and therefore may not necessarily reflect the views of the Agency and no official endorsement may be inferred.

14.11 References [l] Puls, K. W., and M. J. Barcelona (1989) Filtration of Ground Water Samples for Metals Analysis. Hazardous Wuste and Huzurdous Materials 6 (4), 385 - 393. [2] Puls. R. W. and M. J. Barcelona (1989) Ground Water sumpling f o r Metals Analyses. USEPA Superfund Ground Water Issue Paper. EPA/504/4-89/001. [3] Gschwcnd. P. M. and M. D. Reynolds (1987) Monodispcrsc Ferrous Phosphate Colloids in an Anoxic Groundwater Plume, J . Contaminunt Hydrol. 1, 309- 327. [4] Buddemeier, R. W. and J. R. Hunt (1984) Transport of Colloidal Contaminants in Ground Water: Radionuclide Migration at the Nevada Test Site. Applied Geochemistry 3, 535- 548. [5] Ryan, J. N. and P. M. Gschwcnd (1990) Colloid Mobilization in Two Atlantic Coastal Plain Aquifers. Wutcr Resour. Re.?. 26, 307 - 322. [h] Kaplan, D. I., Bertsch. P. M., and D. C. Adriano (1993) Soil-Borne Mobile Colloids a s Influenced by Water Flow and Organic Carbon. Enuiron. Sci. Technol. 27 (6). 1193- 1200. [7] Ronen, D., M. Magaritz and I. Levy (1987) An In Situ Multilevel Sampler for Preventive Monitoring and Study of Hydrochemical Profiles in Aquifers. Ground Warer Monitoring Review 7 (4), 69-74. [XI Puls, R. W. (1994) A New Approach lo Purging Monitoring Wells. Groundwuter Age 28 ( 5 ) . 18-19. [Y] Powcll, R. M. and R. W. Puls (1993) Passive Sampling of Ground-Watcr Monitoring Wclls Without Purging: Multilcvel Well Chemistry and Tracer Disappearance. J . Contrrm. Hydrol. 12, 51 -77.

[lo] Scalf, M. R., J. F. McNabb, W. J. Dunlap, R. L. Cosby, and J. L. Fryberger (1981) Munuulof’ Ground- Wuter Quality Sampling Procedures, National Water Well Assoc., 93 pp. [ l l ] Gillham, R. W.. M. J. L. Robin, J. F. Barker and J. A. Cherry (1983) American Petroleum Institute. API Puhlicxition 4367, 206 pp. 1121 Barcelona, M. J.. J. P. Gibb, and R. A. Miller (1983) A Guide to the Selection of Materials for Monitoring Well Construction and Ground-Water Sampling. Illinois State Water Survey (ISWS), Champaign, IL. ZSWS Contract Report 327, 68 pp. [I31 Barcc1ona.M. J . , J . P. Gibb, J . A. HelfrichandE. E. Garske (1985) Practica/Guide,forGroundWuter Srimpling. EPA/600/2-85/ 104. 169 pp. [I41 Barcclona. M. J., J. F. Keeley, W. A . Pettyjohn, and A. Wchrmann (1987) Handhook: Ground Water. EPA/625/6-87/016, 212 pp. [IS] Kcclcy, J. F. and K. Boateng (1987) Monitoring Well Installation, Purging, and Sampling Techniques - Part 1: Conceptualizations. Ground Water 25 (3). 300-313. [I61 Aller, L., T. W. Bennett, G. Hackett, R. J. Petty, J. H. Lchr, H. Sedoris, D. M. Nielsen, and J . E. Dennc (19x9) Hundbook of’ Suggested Practices for the Design and Ins~ullution of Groimil- Wilier Monitoring Well.?. EPA/600/4-89/034, NTIS # PB90- 159807. 398 pp.

Groundwater Sampling for Metals

30 1

[I71 Ross, S. and I. Morrison (1988) Colloidal Systems and Interfaces, New York, John Wiley & Sons. [ I 81 Puls, R. W. and J. H. Eychaner (1990) Sampling Ground Wuter,forInorganics - Pumping Rate, Filtration, and Oxidution Ef f ct s. Fourth National Outdoor Action Conference on Aquifer Restoration. Ground Water Monitoring and Geophysical Methods, National Water Well Association, Dublin, OH. [I91 Gounaris, V., P. R. Anderson, and T. M. Holsen (1993) Characteristics and Environmental Significance of Colloids in Landfill Leachate. Enuiron. Sci. Technol. 27 (7), 1381 - 1387. [20] Gschwend, P. M., D. A. Backhus, J. K. MacFarlane, and A. L. Page (1990) Mobilization of Colloids in Ground Water Due to Infiltration of Water at a Coal Ash Disposal Site. J . Contam. Hydrol. 6, 307-320. I211 Dunnivant, F. M., P. M. Jardine, D. L. Taylor, and J. F. McCarthy (1992) Cotransport of Cadmium and Hexachlorobiphenyl by Dissolved Organic Carbon through Columns Containing Aquifer Material. Environ. Sci. Technol. 26 (2), 360-368. [22] Sheppard, J. C., M. J . Campbell and J. A. Kittrick (1979) Retention of Neptunium, Americium and Curium by Diffusible Soil Particles. Environ. Sci. Technol. 13 (6), 680-684. [23] Takayanagi, K. and G. T. F. Wong (1984) Organic and Colloidal Selenium in South Chesapeake Bay and Adjacent Waters. Marine Chem. 14, 141 - 148. [24] Sandhu, S. S. and G. L. Mills (1987) Kinetics and Mechanisms of the Release qf Trace Inorganic Contaminants to Ground Water,from Coal Ash Basins on the Savannah River Plant. Savannah River Ecology Lab. Aiken, SC, DOE/SR/15170-1. 1251 Means, J. C. and R. Wijayaratne (1982) Role of Natural Colloids in the Transport of Hydrophobic Pollutants. Science 215 (l9), 968 - 970. [26] Nelson, D. M., W. R. Penrose, J. 0. Karttunen and P. Mehlhaff (1985) Effects of Dissolved Organic Carbon on the Adsorption Properties of Plutonium in Natural Waters. Enuiron. Sci. Techno/. 19 (I), 129-131. [27] Buddemeier, R. W. and J. H. Rego (1986) Colloidal Rudionucfides in Groundwter. FY85 Annual Report Lawrence Livermore National Laboratory, Livermore, CA, UCAR 10062/85-1. [28] Nightingale, H. I. and W. C. Bianchi (1977) Ground Water Turbidity Resulting from Artificial Recharge. Ground Water 15 (2), 146- 152. [29] Eichholz, G. G., B. G. Wahlig, G. F. Powell and T. F. Craft (1982) Subsurface Migration of Radioactive Waste Materials by Particulate Transport. Nuclear Tech. 58, 51 1 - 519. [30] Champlin, J. B. F. and G. G. Eichholz (1976) Fixation and Remobilization of Trace Contaminants in Simulated Subsurface Aquifers. Health Physics 30: 215-219. [31] Champ, D. R., W. F. Merritt, and J. L. Young (1982) Potential for Rapid Transport of Pu in Groundwater as Demonstrated by Core Column Studies. In: Scienttfic Basis for Radioacrive Waste Munagement. Vol. 5 , Elsevier Sci. Publ., NY. [32] Reynolds, M. D. (1985) Colloid.7 in Groundwater. Masters Thesis. Mass. Inst. of Tech. Cambridge, MA. [33] Harvey, R. W., L. H. George, R. L. Smith, and D. R. LeBlanc (1989) Transport of Microspheres and Indigenous Bactcria through a Sandy Aquifer: Results of Natural- and Forced-Gradient Tracer Experiments, Environ. Sci. Technol. 23 ( I ) , 51 -56. [34] Puls, R. W. and R. M . Powell (1992) Transport of Inorganic Colloids Through Natural Aquifer Material: Implications for Contaminant Transport. Enuiron. Sci. Teclznol. 26 (3), 614- 621. [35] Enfield, C. G. and G . Bengtsson (1988) Macromolecular Transport of Hydrophobic Contaminants in Aqueous Environments. Ground Water 26 (I), 64- 70. [36] USEPA (1992) RCRA Ground- Water Monitoring: Draft Technical Guidance, EPA/530-R-93001, November, 1992. [37] Ronen, D., M. Magaritz, U. Weber, and A. J. Amid (1992) Characterization of Suspended Particles Collected in Groundwater Under Natural Gradient Flow Conditions. Water Resour., Res. 28 (9,1279- 1291. [38] Puls, R. W. and R. M. Powell (1992) Acquisition of Representative Ground Water Quality Samples for Metals. Ground Water Monitoring Review, Summer, 167 - 176. [39] Puls, R. W., D. A. Clark, B. Bledsoe, R. M. Powell, and C. J. Paul (1992) Metals in Ground Water: Sampling Artificats and Reproducibility. Hazardous Waste h Hazarduus Materids 9 (2). 149- 162.

302

R. W . Puls

[40] Kearl, P. M., N. E. Kortc, and T. A. Cronk (1992) Suggested Modifications to Ground Water Sampling Procedurcs Based on Observations from the Colloidal Borescope. Ground W u t ~ r Moriiforing Reuieic,, Spring, 155 - 160. [41] Michalski, M. (1989) Application of Temperature and Electrical Conductivity Logging i n Ground Water Monitoring. Groimd Wutc~rMonitoring Reoiew 9 (3), 112- 118. 1421 Puls, R. W., R. M. Powell, D. A. Clark, and C. J. Paul (1991) Faciliiuted Trun.sportqflnor~tinic Contrimiizunts in Groiind Water: Purr 11. Colloidul Transport. EPA/600/M-91/040, 12 pp. [43] Yao. K., M. T. Habibian and C. R. O’Melia (1971) Water and Waste Water Filtration: Concepts and Applications. Entiiron. Sci. Techno/. 5 {I I), I 105- I 1 12. [44] O’Mclia, C . R . (1980) Aquasols: The Behavior of Small Particles in Aquatic Systems. Dzvirorr. Sci. TCJCIZIIOI. 14 (9), 1052- 1060. 1451 Hingston, F. J., A. M. Posner, arid J. P. Quirk (1971) Discuss. Furaduj, Sci. 52, 334-342. [46] Niclsen, D. M. and G . L. Yeates (1985) A Comparison of Sampling Mechanisms Available for Small-Diameter Ground Water Monitoring Wells. Proceediiigs of the 5th National S)wnpo.siutiz und fixposition on A4uifi.r Restorution arid Ground Wuter Monitoring, pp. 237 - 270. izcd Ground Water Sampling Device Matrix. [47] Pohlmann, K . F. and J. W. Hc Grouiid Water Monitoring Rev [48] Yeskis. D., K. Chiu, S. Mycrs, J . Weiss, and T. Bloom (1988) A FieldStudy 0fVuriou.s Sumpling Dcoices rind Their , F f j k ~ son V o / u f i lOrgtinic ~ Coniuminunfs. Second National Outdoor Action Conference on Aquifcr Restoration, Ground-Water Monitoring and Geophysical Methods, NWWA. May 23-26, 471 -479. 1491 Ilolm. T. R., G. K. Gcorgc, and M. J. Barcelona (1988) Oxygen Transfer Through Flexible Tubing and its Effects on Ground Water Sampling Results. Ground Water Monitoring Rtwinv 8 (3), 83. [SO] Paul, C. J. and R. W. Puls (1992) Cor??parisonof Ground- Water Sampling Deviccs Bused o i l fi41rilihrrrtion ol‘ Wuter Quulity Indicator Purrimetcw. National Groundwater Sampling Symposium, Washington D.C., November, 1992. [SI] Tai, D. Y., K. S. Turner and L. A. Garcia (1991) The Use of a Standpipe to Evaluate Ground Water Samplers. Ground Wuter Moniforing Revi~iv.Winter, 125 - 132. [52] Clesceri, L. S . , A. E. Greenberg, and R. Rhodes Trussell (eds.) (1989) StundurdMethodFfor rhr E.:yoiiziiz~tioizof’ Woter and Wtrstc~twtcr 17th Ed., American Public FIealth Association, Washington D.C. [53] USEPA ( 1979) Mctliods ,for Chet17icd Anulvsis of’ Wutrr and Wustrs. EPA/600/4-491020. [54] Kim, J. I . , G . Buckau. F. Baunigartncr, H. C. Moon and D. Lux (1984) Colloid Generation and the Actinide Migration in Gorlebcn Groundwaters. In: Scientific Bu.sis,for Nuclear Wmtc Munugoizcnt. Vol. 7, Gary L. McVay (ed.), Elsevier, NY, pp. 31 -40. [SS] Wagcmann, li. and G . J. Brunskill (1975) The Effect of Filter Pore-Size on Analytical Conccnlrations of Some Trace Elements in Filtrates of Natural Water. Intern. J . Ennviron. A n d . Clietn. 4. 75 - 84. [56] Hem, J. D. and C. E. Roberson ( 1 967) Form and Stubility c?f’uluti?inium11yi”oxide cornplexes in rliliitr. .so/ztfioii. U.S. Gcol. Water Supply Pap. 1827-A, A24. (571 Kennedy, V. C. and G . W. Zellweger (1974) Filter Porc-Size Erfects on the lysis of Al, Fe, Mn, and Ti in Water. Writer Resour. Res. 10 (4). 785-789. [S8] Sheldon, R. W. and W. H. Sutcliffe, Jr. (1969) Retention of Marine Particles by Scrcens and Filters. Lirntiol. Ociwnogr. 14 (31, 441 -444. [59] Johnson, R. D. and P. J . Wangersky (1985) Seawater Filtration: Particle Flow and Impact Considerations. Litmol. Occnnogr. 30 ( 5 ) . 966-971.

C. Soils and Sediments

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

15 Representative Soil Sampling Otto Franzle

15.1 Selection of Representative Soil Samples The selection of area-specific and ecosystem-specific sets of soil specimens has emerged as one of the most complex and challenging problems facing the inauguration of ecosystem research programs, real-time monitoring of the environment and environmental specimen banks. Soils are deemed t o be of particular importance since they form essential regulative compartments of terrestrial and freshwater ecosystems, and their chemical and physical properties and the related microbial activity make them sensitive indicators for a very wide range of environmental pollutants.

15.1.1 Methodology of Statistical Design Like many other spatially differentiated phenomena soils exhibit such a variability that only a careful and systematic primary study of their particular distribution functions can ascertain that a specimen selected is representative. In this context the term “representative” (1) means reproducing faithfully the properties of sets of phenomena in terms of characteristic frequency distributions, and (2) relates to specific spatial patterns. The latter aspect merits special attention since soils, like many other things, are not discrete independent and unambiguously identifiable objects or entities; the habitual and well-known statistical procedures cannot be applied to them as a consequence. The particular problems relating to areal data such as mapping units on soil maps “concern (1) the arbitrariness involved in defining a geographical individual, (2) the effects of variation in size and shape of the individual areal units, ( 3 ) the nature and measurement of location” [l]. Difficulties encountered in separating individual areal units from a continuum like soil cover are most frequently, and at least partially, overcome by the selection of grid squares as the basic units, geographical characteristics being averaged out for each grid square. Since grid squares are all of the same shape and size their use eliminates variability in these properties and thus solves the second problem. The most common solution of the third problem, which is peculiar to geography, is to make relative location as measured by spatial continguity the dominant variable of analysis. It can be accomplished by means of special diversity analyses or regionalization procedures which are based on comprehensive geographical data matrices whose elements are

306

0. Frunzle

derived from the digital evaluation of soil and related maps, i.e., maps of geology, national land capacity, land use, and pore space of soils. The scale varies according to the size of the area investigated, i.e., normally between the 1 : 1000000 and 1: 25000 scales. The first step in selecting representative soils is the determination of their acreages, i.e., a simple frequency analysis in terms of descriptive statistics. The next step is to define the characteristic spatial patterns of soil distribution by means of neighborhood analysis. The methodology [3] basically consists in determining the individual nearest-neighborhood relationships of each grid point, i.e., the positive or negative spatial autocorrelation which is a distance-weighted measure for each point in relation to 80 neighbors. The resultant data matrix permits to define average association frequencies of all soil units of the original maps, which, in turn is the basis for a comparison of each individual grid point as defined by positive or negative autocorrelation with these average frequencies. The vectorial distance of each grid point from the corresponding soil average is a measure of similarity or representativeness. In terms of spatial structure it ensues that those soil units are most representative which differ least in their neighborhood relationships from the average association pattern of the respective soil type. The sampling sites of these representative soil units are more precisely determined by subsequent application of the above analytical techniques to large-scale maps the results of which are eventually corroborated by inspection in the field [2, 31, including larger-scale mapping and variogram-analytical verification of experimental samples taken to ultimately define the probably representative soil specimens.

15.1.2 Small-Scale Variability of European Soils in the Light of Frequency Statistics As a result of the primary digital evaluation of the base maps the data are nominally scaled in terms of descriptive statistics. Thus the distributional characteristics of the variable soil and the genetically related other variables considered may be defined in terms of frequency distributions and by means of crosstabulation procedures.

15.1.2.1 Determination of Regionally Representative Soils in Germany by Means of Crosstabulation and Neighborhood Analysis Tab. 1 shows the relative acreages of the ten most frequent soils in the Federal Republic of Germany, which corresponds to a pre-evaluation indicating a first trend but not allowing more precise statements about the regional representativeness or the location of sampling sites.

Rrpresentutive Soil Sumpling

307

Tab. 1. The Relativc Acreages of the I0 most Frequent Soils of the Federal Republic of Germany

I 2 3 4 5 6 7 8 9

Albic Luvisols in association with Dystric Planosols and Gleysols Orthic to Calcic Luvisols Dystric Cambisols, Dystric Planosols Rcndzinas, locally Chromic Luvisols Orthic and Calcic Luvisols, locally Calcaric Regosols and Planosols Dystric Carnbisols, locally Planosols Eutric Gleysols, Gleyic and Eutric Cambisols Orthic and Gleyic Podzols Dystric Cambisols, frequently in association with Rankers 10 Orthic Podzols

7.4% 7.4% 6.7% 6,4% 5.2% 4.7% 4.4% 3.9% 3.1% 3.5%

After defining the relative importance of soil types in terms of acreage contingency table analysis or crosstabulation [4] define relationships with associated variables, e.g., parent material, land use, etc. Tab. 2 summarizes the results of crosstabulation, indicating, for example, that an Orthic Luvisol on loess comprises an optimum of representative qualities. On the basis of the above statistical pre-evaluations regionally representative sampling sites are defined by means of specific spatial statistics among which nearest-neighborhood analysis is particularly useful. It is specifically designed for measuring patterns in terms of their arrangement in two or more dimensions [5]. It involves calculations of the nearest neighbor of all points and their scores defining Tab. 2. Reprcsentative Soils of the Fcderal Republic of Germany (arranged in decreasing order of acreages)

Main soil group

Rclative* acreage

Parent material

Land use

Land Drainable capa- pore volume city of rooty horizon

Orthic Luvisols

28.1%

Loess

50

medium

Cambisols

23.9%

Grain production Foresty

38

medium

Podzols

14.0%

30

high

45

high

38

high

57

medium

30

high

58 72

high medium

32

mcdium

Rcndzinas and associated soils Planosols, Gleysols p.p.

6.3%

Fluvial Fluvisols

6.2%

Dystric Lk Eutric Histosols Marine Fluvisols Chcrnozcms

4.2%

Middle Buntcr Sandstone Glacifluvial or Saalian deposits Upper Jurassic Limestone Gypsum (Keuper formation) Various fluvial deposits Peat ( > 2 m thick)

2.3%, 0.4%

Marinc silty deposits Loess

Li thosols

0.2%

Alpine carbonate rocks

8.2%

Root crop production Grain production Grain production Root crop production Grassland Grassland Root crop production Grassland

* Total 93.8%; the remaining 6.2% comprise other soil groups, settlements, rivers and lakes.

30x

0 .P r i d e

Lithosols

Others

Dystric & Eutric Histosols

t

M a r i n e Fluvisols

Fluvial Fluvisols

Highly variable soils

Podzols

Planozols

Orthic Luvisols

9

Carnbisols

v)

r

Chernozerns

r

Rendzinas

u) r

u)

0

-0 In

ln

0 C

-am Fig. 1. Soil ncighborhood matrix

-In

0 N

v) L

0,

U

.c

0

0

a

c

Representative Soil Sampling

309

how many per cent of the cases have a neighbor of the same type, which would mean a positive spatial autocorrelation, or how many per cent of the cases have a neighbor of another type, which would indicate a negative spatial autocorrelation. In Fig. 1, the soil neighborhood matrix is a summary reflection of these facts and shows how often one main soil group is associated with itself or with others, limiting the spatial relationships to the 4 most frequent neighbors of each group. Such a matrix is not necessarily symmetric since the acreage of the various soil groups differs considerably. For example, the Orthic Luvisols have the most marked neighborhood relationships to Chernozems but these, in turn, are not among the 4 most frequent neighbors of the Orthic Luvisols. The final step of spatial analysis involves searching for the areas which correspond in their actual soil patterns most closely to the average association pattern of the respective dominant soil types of Fig. 1. After determining the regionally representative sampling sites for each soil the corresponding occurrences primarily identified on the small-scale soil map have to be localized more precisely on maps of the 1 : 25000 scale and the findings finally corroborated in the field. Tab. 3 summarizes the locations of regionally representative soils in the Federal Republic of Germany. Tab. 3.

Locations of Regionally Representative Soils in the Federal Republic of Germany (arranged in decreasing order of acreages)

Soil typc

Geographical coordinates

Albic Luvisol Dystric Cambisol Orthic Podzol Rendzina Dystric Histosol Eutric Fluvisol

13" 10' E, 48" 23' N 9" 1 0 E, SO" 1 9 N 8" 13' E, 51" 54' N 12" 0 0 E, 49O 07' N 8' 09' E, 53" 07' N 9" 0 4 E, 54" 2 4 N

In terms of acreage these soils represent more than 75% of the total German soil inventory with a correspondingly wide span of pedophysical and chemical properties. On the regional level, finally, i.e., related to the Bundesland Schleswig-Holstein a combination of the following soils exhibits a maximum spatial representativity : Gleyic Luvisol, Ferric-humic Podzol, Dystric Cambisol, Histic Gleysol and Eutric Hortisol. They represent about 80% of the soil cover of Schleswig-Holstein in terms of acreage, and no less than 70% of the German soil inventory.

15.1.2.2 Selection of Representative European Soils for Sorption-Testing Purposes In the framework of chemical testing programs the determination of the adsorption/ desorption behavior of environmental chemicals in soils is particularly important [6]. With regard to the situation in the European Union (EU) it ensues from the comparatively high costs of sorption tests on the one hand and the high degree of spatial variability of European soils on the other that the selection of test soils must be thoroughly

310

0. Friinzle

oriented by the above criteria of representativeness. Furthermore, the identification of representative soils in the member states of the European Union has to make appropriate allowance for the fact that the limited number of soils best suited for testing procedures ought to represent either a maximum part of the territory of EU member states or the wide variability of relevant parameters controlling the sorption processes in soils. Ideally, a combination of either strategy should be envisaged. The procedures described in the preceding sections permitted to identify five soil types on the basis of the F A 0 Soil Map of Europe (1965) as representative in the precise geostatistical sense of the term: - The most common soils of EU member states are the Eutrir Luui.sols (Boralfs), covering 12.8% of the whole area with a distinct maximum in the northern parts. The regionally representative sampling site is located west of Caen (Normandy, France), which again stresses the outstanding importance of loess soils which display a very wide spectrum of chemical and physical soil characteristics while they developed from a largely homogeneous substrate. - The Eutric Cambisols are typical representatives of the soil cover in southern regions, reflecting 6.8% of the European Union. The corresponding sampling site is located in Sicily southeast of Palermo. - In the light of frequency statistics the Dystric Cumhisols cover nearly the same area (6.5%) as the Eutric Cambisols, but they differ in physical-chemical properties as well as in their specific distribution pattern. While the latter are concentrated mostly in costal areas of the EU’s southern parts, the Dystric Cambisols are typical of higher latitudes and altitudes. The sampling site is near Cardiff (Great Britain). - Rrnrlzinas - distributed over 6.1 %, of the area - reflect the group of moderately developed soils on calcareous material with the representative sampling site in Greece west of Korinthos. - Orthic P o d ~ d (Orthods) s have an acreage of 4.2O/0 of the EU territories occurring mostly in the northern parts. The sampling site for the ideal Podzol is located in the Federal Republic of Germany near Lauenburg (Schleswig-Holstein). The rcsults of the spatial analysis were subsequently made subject to a two-step corroborating examination by means of larger-scale soil maps and direct visual inspection in the field, including a sampling campaign and subsequent comprehensive laboratory experiments. They furnished ample and conclusive evidence that the soils selected covered the wide range of sorption properties which have to be taken into account for comparative testing purposes [7, 81.

15.1.3 Large-Scale Soil Variability in the Light of Variogram Analysis Clearly, the quality of soil maps constitues the crucial point in the application of this type of diversity, analysis for preselection purposes. Also official surveying instructions for larger-scale soil maps may dismiss the subject of soil variability by

Rep resent u t ive Soil Sumpling

311

stating that one auger sample for depth and horizon determinations for 50 or even 200 running meters gave a reasonable estimate of soil properties and boundaries [9]. Reynolds [lo] has shown, however, that to estimate soil depth, pH, moisture and organic matter populations with an accuracy of 1 % might require 10-689 individuals and of 0.2% 196- 17227 individuals. These figures show that the determination of representative soil propertytopography relationships requires a considerable number of random samples in order to comply with the demands of frequency statistics. Even if Reynolds' [l 11 assumption is true that size of area may only influence soil variability within certain broad variability classes, so that the data presented here probably relate to areas of at least 1000m2 in size, the quest for a more practical method than random sampling remains imperative. Regionalized variables appear to adequately characterize the spatially distributed and structured phenomena under consideration, and consequently variogram analysis is the appropriate method.

15.1.3.1 Variogram Analysis Values of a measured variable, e.g., a diagnostic soil property, are usually punctiform but are to be indicative of spatial interrelations. Therefore, a random sample has to be considered as the measured value of an assumed distribution function Y (x) of a characteristic two- or three-dimensional vector x. Customary mathematical functions are insufficient to give an adequate representation of a regionalized variable because of its -

high degree of complexity, high degree of not infrequently small-scale variability, various correlations between neighboring points.

A useful statistical method would therefore have to inform about the following problems: (1) Is it possible to decide upon the existence of a spatial distribution function on the basis of the available random samples? (2) If such a function exists: How is a difference vector related to the mean variance of all vectors x i , y i for which x i - y i = k? (3) To which (spatial) extent is a random sample representative within the limits of the imputed distribution function'? The most powerful statistical tool available to this end is variogram analysis [12,13]. The Mathematical Concept

A regionalized variable can be considered as the realization of a random function Y. Usually, this function is assumed to be stationary; this implies: (1) The expectation (rn) of Y at any point x is constant and independent of x. E [ Y ( X ) ]= m

(1)

0. Frunzie

312

(2) The covariance function of any pair of points x and x on the vector h and is independent of x . E[Y(x)Y ( x

+ h)] - m’ = K ( h )

+ h depends exclusively (2)

In many cases merely the increments of the functions are supposed to be stationary. The intrinsic hypothesis for a vector h concerning expected value and variance is:

+ h) - Y ( x ) ] = 0 Var [ Y ( x + h) Y ( x ) ] = 2y(h) E [ Y (x

-

(3) (4)

The so-called semi-variogram (further on designated as variogram), i.e., the function of the vector h is defined as: y(h) = 1/2 Var [ Y ( x

+ h)

-

Y (x)]

(5)

From Equs. (3) and (4) ensues that y(h) = 1/2 E [ Y ( x

+ h) - Y(x)]’

(6)

In case of discontinuous data the expected value (h) can be estimated by the formula : 1 ” (h) = [ Y ( X i h) - Y (Xi)]’ (7) 2n i = l

c

+

(n = number of pairs of points). The points are situated in either a one-, two- or three-dimensional space. In a two-dimensional space, as dealt with here, the coordinates h , and h, determine the vector h. Hence, the variance of measured values is dependent on the distance and the direction of the difference vector h. In practice the variogram is usually computed for 4 main axes (Fig. 2) in order to account for possible directional effects. Distinctions of the range in different directions (anisotropy) then enable a more detailed interpretation. A mean variogram that is independent of such directional effects is usually computed on the basis of the four directional variograms. A variogram is geometrically characterized by two boundary criteria which are called “sill” and “range” (Fig. 3). The latter denotes the maximum extent of influence, while sill is the analogous limiting value of influence as measured on the ordinate.

K’ 2

Fig. 2.

Main variogram axes.

Rep resen ta t ive Soil Sampling

313

I

Fig. 3. Sill and range of a variogram.

-range-

The Interpretation of the Variogram The Zone of Influence It follows from the foregoing that a variogram does not necessarily have a maximum or a level of stabilization. Due to this criterion two basic types of variograms are distinguished (Fig. 4). In Fig. 4a the maximum range is reached when the correlation between Y (x)and Y(x + h) becomes nil.

+ h) - Y(x)] = 1/2 [Var (Y(x + h)) + Var ( Y ( X ) ) ]

y(h) = 1/2 Var [Y(x

-

2c2 2

(8.1) (8.2) (8.3)

(T2

In Fig, 4 b the zone of influence extends beyond the area examined. Mathematical Models of the Curve

In order to best fit the curve of the variogram under construction to the sequence of points primarily obtained as a result of the computation of the variance function, various mathematical models are used, the most important of which are briefly described. (I) Power functions: y(h) = C IhJ” with 0 < 1 < 2

(9)

y(h) = C Ihl is the special case of a linear model

bounded

unbounded

Fig. 4.a, b. Bounded (a) and unbounded variogram (b).

3 I4

0.Friinzle

(2) Spherical model:

(3) Exponential model: y(h) =

c

1-

u

z 1/3 of range

(4) Gaussian model:

Often, for a more exact fitting the models must be combined. The variance of the sample is approximately the same as the value of the sill as far as bounded variograms are concerned. The model of the curve is primarily important for the determination of the “nugget-effect”, a phenomenon that has to be dealt with when the curve’s behavior near the origin is considered. The Curve’s Behavior Near the Origin - A

parabolic shape shows a high degree of continuity of the regionalized variable. It is differentiable. - A linear shape shows continuity “in average” according to Matheron [12]. - The curve does not intersect the abscissa at the origin. The “nugget-effect” reveals great irregularity and can be caused fundamentally by either an extremely discontinuous distribution in the immediate neighborhood of the sample taken or by observational errors. Supplementary information is consequently needed. - A straight line parallel to the abscissa indicates that there is no correlation between any points Y (x h) and Y ( x )whatever their distance might be. The sample does not show any spatial structure.

+

Anisotropies Distributions characterized by different variabilities in different directions are reflectcd in the resulting variogram and called anisotropy. - Gcometrical or affine anisotropy exists whenever elliptical zoncs of influence can be deduced in a two-dimensional space. - In a three-dimensional space variations might also appear in the vertical direction; this is the case of stratified anisotropy. Other Structures Some variograms display substructures as regards the limits of their maximum ranges. Typical ones are (Figs. 5 - 7):

315

Representative Soil Sampling

(1) nestcd structures

h

Fig. 5. Nested variogram. a1

a2

a3

(2) periodical structures

Fig. 6. Periodical variogram.

(3) “hole-effect’’ structures

Fig. 7. “Hole-effect’’ variogram.

h

15.1.3.2 Model Applications of Variogram Analysis Two examples may finally illustrate the preceding theoretical considerations. The first is based on the spatial analysis of the 1975 data of the immission monitoring network of the Cologne industrial belt. The variogram shows the SO, distribution in the area north of the city, which is characterized by prominent industrial sites with remarkably high SO, emission rates at Merkenich, Wiesdorf and Dormagen. The network consists of a 1 km2 grid for which a mean value is derived from the measurements at the four grid points (Fig. 8).

0. Frunzle

316

Variogram o f SO, pollution (Cologne)

Directions 3

0.25E-03

0.20E-03

3 Mean

0.1 5 E- 0 3

0.10E-03

0.50E-04

\

ly;;

\

i;

\ \

\

‘..2 I

0

f+OO

0.80E+O.?

0.16E+0.2 krn

Fig. 8. Variogram of SO, pollution in the northern part of the Cologne industrial belt.

The variogram of the SO, distribution shows the following characteristics: -

two sills, directional variation, Gaussian model, no “nugget-effect”.

The sills at distances of 4 km and 13 km can be explained by two nested structures of which the inner one, with SO, immissions above average, covers the neighborhood of the industrial plants. Consequently, the statistically appropriate maximum distance of SO, sampling is 3 -4 km for the highly industrial core areas and 12- 13 km for the adjacent peripheral parts of the industrial belt where immission concentrations are distinctly less. Furthermore, the variogram shows that the SO, distribution varies in different directions. The particularly marked immissions in directions 3 and 2 (N-S and NE - SW) are accounted for by the predominant winds which are canalized by the Rhine-valley. On the contrary only slight variations with growing distance can be noted in direction 1 (E - W). The second example is derived from comprehensive soil investigations aiming at a detailed map of the anthropogenically induced PO:- distribution in the A-horizons

Representative Soil Sampling

317

of soils in Schleswig-Holstein. The variograms in Fig. 9 are based on the following data [14]: 1104 10 m 158.0 ppm P 65.8 11.0 ppm P 570.0 ppm P SW-NE S -N SE -NW W -E

Number of samples Distance Mean Standard deviation Minimum Maximum Direction 1 Direction 2 Direction 3 Direction 4 _____~

____

Variogram

sill

range

nugget effect?

1 2 3 4 5 (mean)

3680 4360 4800 5900 6240

100 m 160 m 240 m 420 m 420 m

Yes Yes Yes Yes Yes

bounded?

The relatively high “nugget effect” corresponding to a variance of 2800 is due to discontinuities in the immediate vicinity of the grid points sampled. Yet all variograms have well-defined sills and ranges much above the grid square dimen-

7000

4

Fig. 9. Variograms of PO:distribution in soils near Bosau, Schleswig-Holstein, F.R.G. (after [14]). Distance (m)

. . . . . .measuring points Fig. 10. M a p of phosphorus contents in soils near Bosau, Schlcswig-Holstein, F.R.G. (after [ 141).

sions. Hence the construction of a detailed P map is possible on the basis of the 10 m grid, and even distances up to at least 50 m (i.e. half the lowest range of the above variograms) would not essentially diminish the precision because metric interpolations between individual grid values remain valid. Maps with a grid basis tested by variogram analysis (which in turn requires a minimum of 40 - 50 points evaluated) are indispensable for the planimetric deduction of valid areal mean values. As such or as elements of more comprehensive areal means they constitute regionally representative values. In view of this it may suffice to say that spatially oriented extrapolations of material and energetic fluxes from punctiform data are not only of lesser quality but simply erroneous if the appropriate dimensional structure of the underlying grid has not been ascertained by means of variogram analysis. In the present case it served prospective archeological purposes by developing a soil-phosphate map whose isoplcth structure was to be indicative of the outlines of an abandoned medieval Slavic village. (In the meantime, excavations have corroborated in detail the pertinent deductions.) In view of the above marked “nugget effect”, however, a far higher number of grid points or measurements, respectively, were necessary to derive reliable isopleth maps with a resolution appropriate for the distinction of individual houses, huts and stables.

Representative Soil Sumpling

319

15.2 Conclusions An analysis of relevant literature (cf., e.g., the comprehensive review by Reynolds 1975 [ I 11) shows that studies relating the magnitude of soil or rock properties to topographical parameters like slope angle or aspect frequently neglect to indicate the limits of their statistical populations, and also fail to assess the degree of variability of the often undefined population. (An analogous statement would apply to numerous studies on vegetation or vegetation-soil relationship, or in the realm of geozoology.) It should be clear that entirely spurious relationships can result if samples consist only of a few individuals and where sample means obtained are not representative of population mean values. Therefore, the selection of soil samples for environmentally relevant analytical purposes should be based on the following five-level approach. (1. I ) Frequency analysis, (1.2) Neighborhood analysis, (1.3) Definition of representative structural units on the basis of small-scale soil and related maps, satellite images, etc. (2.1 -2.3) idem, on the basis of large-scale maps, stereo couples, etc. (3.1 - 3.4) Visual inspection, high-resolution mapping, sampling, chemical analysis. (4) Variogram analysis of samples (5) Definition of representative samples with regard to specific soil properties or test purposes, respectively.

Both the importance and costs of many environmental assessment projects urgently recommend such a sequential approach in order to obtain spatially valid data. These, however, are the essential prerequisite for pertinent extrapolations, in particular in the wide fields of both pure and applied ecology, ecotoxicology or environmental chemistry.

15.3 Summary Defining representative soil samples with regard to specific test or banking purposes requires a composite selection methodology in terms of both frequency and spatial statistics. The first step in selecting soils for test purposes is the determination of their acreage, i.e., a simple frequency analysis based on the digital evaluation of soil maps and related areal information. The next step is to define the characteristic spatial patterns of soil distribution by means of neighborhood analysis. The methodology basically consists of determining the individual nearest-neighborhood relationships of each grid point resulting from the preceding digitizing procedure. The resultant data matrix allows the definition of average association frequencies of soils, and the vectorial distance of each grid point from the corresponding soil average is a measure of similarity or representativeness, respectively. The sampling sites of these representative soil units are more precisely determined by subsequent

320

0. Frlinzle

application of the same analytical techniques to larger-scale maps the results of which are eventually corroborated by visual inspection and mapping in the field. Within a sampling plot thus defined variogram analysis should finally be applied to adequately characterize the large-scale variability of soil properties on the pedon and polypedon levels. The above methodology is exemplified on the continental (EU Member States), national (Germany) and regional levels, and the resultant locations of representative soils for analytical purposes are given. Acknowledgements. I am particularly indebted to the Federal Environmental Agency (Umweltbundesamt)/Bundesministeriumdes Innern and Bundesministerium fur Umwelt, Naturschutz und Reaktorsicherheit of the Federal Republic of Germany for valuable grants which permitted the implementation of the comprehensive research schemes which the present paper is based upon. Mrs. Weller prepared the manuscript.

15.4 References [I] Mather, P. M. (1972) in: Spatial Anulysis in Geornorp/zology, Chorley, R. J. (ed.). Mcthuen, London, pp. 305 - 322. [2] Frlnzlc, 0. (1984) in: I.:nvironrnentul Specimen Runking and Monitoring as Rduted to Bunking, Lewis, A,, Stein, N. (eds.). Proceeding of the Intern. Workshop, Saarbrucken, May 10- 15, 1982. M. Nijhoff, Boston, The Hague, Dordrecht, Lancaster, pp. 164- 179. [3] Frlnzle, O., Kuhnt, G . (1983) Regional reprasentative Auswahl der Boden fur cine Umweltprobenbank - Exemplarische Untersuchung am Beispiel der Bundesrcpublik Deutschland. Forschungsberichi 106 05 028 im Umweltforschungsplan des Bundesministers des Innern/Umweltbundesamtes. Berlin, pp. 74. [4] Nie, N . H., Hull, C. H., Jcnkins, J. G., Steinbrenner, K., Bent, D. H.(1975) SPSS Statistical Puckugefor the Social Sciences. McGraw-Hill, New York. [5] Ebdon, D. ( I 977) Statistics in Geogrphy. A Practical Approach. Blackwell, Oxford. [6] OECD (Organisation for Economic Co-operation and Development) (1981) U E C D Test Guideline 106 A . Paris. [7] Brummer. G., Frlnzle, O., Kuhnt, G., Kukowski, N.,Vetter, L. (1987) Fortschreibung der OECD-Prufrichtlinie ,,Adsorption/Desorption" im Hinblick auf die Ubernahme in Anhang V der EG-Richtlinie 79/83 1 : Auswahl reprasentativer Boden im EG-Bereich und Abstufung der Testkonzeption nach Aussagckraft und Kosten. For.sr,hungshericht 106 02 045 im Umweltforschungsplan des Bundesministers fur Umwelt, Naturschutz und Reaktorsicherheit/Umweltbundesamtes. Berlin, pp. 197. [8] Vetter, L. (1989) Evaluierung und Entwicklung statistischer Verfahren zur Auswahl von reprlsentativen Untersuchungsobjekten fur iikotoxikologische Problemstellungen. Doctorul Thcsis, Kiel. [9] Arbeitsgemcinschaft Bodenkunde der geologischen Landcsiimter und der Bundesanstalt fur Bodenforschung (1 97 1) Kurtierunlritung, Anleirung und Richrlinim zur Herstellung der Bodenkarte 1 :25000. Hannover. [lo] Reynolds, S. G. (1971) A study of the influence of topography on certain soil properties with special reference to soil property variability. Unpub. Ph. D. ThcJsis. University of Bristol. [ I I] Reynolds, S. G. (1975) Soil property variability in slope studies: suggested sampling schemcs and typical required sample sizes. Z . Geomorph. N.F. 19, 191 -208. [I21 Matheron, G. (1963) Principles of geostatistics. Economic Geology 58, 1246- 1266. [ 131 Delfiner, P. (1975) Geosruti.sticd Estimation of H.ydrocurbon Reserves. Fontaineblcau. [I41 Zolitz, R. (1980) Bodenphosphat als Siedlungsindikator. OFFA-~rgu'nzungsrPi/zP5, Neumunster, 91 pp.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

16 Problems and Results in the Development of International Standards for Sampling and Pretreatment of Soils Andreas Paetz and Gerd CroJmann

16.1 Introduction In 1986 the new Technical Committee 190, Soil quality, was founded in the Hague, The Netherlands, within the International Organization for Standardization. The scope of this technical committee (TC) reads: Standardization in the field of soil quality, including protection and classification, definition of terms, sampling of soils, measurement and reporting of soil characteristics. Exluded are -

limits of acceptability for soil pollution civil engineering aspects (as dealt with by ISO/TC 182, Geotechnics).

Sampling of soil became one of the top-priority projects of ISO/TC 190 from the beginning and three work items were set up to meet this project: (1) sampling strategy; (2) sampling techniques; (3) preservation of samples.

There was a fundamental discussion in the beginning whether or not preparation of soil samples should be part of the work of the subcommittee on sampling. It was then decided to install a separate working group on standardization of the preparation of soil samples which should work in close cooperation with working groups covering determination of chemical substances and the subcommittee on sampling. The range of sampling aspects that have to be considered by ISO/TC 190, Soil quality, includes the sampling of soil and related material for chemical analysis, both on inorganic and organic parameters, for biological examinations, and for determination of physical soil properties. The task of sampling as it is understood by ISO/TC 190 is not restricted to purely obtaining soil in the field but also includes the selection of appropriate spots in the field where in sifu measurements may be carried out as a part of site investigation programs. From the above it is obvious that sampling of soil for the determination of trace elements is just one task out of a number of very different attitudes to sampling. Within the standardization work of I S 0 it is usual to develop an International Standard by the harmonization of existing national standards. The first stage of

322

A . Yuetz

iitiil

G C‘ro~r~rcurrr

work, therefore, was to review the national standards of a t least those countries actively participating in the work of ISO/TC 190. It turned out that the number of national standards was rathcr sinall and, that although the technical content was similar in many ways, it did not meet all objectives of sampling as identified by the working group. Existing national standards did cover classical sampling procedures for argricultural purposes and only a few focussed on contaminated soil [ I , 2, 31. During the following stagc other guidelines, scientific papers, text books, and information on sampling were collected and taken into consideration. Based on this information and following contact with other organizations and institutions six different parts of an International Standard on soil sampling were identified and along with these an International Standard dealing with a vocabulary on soil sampling (see Tab. I ) . Tab. 1. International Standardson Sampling of Soils in Preparation by ISO/TC 190,Soil Quality

I so

Title of Interntionnl Standard

Current stage of preparation (March 1994)

10381-1

Soilquality - Sampling - Part 1 : Guidanceon the dcsign of sampling programs

Committee draft

10381-2

Soil quality - Sampling - Part 2: Guidance on sampling techniques

Committee draft

10381-3

Soil quality safety

10381-4

Soilquality - Sampling - Part 4:Guidancconthe proccdure for the investigation of natural, near natural and cultivated sites

Committee draft

1038 1-5

Soil quality - Sampling - Part 5 : Guidance on the procedure for thc invcstigation on soil contamination of urban and industrial sites

Working document

10381-6

Soilquality - Sampling - Part 6 : Guidanceon the collection, handling and storage ol‘soil for the assessment ofaerobic microbial processes in the laboratory

International Standard

1 1 074-2

Soil quality - Vocabulary - Part 2: Terms and definitions relating to sampling

Committee draft (in preparation)

~

Sampling

-

Part 3: Guidance on

Committee draft

16.2 What is Soil? ISO/TC 190 has defined soil as: “The upper layer of the Earth’s crust composed of mineral particles, organic matter, water, air, and organisms” [4]. This is the traditional way of defining soil as has been done in the past by soil science. In the case of soil protection ISO/TC 190 has agreed that attention must

Internutional Standurds for Sampling

323

also be paid to a wide variety of other materials which are either applied to soils to support certain soil functions, e.g., soil conditioners and growing media, or which may be deposited and possibly cause hazards to soil, man, and the environment. Resulting from this definition soil samples are prone to large variations due to the heterogeneity of the material. This can be exacerbated by inappropriate soil sampling procedures which reduce the value of soil samples for predicting concentrations of e.g., heavy metals in soil.

16.3 Soil Sampling Objectives ISO/TC 190 identified four general objectives for sampling:

( I ) sampling for determination of general soil quality, e.g., investigation of nutrient demands of agriculturally used areas; (2) sampling for preparation of soil maps, e.g., for general description of soil, land appraisal, soil monitoring sites; ( 3 ) sampling to support legal or regulatory action, e.g., to establish base-line conditions prior to activities which might affect the composition or quality of soil or in response to the input of an undesirable material which may be from a point or a diffuse source; (4) sampling for hazard and risk assessment, e.g., to identify hazards associated with soil contaminations. This kind of sampling may have to comply with legal or regulatory requirements. Depending on the principal objective(s) it will be necessary to determine for the body of soil or part thereof -

the nature, concentrations, and distribution of naturally occurring substances as well as of extraneous substances, e.g., contaminants of anthrophogenic origin.

For site investigations the physical properties and variations and the presence and distribution of biological species may also be of interest.

16.4 Requirements on Sampling An inventory was made among experts to define the requirements on sampling and samples obtained in a sampling project. Of paramount importance is the representativity of the sample for the whole body under investigation. It is obvious that several factors affect representativeness, the most influencing one is the degree of preliminary information on the site to be sampled. In many cases, soil maps or geological maps are available and, depending on the scale of these maps, subdivision and description of different soil types exposed in the field may be very detailed. In addition, soil profiles may be given in the map

324

A . Puetz and G . Cn@wuzn

or accompanying guiding information. In such a case selection of target areas for sampling is more simple. Where no such maps or similar information are available this basic information has to be prepared to some extent, otherwise all information received by analysis of samples obtained at unknown land is of little or no value. The sampling strategy, e.g., selection of sampling pattern, choice of equipment, transport and storage of samples must be so as to ensure that samples taken meet the aim of the investigation. For example, a sampling pattern commonly used in agricultural soil investigations may not be applicable for investigations regarding the protection of soil. The sampling procedure should be reliable, accurate and cost-effective. Through the various discussions it seems that reliability and accuracy on the one hand and cost-effectiveness on the other hand are contradictory. The statistical approach is to first determine the necessary number of samples or sampling points to properly define a sampling area by means of preliminary sampling and testing [5, 61. Although this is correct in general and fits well to recent developments in improving quality management in sampling it very often cannot be connected with financial resources available to customers of sampling programs. This points directly to the need for better coordination between the sampling stage, the preparatory stage, and the analytical stage of a soil investigation program. Responsibilities must be clear from the very beginning of the program to avoid misunderstandings between these three stages of work. First, and most important, the customer must definitely state the aim of the investigation, including all parameters to be examined, the time and financial sources available for the program. For the further design of the sampling program it is a basic requirement to know about all chemical analyses that will be carried out on the samples. Only if this is ensured the customers may expect a soil sample appropriately obtained in the field and transported to the laboratory at specified conditions. Having arrived at the laboratory and usually before the analysis is started some kind of sample pretreatment has to be done. Here, it is essential to distinguish between samples to be analyzed for inorganic compounds or soil characteristics such as pH, electrical conductivity, and cation exchange capacity, and those samples taken to be analyzed for organic contaminants. Traditional methods both of sampling and sample preparation did not consider this distinction to be necessary. This was basically due to the lack of information on how organic compounds respond to some kinds of treatment. It must not be forgotten to mention that analytical procedures for some inorganics and organics still need special attention because even now they cannot be treated sufficiently by following the relevant I S 0 standards on pretreatment. In some cases, no treatment may be the right treatment. In other cases, special treatments are described in the relevant analytical standards. Standardized analytical methods for the determination of inorganic elements in soil have existed at a national level for decades but mainly focus on determination of nutrients (Ca, K, P, Mg, Fe, Mn). With the increasing interest in environmental analysis standardized methods for the determination of inorganic contaminants were requested and there were not many of them. Adaptations were made by applying water analytical standards to soil analysis, and many of these mutated

International Standards ,for Sumpling

325

water standard methods are still in existence in many countries, and a great amount of soil data exists which is still based on these methods. Nowadays, methods are being developed which are especially suited to and tested on soils, because a lot of data based on water analysis have proved to be unrealistic. The aim behind better coordination between sampling, sample preparation, and sample analysis is to minimize the total error related to an investigation the main part of which is introduced by sampling. The inhomogeneity of a soil sample introduces a greater error than the analytical procedure in the laboratory. To decrease the error introduced by sampling I S 0 10381, Part 1 gives general information about all those critical aspects of sampling which contribute to error in case they have not been taken into consideration during the design of a sampling program. In the following, some of the critical aspects of sampling are reflected.

16.5 Preliminary Investigation A preliminary investigation should be carried out before any sampling or analytical program is specified or undertaken and the effort devoted to it will depend on the objective of the investigation. The principal objectives of the preliminary investigation are to gain knowledge about the present condition of the site, and of past activities on the site and neighboring land which may have affected it. A preliminary investigation usually consists of two stages, namely (1) the desk top study, which comprises the collection and examination of information on the site history, taking into consideration all kinds of maps, site plans, photographs, satellite imagery (if relevant), site records, and enquiries; (2) the site visit or reconnaissance, either as a result of the desk top study or independent from this. Depending on the local variability of the site and especially when suspect sites are investigated only experienced personnel should be chosen for this task. In addition, a limited amount of sampling may be carried out. The results of the preliminary investigation should be reported including factual findings and stating conclusions for the further program of investigation.

16.6 Selection of Sampling Patterns Sampling patterns are based on the estimation of the distribution of soil constituents - chemical substances, in most cases - on an area or on the type of substance input. In agriculture, a small number of convenient non-systematic sampling patterns are established. These are predominantly applied to obtain information on nutrient demand, i.e., to predict the amount of fertilizer to be applied to the soil. Other

326

A . Paetz utid G . CrGJmunn

applications in more recent times are the determination of pesticide residues or distribution of heavy metals. The general premise is that the distribution of soil constituents is relatively homogeneous. Very often these non-systematic patterns are shaped like the letters N, S, W, or X, but also sampling along a single diagonal or along a zig-zag traverse is commenced. Sampling points are fixed by taking samples after several paces and continuing this procedure until the necessary number of samples is obtained. Circular grids (polar grids) are often used for delineating local contaminations such as storage tanks but also to indicate the influences around a regional emitting source, e.g., precipitation from industrial plants [7]. Sampling points usually are located at the sections of concentric circles and the lines of the eight main points of the compass. Circular grids might imply a uniform extension of contamination in all directions which is not the case usually. Preferred directions, e.g., due to the main wind direction in case of airborne contaminants, should be considered in modifications of the circular grid such as increasing the number of sampling points in critical directions. In many cases, a regular grid is selected. Regular grids may be set up easily and grid dimensions readily varied. Also, interpolation between sampling points is easy. The grid dimensions, the assigned spacing, will differ according to the objective of sampling, e.g., to collect samples of average degree of a chemical substance, to locate isolated sources (hot spots) of contamination, - to establish the extension ofcontaminated zones both horizontally and vertically. ~

-

Random sampling in its purest form is rarely used in soil investigation programs although it may be applied especially in cases of presumably irregular occurrences of contaminated zones. Compared to that stratified random sampling has some advantages, mostly because of the preliminary subdivision of the site under investigation into a number of grid cells. Both random sampling and stratified random sampling have in common that setting up the grid and localization of sampling points is not as easy as with regular grids. Moreover, interpolation between sampling points is difficult and variation of the initial grid really causes problems. Grids based on unaligned random sampling are comparably complicated. Since the layout of designed grids in the field is rather time consuming and therefore expensive, and the exact localization of sampling points is essential for many investigation programs, the regular grid is most frequently selected, due to the experience of the IS0 working group, although everybody is well aware that the more sophisticated grids are closer to the approaches of statistics. Often requested was a figure for the optimal spacing of such a regular grid, or, in other words: Is there a prcferred spacing that covers at least the main objectives of sampling? The answer can only be no, since changes in local conditions must always be considered, e.g., variations of geological, pedological, and topographical features. A grid with squares of 50 m x 50 m applied in plain regions or in mountain regions covers particularly different areas (expressed in square meters) depending on the local topography. Either grids have to be adapted to landscape or results of an investigation covering both such regions have to be evaluated with utmost care.

International Standards for Sampling

327

But there are other sampling grids. More recently, also the application of equilateral triangular grids has been discussed, the advantage of which is also primarily a statistical one: the area of a given site remaining unsampled is thus smaller compared to rectangular grids of the same size applied on this area. Though the difference between the two areas remaining unsampled is not striking (- 18%) it might be of interest for sampling investigations to be carried out a t larger sites. If contaminations are following a line, e.g., caused by leaking pipelines, sampling can be conducted in the soil overlying the pipeline, or, if not practical, close to the pipelines or the respective structure. To date, three-dimensional sampling grids have rarely been applied in soil investigation. Positive experience is known from mining geochemistry but the procedures used there cannot be converted to soil since the objectives are different. The third dimension is covered best by use of trial pits. The statistical approach of three-dimensional grids is discussed in [5]. Apart from all sampling grids another type of sampling is practiced, called judgement sampling. This includes also sampling special spots of interest based on preliminary informations of a site. Such a sample may not be in direct relationship with a grid probably applied at this site as well, but nevertheless it can contribute to the aim of investigation. Judgement sampling can also mean that the sampler obtains soil samples based on his own experience and clearly indicates whether or not this sample is representative [6]. This kind of judgement sampling sometimes is attributed to random sampling but it must not be confused with the statistical random sampling described above. I S 0 10381, Part I does not give any direct preference to one of the methods mentioned but rather describes the advantages and disadvantages of these methods. It is in Parts 4 and 5 of this International Standard that examples for applications are given.

16.7 Sampling Depth Once an appropriate grid is selected, the depth of sampling has to be fixed. Two principal approaches have to be considered : (1) metric (depth related) sampling, (2) soil-horizon related sampling.

For agricultural purposes and screening analysis of potentially contaminated land often metric sampling is carried out. Usually, soil is sampled at a few fixed depths, e.g., at agricultural sites in the root zone between 0 to 0.3 m, at children’s playgrounds between 0 to 0.35 m. Intervals between these sampling depths may be equal or different, depending on the objective and local conditions. Metric sampling may also be appropriate in cases where the soil profile is not very much differentiated.

32X

A . Puerz cind G . Crij/hnunn

For environmental considerations and especially for matters concerning soil and groundwater protection I S 0 10381, Part I recommends sampling of soil horizons which is more representative and either exposed in nature, in a trial pit, or from core samples. Mixing of both techniques within one project should be restricted, and in any case the reasons should be reported.

16.8 Sample Quantity The sample quantity necessary was a matter of long discussion within ISO/TC 190. The first approach based on a table taken from the German Standard DIN 18123 [8] including lower limits for sample quantities used for particle size analysis. Slightly modified, these limits were for soils of, e.g., < 2 mm 1000 g and > 60 mm 18,000 g, respectively. Thus these limits may still be relevant for this kind of survey it was quite clear that these quantities do not reflect common practice in trace element analysis, and introducing these figures as standard values would lead to uncontrollable masses of soil in the laboratories. In conjunction with I S 0 11464 [9] it was resolved to recommend that at least 500 g of fine soil ( < 2 mm) as sampled in the field should be obtained for chemical analysis. The lower limits discussed were between 200 g and 1000 g but the majority of experts had the opinion that 200 g might not be enough, especially where archive samples are required. Losses of material during sample pretreatment have to be considered, too. The recommendation of 500 g leaves space for other options as required for specific objectives, e.g., the environmental specimen bank program of Germany, which is currently being developed, specifies samples of 5 kg for humus layers as sampled, and samples of 8 kg for A , horizons as sampled [lo].

16.9 Single Sample or Composite Samples Depending on the kind of chemical substances and the aim of the investigation it is essential to decide whether or not it is appropriate to prepare composite samples. If the average concentration of, e.g., heavy metals, of a rather homogenous site has to be known it is common to prepare a composite sample from a number of increments obtained at this site. The number of increments varies tremendously between countries, the range figured out by ISO/TC 190 is 15 to 50 increments to be joined in one composite sample. The total number of composite samples depends on the total size of the site investigated. Part 4 of I S 0 10381 specifies values given in Tab. 2. The situation is different for the analysis of e.g., volatile organic compounds or volatile mercury compounds. Based on the experience of experts in ISO/TC 190, composite samples should be avoided as due to the process of mixing a decrease in the content of volatiles can occur which leads to wrong evaluation criteria.

Internotional Standards for Sampling

329

Tab. 2. Number of Composite Samples in Relation to the Area of the Site (as recommended in I S 0 10381, Part 4, Committe draft version February 1994) ~~

~

Number of composite samples

Area in ha

0-2

> 2-5 > 5-10 > 10- 15 >15-20 >20-30

16.10 Preservation of Soil Samples Compared to the sampling of water where a lot of preservation methods exist which usually are applied [ll],such methods are rarely used for the preservation of soil samples. Especially the addition of certain chemicals to the sample to avoid variations of the sample condition is almost inapplicable without disturbing the soil structure or affecting other chemical soil properties. Among the few methods of preservation of soil samples cooling or freezing is widely used. The general rule to preserve soil samples in a more or less unchanged state is to transport them to the laboratory as soon as possible or to carry out necessary determinations immediately, possibly on site.

16.11 Use of Appropriate Sampling Tools and Containers The selection of suitable sampling tools in terms of the best possible technique to obtain a sample in the field is self-explanatory. However, special emphasis also should be given to the fact that the material the tools are made of do not introduce contamination into the sample, e.g., nickel or chromium plated tools may be subject to abrasion due to intensive use, sieve sets made of copper are known to introduce copper to the samples. Power-operated sampling tools may contaminate the direct surroundings of the sampling place by lubricant oils etc. After the sample has been removed from the main body of soil appropriate containment should be provided. Plastic bags may be suitable for material to be analyzed for traces of heavy metals or nutrients, they are definitely unsuitable for samples to be analyzed for organic compounds. The selection of containers has to take into account several factors beside the problems of cross contamination, e.g.: -

resistance to extreme temperature, resistance to breakage,

330

A . Ptrctz and G . Cr@ntmn

-

water and gas tightness, availability, potential for cleaning and reuse.

Cost effectiveness is a key word in this respect, and experience has shown that in the past containers were selected for financial reasons especially based on storage, transport and reuse facilities and not taking into consideration aspects of cross contamination or losses of volatilcs due to leaky containers. Containers once used for storage of contaminated soils should preferably not be used again because cleaning these containers may cause incalculable risks to human health. If in doubt, special disposal of these containers is recommended and national legislation on waste disposal to be observed. ISO/TC 190 has also considered sample container preparation and cleaning procedures as solicitated by EPA [12] but resolved that this practice is not subject to International Standardization at this time.

16.12 Sampling Report The sampling report should be as detailed as necessary for the special aim of investigation. This may include a very detailed description of the site but may also be restricted to a few cssential provisions necessary for the final evaluation of the analytical results. Based on recent developments in Germany [13] a minimum set of data was introduced in I S 0 10381-1 (and taken up again in parts 4 and 5 of this Interntional Standard) including title data, e.g., sample number, coordinates of the locations; site data, e.g., minimum information of the site utilization; - sampling procedure, e.g., information in the depth of sampling, the sampling tools used ctc.; - transportation and storage, e.g., material of container, cooling or freezing procedures or other measures of stabilization, time of transportation. -

-

16.13 Quality Control Emphasis is given to quality control which does not start in the laboratory but when beginning the design of a sampling program. Although, up to now, there is no standard procedure existing that meets all aspects of sampling the main requirement to sampling is the possibility of backtracing sampling procedures to enable the investigator to check for inconsistencies of analytical results. It is therefore esscntial that the sampling report is available at every stage of the investigation.

Internutionul Stundurd~7,fbrSampling

33 I

16.14 Pretreatment of Soil Samples A major issue concerns the pretreatment of soil samples before chemical analysis is conducted. Just a few national standards and guidelines exist giving detailed information on this matter or even describing conclusions which have to be drawn from different kinds of pretreatment in the calculation of the analytical results or the evaluation of these results. Generally, two different methods of pretreatment are regarded by ISO/TC 190: (1) pretreatment of soil for physico-chemicai analysis, (2) pretreatment of soil for the analysis of organic compounds (also referred to as organic contaminants). Developments with regard to physico-chemical analysis resulted in I S 0 11464 [9]giving guidance, not strictly procedures, on drying, sieving, subsampling, breaking, and milling of soil samples. The following reflections in the course of developing this International Standard caused controversal discussion between soil scientists and chemical analysts: -

-

-

Should only the soil particle fraction of < 2 mm (commonly designated as fine soil) be used for chemical determinations? If yes, how to deal with the remaining larger amounts? If no, how to further pretreat the larger amounts? If it is the < 2 mm size to be focused on, should this fraction then be further decreased in size and what size then will be necessary? Should soil samples be dried in the oven, in the air, or perhaps freeze-dried? Will the result of determination assumed later be expressed as fresh soil, oven-dried soil or air-dried soil? Will this expression be related to volume or mass of soil?

Many of the preparatory and analytical techniques for soil materials have been developed in geochemistry and have later been adopted by soil analytical laboratories with some minor modifications only. One example is the widely used particle size of < 100 pm, sometimes also referred to as analytical fineness, in the analysis of soils and sewage sludges [14]. It is the total or so-called total content which is being analyzed in these cases after digestion of relevant test portions to predict future accumulations of heavy metals from sewage sludges applied to agricultural land, and to give advice of further sewage sludge applications to soils. Besides the discussion that neither a size fraction < 100 pm nor the determination of total contents give realistic results with regard to ecotoxicological aspects, it has been shown that at least for some types of soil the influence of grinding soil is less than expected and there are no statistically significant differences between soil just sieved to < 2 mm and this fraction further decreased in size by milling to < 150 pm [15]. Hence, it follows that excessive milling seems not to be necessary for a lot of standard chemical determinations. TSO/TC 190 decided to add a corresponding note to I S 0 11466 [ 161, an International Standard describing the extraction of trace metals from soil using aqua regiu. However, ISO/TC 190 also resolved that further testing is required to find certainty about which soil types and elements are not affected by grinding and milling.

332

A . Ptretz nncl G . Criijmcmn

Finally, it is stated in I S 0 11464 that milling soil to <250 pm is sufficient for most chemical analyses to be carried out. Alternatively, and if requested by the customer, sieving to < 2 mm will be sufficient. Accordingly, the mass of the test portion used in the extraction or digestion has to be adapted. I t was the opinion of analytical chemists also to break and mill soil material of sizes larger than 2 mm for any kind of determination of the total content. Knowledge of the total content is needed only for some very few objectives, and much more interesting is the mobile and mobilizable content of trace elements in soil. These compounds are hardly correlated to stones or other bigger fragments in soil but to the fine soil fraction. The general recommendation of I S 0 11464 is to select material > 2 mm from the sieve and to record the mass and/or volume and kind of material. Whether or not this part of the sample can be discarded then depends on the objective of the investigation. The results of the analytical determination should be related to oven-dried soil which is the most common practice in soil investigation programs. This requires a determination of the dry matter and the water content of the sample, preferably according to I S 0 11465 [17]. The routine drying procedures recommended in I S 0 11464 for large sample series are either drying the samples in air, or in an oven at 40 "C. I S 0 11464 gives preference to oven-drying because the increased speed of drying limits changes of the sample condition due to microbial activity. It is also important to state if the results of analysis are related to soil mass or soil volume. The latter is to be preferred for ecological reasons. The second method of pretreatment is currently being developed for samples which have to be analyzed on organic contaminants. The properties of organic (micro-)contaminants may differ very much. They can be very volatile to non-volatile, they may decompose at high temperatures, they may be biodegradable or UVdegradable, and they have different solubilities in water. Again, ISO/TC 190 is well aware that there cannot be a uniquc method of pretreatment, but a framework can be provided [18]. The method of pretreatment depends on the volatility of the compound or group of compounds to be analyzed and two categories are distinguished in the future standard: -

volatile compounds, i t . , hydrocarbons having a boiling point below 300 "C (at a pressure of 101 kPa), moderately volatile compounds, is., hydrocarbons having a boiling point above 300 "C (at a pressure of 101 kPa).

For volatile compounds, it is recommended that the samples should be cooled, the extraction solvent possibly added in the field, and the preparation of composite samples be avoided (a composite sample may be prepared later by joining extracts from the individual samples). For the accurate determination of moderately volatile compounds the sample is chemically dried at low temperatures using sodium sulfate and talcum powder. The drying time is between 12 to 16 h. It should be noted that the amount of sodium sulfate needed will incrcase with a higher water content of the sample. This especially

International Standards ,for Sampling

333

regards samples obtained from sludges and sediments. Following this drying stage and before cryogenic crushing can be carried out, the samples have to be cooled to - 196 "C using liquid nitrogen. Average cooling time is about 30 min. After complete cooling the sample is crushed using a suitable apparatus. Then analysis should be performed as soon as possible.

16.15 Summary Regarding the approach of developing International Standards for the examination of soils recent results are reported giving an indication of the various problems going along with this task. The first problem is always introduced by the soil itself. In general, the heterogeneity of soils in terms of, e.g., particle size distribution, soil-water regime, and chemical composition, makes it impossible to standardize just one procedure of sampling and pretreatment. This would lead to a standardized introduction of error into trace element analysis. Secondly, there is the fact that many of the procedures which are in development by ISO/TC 190, Soil quality, are pioneers' work rather than the result of harmonizing existing national standards. The third major problem concerns the expectations of the customers and contractors of soil investigation programs. Methods should be as reliable and as easy as possible, to facilitate application in many countries and for a wide range of objectives. ISO/TC 190 has tried to consider all these problems and to provide answers to them. Although the main work in the fields of soil sampling and pretreatment has been done future experience will certainly identify aspects which are not or not sufficiently covered by the International Standards on sampling and pretreatment of soil samples which will be available to the public in 1993 and 1994. Acknowledgements. The authors would like to thank all colleagues of ISO/TC 190 participating in the development of International Standards for the sampling and pretreatment of soils. Special thanks are regarded to Ron Bosman of TNO, The Netherlands, and to Colin Coggan of Clayton Environmental, United Kingdom. Thanks are also dedicated to Christiane Schuffler for typing of the manuscript and to Barbara Clark for improving of the English.

16.16 References [ I ] British Standards Institution (1988) DD 175 Code ofpracticefor the identification qfpotentially contaminated land and its investigation. 121 Nederlands Normalisatie-instituut (1991) N V N 5740 Bodem, Onderzoeksstrategie bij verkennend onderzoek.

334

A . P u ~ t zrind G . Cri$minin

[3] Amt fur Standardisicrung, Menwesen und Warenprufung (1985) TGL 42316 Nutzung uiid Schutz des Bodcns; Bodenvcrunreinigung; allgemeine Fordcrungen an die Entnahme von Bodenproben. (Authors' remark: T G L 42316 was a national standard of the former German Democratic Republic (GDR)). 141 International Organization for Stand;irdization (1993) /SO 11074-1 Soil quality - Vocabulary - Part I : Terms and definitions relating to the protection and the pollution of the soil (at present at the stage of Draft International Standard). [5] Nothbaum, N., Scholz, R. W. (1992) Prohcnplonurig und Diitencinalyse hei korztutninierten Bijden. Projektbericht, erstellt im Auftrag des IWS-TU Berlin (Unterauftrag im F uiid E-Vorhaben ..Entwicklung eines Konzeptes zur Ablcitung von Sanierungswerten fur kontaminierte Boden", T U Z 10703007(06)) 126 p. [GI Stcin A,, Wopereis. M. C. S. (1992) Sampling strategies and geostatistical techniques. in: Mun~iiil, f h r soil 17j~dra~ilii~ measurements,ji:eld nionitoring techniques and sampling strategies f o r ricp h u m / cropping sy.rtems, research program on soil management for increased and sustainable rice production (Draft). The Winand Staring Centre for Integrated Land, Soil and Water Research. [7] Bayerisches Geologisches Ldndcsamt ( I 992) M c r k h h t f i r die Anlage von BodenmeJnetzm Z U I Bi~ohaclii~ing rind Beii'c>i,~siclierung hei P r o h l i ~ m s t ~ ~ ~ ~ t i i i r i ~ ~ n t i ~ n . [8] DIN Deutsches Institut fur Normung e.V. (19x3) D / N 18123 Subsoil; testing of soil samples; determiiiation of the particlc size distribution. [Y] International Organization for Standardization (1 992) I S 0 11464 Soil quality - Pretreatment of samples for physico-chemical analysis (at present at the stage of Draft International Standard). [ 101 Sprengart. J., Wagner, G. (1993) Unii~'I.lrp,ohenhanl~drs Bundim - Richilinie zur Prohi~nuhriii~ und Prohenrrirfbereiiurzg. lnstitut fur Biogeographie, Zentruin Uniweltforschung. Universitiit des Saarlandes. [ I I ] International Organization for Standardization (1985) I S 0 5667-3 Water quality - Sampling - Part 3: Guidance on the preservation and handling of samples. [ 121 EPA ( 1990) E P A ,spwifi'cutionsand guidance for ohtuining contaminant-jrre sumplc cnntainer.c. EPA Region 10, Seattle, WA. [ 131 Bodcnschutzzentrum des Landes Nordrhein-Westfalen (199 1) Mindestdutensutz Bodi~nuntivscichungen, AbschluBbericht des Arbeitskreises Mindcstdatensatz Bodcnuntersuchungen der Sonderarbeitsgruppe Informationsgrundlagen Bodenschutz. [I41 Kliirschlammverordnung (Abf'KliirV) vom 15. April 1992. - BGBI. 1992, Teil 1, 912-934. [I51 Houha, V. J. G., Chardon, W. J., Roelse, K. (1993) /nflztence ofgrinding qf'soil on apparent clieniii~uli~otnpo.sition. DLO-lnstituut voor Bodemvruchtbaarheid, Nola 263. [Ih] International Organization for Standardization (1992) /SO 11466 Soil quality - Extraction of [race metals using aqua regia (at present at the stage of Draft International Standard). [ 171 International Organization for Standardization (1992) I S 0 11465 Soil quality - Determination of dry matter and water content on a inass basis - Gravimetric method (at present at the stage of Draft International Standard). [ 181 lSO/working document: Soil quality - Sample pretreatment for determination of organic contaminants in soil. Sccond draft, Document ISOjTC I9O/SC 3/WG 9 N 45, May 1993.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

17 Fixed and Hypothesis-Guided Soil Sampling Methods - Principles, Strategies, and Examples Roland W . Scholz, Norbert Nothhuunz, and Theodor W. May

17.1 Introduction The development of an appropriate statistical soil sampling method is of prime importance for the quality of data and for the evaluation and the interpretation of the results. For the problem field of soil evaluation in the frame of strategies for the protection of the environment, several attempts have been made by work groups established by DIN (Deutsche Industrie Norm, German industrial norm) and I S 0 (the International Organization for Standardization)to develop guidelines for soil sampling. In the practice of environmental soil evaluation, however, potential sources of errors residing in taking, transporting, and processing samples and in analysis are as yet not taken into account seriously enough. Although sophisticated statistical sampling procedures are applied in geology, this method is in its infancy in environmental practice. One of the reasons is that environment analysts (in most cases chemists, biologists, geologists, or toxicologists) consider themselves as experts for the object level, e.g., for biochemical processes, while according to training and genesis, they are not ex~ planning of experiments and sampling protocols perts for the so-called sign I P V C The as well as the quality control of laboratory analysis, however, are mathematicalstatistical methods. Their conceptional and constructive bases require objectindependent operations at the sign level. If too little attention is given to planning the experiment, deficits in the economy of procedure, but also in the representativity of data, in the sufficiency and indicativity of results must be suffered. Conversely, environmental analyses are of course worthless if they are not related to previous knowledge and to the model of the object. In principle, planning of soil sampling must be understood as being dependent on previous information and objectives of the study. At this general level, there will certainly be no disagreement. Decisive, however, about planning a sampling and protocol is the interest of the assessment. The first column of the schema in Fig. 1 includes various types of previous knowledge and pollution hypotheses. If we consider the knowledge desired, different goals of knowledge having an influence on the soil sampling method can be distinguished independently of parameters and direction of the inquiry (i.e.,on the sign level). With regard to the knowledge desired, we distinguish four cases: (1) Determination of a mean value

The agricultural sector is frequently interested (as in the case of sewage sludge disposal) in the determination of mean values. This is often a low-cost procedure: by using mixed samples and an optimal distribution of the sampling points,

336

R . W. Sciiolz, N . Nothhaurn, und T. W . M a y

I

Vevious Knowledge I Wlution Hypotheses

I

I1

Knowledge Desired Values

Determining Natural Nutrient/ Agricultural Use Determining Polluting Agents Imrnission from Industrial Areas

‘Point’-ShapedSources (Balteries, Barrels)

v

W

W

0

J

The arrows indicate the direction of increasing cost

Fig. 1. Schema concerning the influence of starting point/previous knowledge o n the complexity of thc soil sampling plan to he choscn based on a specific cxample.

reliable statements concerning the mean contamination are possible even on the basis of a small number of soil samples. (2) Defernfination of nlaximurn values

Maximum values are of interest if an immediate hazard emanates from the soil system, e.g., in case of skin contact or ingestion. For such questions, the rcquirements for the soil sampling plan are higher. Using mixed samples usually is not justifiable, and a soil sampling pattern must be used which offers sufficient certainty of “discovering” and assessing maximum values, respectively. (3) Deternzinution of distributions

Determining value distributions is of significance for testing immission hypotheses. If knowledge about the distribution or profiles of contaminants is desired, a particularly comprehensive and careful sampling is necessary. (4) Verification of pollutionlcausal hypotheses Often, the objective is an assessment of the relative contribution of different sources of pollution. Before we deal with fixed and hypothesis-guided soil sampling methods, we would like to introduce a checklist of criteria which have to be met in every case of planning soil sampling (Tab. I). Some criteria have been taken from statistical experiment planning and from psychological test theory. They have been specified for ecological sampling.

Soil Sampling Methods

Tab. 1.

337

Checklist of Criteria for Ecological Sampling/Study Planning

A. Can the knowledge desired be obtained by the study planned? Representativity :

Does the sampling plan yield a good image of the real soil contamination, or is a systematical bias (overestimation or underestimation) of the data to be expected?

Sufficiency:

Are there sufficient data? (Is it necessary to calculate the sample size in order to obtain reliable data, given a specified error probability and an assumed critical effect size?) Is the sample big enough to allow for sufficiently precise statements about mean or variance of the actual data? Have all the relevant parameters been included? Is it possible to derive statements about spatial distributions or temporal changes?

Indicativity :

Do the soil sampling plan and the method of analysis chosen also permit the detection of pollution levels which are only slightly higher than the limit? Are the sum parameters selected indicative for specific contaminants? Are the selected contaminants or the analytical method indicative for the effect in question?

B. How reliable is the result of the study? Objectivity :

Reliability:

Validity:

Is the method of taking soil samples independent of the sampling person? Is the method of taking samples reconstructable (localization of mixed samples, localization of grid points)? Is the sampling error quantified, e.g., by a sufficiently large number of sampling repetitions? Is there an indication as to how much uncertainty in the data must be expected because of random effects on sampling, e.g., by sampling errors? Was the laboratory/analysis error quantified by repeated analysis of reference samples? Are there indications as to how much uncertainty in the data must be expected because of random effects of sample preparation and of the analytical method? Do special circumstances during sampling (rain, dryness), during sample transport or storage affect the data? Is the selection of the parameters determined sufficient to obtain the knowledge desired? (Can pollutant concentrations in vegetables, for instance, be predicted from known soil concentrations? Is the knowledge of additional soil parameters necessary? Relative importance of contaminant speciation for bioavailability and mobility?) Have all relevant intermitting variables (e.g., soil parameters) becn taken into account?

C. Can the knowledge desired be obtained at smaller cost? Economy:

Are more samples obtained than statistically necessary? (Calculating case number!) Is it possible to replace a complete sampling of several areas by an appropriate sampling plan? Is it possible to economize by early exclusion of certain hypotheses by n.eans of a sequential procedure?

338

R. W . Scliol;, N . Nothhuurn, und T. W . Miry

17.2 “Fixed” Sampling Plans/Grid Plans The principle of fixed sampling plans includes the application of a generalized grid plan to a specified situation in order to bc able to assess relevant characteristic data like mean, maximum, or distribution of a contamination. Criteria for selecting suitable fixed sampling plans can be derived from the previous information supplied and from the kind of knowledge desired. Fig. 2 shows how the selection of various types of grid plans is dependent on assumptions about the contamination distribution in the area to be sampled. As can be seen from the figure, there is no optimal grid plan, the plan must rather be chosen depending on the problem and the answers obtained according to the above checklist. The advantages and disadvantages of various fixed soil sampling plans arc illustrated by a real example: In the outskirts of a big city, building ground was offered on the site of a former ammunition factory. An environmental research institute subdivided the lot into about 100 plot areas. From each of these plot areas

Rectangular Grid / Bottle Rack Grid

I

Homogenous Contamination

Rectangular Grid / Bottle Rack Grid

Centered Contamination

Polar Grid / Rectangular Grid / Bottle Rack Grid

Contamination along a Line

7

Modification of the Rigid Grid Dependent on:

I

I

Particularitiesof the Area Type of Objective / Hypotheses Quality and Reliability of Previous Information

Fig. 2. Selecting and adapting a fixcd soil sampling plan according to previous information and to objcctives.

339

Soil Surnpling Methods

7a

KT-J

3b. Random grid

3a Grid plan used in the example

0

F - 7 0

0

0

a

0

a

L'1 3c. Limited random gnd

3d. Rectangular grid

3e. Bottle rack grid

Fig. 3a-f.

3f. Diagonal grid

Examples for various fixed grid plans

a sample of 10 holes was taken. The grid plan in Fig. 3a was considered to be favorable and applied. Now it was suspected that a residual stock of TNT (trinitotoluene) had been buried during demolition work in a former crater-shaped hole. The potential house builders wanted to know the probability of discovering such a crater by a given sampling procedure. As there was no information concerning the location of the crater, there was an equal probability for all areas of certain plots in the vicinity of the former factory (assumption of equal probability). In order to assess the probability of really hitting the crater with the sampling, (model)

340

R W . Sckok, N . Nothhuum, and T. W . May

assumptions had to be made. To simplify matters, we assumed that the crater was circular. Besides, we assumed that a mixed sample would be indicative if at least one sample hole hit the crater area. Thus we also neglected the depth of the contaminant location. For a crater of 40 square meters (approx. 160 square feet) and for a mixed sample taken from a total of 10 sampling holes, and a plot area of 2Ox20meters, this applied plan yields a hit probability of 87 percent! I). The hit probability was fiir determined by means of the simulation program “~e~~erwuhvscheinlic,hkeit AuJtrugungs-Flachen ( TRAF)“ developed by the “Gesellschaft fur Organisation und Entscheidung”. The house builders concerned also made their own calculations. Their assumptions (presumably developed by mathematics teachers) were that the sampler had selected the sampling holes purely at random and stochastically independently, as shown in Fig. 3b. Such a plan, for instance, leads to a hit probability which is 25 percent smaller, amounting to about 65 percent. Such a soil sampling plan, of course, is not only unrealistic, but also suboptimal. This can be easily made plausible by the fact that random sampling also will take samples at spots very close to each other, thus dismissing potential information. This effect could be avoided by subdividing the area, as shown in Fig. 3c, and then proceeding to random sampling of the individual plots. This plan again is suboptimal, as the lattice grid in the examples shows. The optimal grid plan under the assumptions made is the “bottle rack plan” of Fig. 3e. The mean hit probability of this grid plan is 97 percent, and thus 12 percent higher than according to the plan 3a actually used, and a full 49 percent higher than if the random plan 3 b had been used. When assessing the economy of the “bottle rack grid”, however, the time required to fix the sampling points must be taken into account as well. Occasionaly, a plan optimal under statistical aspects can be less economical than another, if the cost in terms of the time required for establishing the sampling points exceeds the cost for additional sampling holes and analyses. At least for cost-intensive sampling (e.g., deep-hole sampling) or expensive analytics, however, the ,,bottle rack grid” should be applied in any case. Fig. 4 shows the probabilitics for finding a contamination crater with a given radius as a function of the grid density for the “bottle rack grid” and the rectangular grid. It can be seen that the increase in hit probability by using the “bottle rack grid” is also dependent on the grid density. In the examplc quoted (area 10,000 square meters, crater radius 10 meters), a maximum is found at a grid density of about 40 sampling points (see Fig. 3 - 5). Increasing the number of sampling points in case of mixed samples creates the problem of diminuation of the contaminant content as the number of sampling points increases. In the worst case, the mixed sample is not even indicative when one of the sampling points is situated within the crater, as the contaminant content is reduced by the large number of samples taken from the uncontaminated area. ’) This term simplifies by assuming that each plot is surrounded by neighbor plots subjected lo

identical sampling.

Soil Sampling Methods

34 1

Rectan ular Grid 0,4

2

0,2

.......................................

.......................................

00

20

40

60

80

100

Number of Grid Points

Fig.4. The probability of detecting a contamination crater as a function of the grid density, illustrated for a rectangular and a “bottle rack” grid; crater radius 10meters on an area of 10,000 square meters.

Further questions and problems: (1) Determining the hit probabilities dependent on - the number of sampling points or the crater size (taking into account the indicativity of mixed samples), - the grid plan selected, - the shape of the contaminated area, - optimizing margin sampling. (2) Evaluating the efficiency by means of gain/loss functions for - costs of sample taking and analysis (including the cost of establishing the sampling points), - costs incurred if a contamination crater escapes discovery.

.....................................

.....................................

....................................

....................................

Number of Grid Points Fig. 5. Probability difference between “bottle rack” and rectangular grid for the detection of a crater as a function of the number of grid points; crater radius 10 meters on an area of 10,000 square meters.

342

R. W. Srholz, N . Nothhaum, and T. W. M ay

17.3 Soil Sampling Plans Guided by Hypotheses For the identification of possible sources of contamination or in order to clarfiy the variance shares of various interactions, fixed soil sampling plans are not sufficient. The testing of specific, determined hypotheses concerning initial pollution by contaminants and later spreading and reduction require soil sampling plans guided by hypotheses. It is necessary to find samples critical for a hypothesis decision, and to systematically exclude possible alternative explanations. The testing of specific hypotheses, however, does usually not occur at the beginning of studying certain ecocompartments, but is, as a rule, the result of a sequence of studies which have served to sophisticate the questions asked and the sampling plan. We should like to illustrate the quasi prototypical development of a hypothesis-guided soil sampling plan by another example, by the studies concerning fl-HCH (beta-hexachlorocyclohexane) contaminated parts of the East-German community of Bitterfeld during the years 1990- 1991 (see Scholz ct al., 1991). After the reunification of Germany, an ecological status investigation was started in Bitterfeld towards the end of 1989, since it has to be assumed that the evident changes in the environment possibly had had an effect on the population. The starting point was an indicator study in which at first a small number of samples of different foodstuffs were unsystematically drawn. These were foodstuffs which were available and consumed in critical areas. Within a general screening analysis for different groups of contaminants, only one single milk sample showed a critical load of 8-HCH. The goal of the second study was to investigate first whether the increased contaminant level obtained was a singular event, or not (quantification of load study). Besides, possible sources and paths of contamination had to be explored in order to find the responsible source and to be able to reduce the foodstuff contamination. Hence, the first two studies were mostly explorative. Their objective was to open .access to the problem field and to establish a basis for a study intended to verify hypotheses. In the second study, there was no systematical sampling plan either. In order to ascertain and to further differentiate the P-HCH load, milk and tissue samples were taken from several herds. To explore the contamination path, local authoritics took samples from river water, pasture soil, and stable wall paint. The attempt was to obtain at least one critical sample for each potential pollution path (hypotheses). The milk obtained from the several herds in the area under investigation was contaminated to different degrees, in some cases critically. Thus, the distribution of contaminants across the Bitterfeld area was not homogeneous, as it is typical for immissions. Suitable maps showed that all the herds with a high load grazed near the river (hypothesis 1) or near a fl-HCH waste disposal area (hypothesis 2). In a third study, hypothcscs of immission were tested by a statistical layman. The task was to test whether extent and sources of the contamination can be determined. Sampling unfortunately was not based on a sampling plan relying on methods of analysis of variance, and thus certain questions could not be tested because of

Soil Sampling Methods

343

incomplete sampling (see Nothbaum and Scholz, 1991b). Substantial contaminations were only found in the river’s flood zone. The fourth study was to classify the areas under consideration by using statistical experimental designs and to test the hypotheses of contamination, as several polluters were under suspicion. By means of a three-factor design of analysis of variance with the factors pollution A and pollution B (both dichotomous), as well as a 4-category soil level classification as indicator for flooding frequency, mixed soil samples were drawn from 33 of a total of 51 plots. The results reveal a significant influence of the “new” pollution and soil level on contamination. Except for one all plots could be classified as “suitable for grazing” or as “unsuitable for grazing”. Plots not sampled were assessed by using the median of all values from the same area.

17.4 The IHEARU Schema Each of the enquiries sketched in Fig. 6 is subdivided into six steps. We call this subdivision the IHEARU schema. Basically, this schema can be applied to any

1. Indicator Enquiry I General Screening

2. Quantificationof Pollution and Investigation of Sources

I: General concerns in the population / ‘obvious’changes of the environment

H Presence of contaminations / damage E: No explicit sampling . _ plan _ A: Unsystematical foodstuff samples

R: B-HCH contamination in milk

1 U: Further studies I

1L

I I

polluters H: Contamination dependent on pollution €5 Variance analytical design A: Sampling in flood area R: Classifying areas according to pollution and flooding U: Giving up certain areas as pastures .

I

I: Checkingtheresults of enquiry 1. / Exploration of sources of pollution H Milk pollution by refuse dump / sewage impact E No entirely systematical sampling plan A: Sources: soil, water, stable paint; acceptors: herd milk, tissue. R: Soil pollution; milk contaminations according to localitv U: Enquiry &to diverse river areas /polluters

on different pastures? H: Various contaminations by various polluters E Monovariableplan (pair samples) A Samples in the river and dump area R: Increased pollution in the highwater area of a

I

I

U: Assigment for systematical classifcation of areas

1

344

R . W . Srholz, N. Nothhaum, and T. W . M a y

I

H

E

A

R

U

Initial information /

Hypotheses

Enquiry

Adequate

Results

follow-Up

design

sampling plan

starting point / goal

ecological study, but it should be implemented particularly in case of sequential studies in order to record the results of the various studies, as these are the starting points for the next. The IHEARU schema describes the enquiry’s starting point/Initial information(1) and Hypothesis/goal (H). It shows which type of Enquiry (E) has been used and hence whether the sampling plan is Adequate (A), or not. By making Results (R) and follow-UP (U) explicit, is shows the status of knowledge attained and whether further enquiries will be necessary (Fig. 7).

17.5 Conclusions The case study “HCH-Contamination Bitterfeld” serves to illustrate the sequential character of the search for information in the analysis of cause-and-effect relationships in the environment in two ways. Firstly, various studies are based on each other in the sense that the results/ follow-up of the previous serve as the starting points for the next, showing that knowledge about statistical significance of relationships is required already for study two, and that the statistical effort increases from one study level to the next. The studies on the cause-and-effect relationships in the environment demonstrated that an environmental expert, working on the modeling and conceptualization of environmental effects, should not only be familiar with the theory of testing hypotheses, but also with multivariate schemata. These include completely crossed designs of analysis of variance and in particular the concept of interaction in order to be able to work competently, say, on the problem posed in the second example. These statements do not mean that the calculatory methods have to be mastered (these are a natural domain of the experts for the sign level, i.e., the mathematicians/statisticians and can be delegated to computersupported systems besides), but rather refer only to the qualitative schemata of the stochastical methods quoted. It is obvious that intuitive knowledge is not sufficient for a complex task like “hypothesis-guided search for information in the environment”. In addition, it was demonstrated that the choice of a suitable grid plan may also lead to complex statistical questions, as the example of fixed sampling plans has shown.

Sail Sampling Method~y

345

17.6 References Hortensius. D., Bosman, R., Harmsen. J., Wever, D. (1991) Entwicklung standardisiert. Probenahmestrategien fur Bodenuntersuchungen in den Niederlanden. NAW V12 02 -91. Landesanstalt fur Umweltschutz Baden-Wurttemberg. Inst. fur Altlastensanierung (1988) Proberiahme Boden, Abfall, Crundwasser, Sickerwasser .fur die chem.-physikal. Untersuchung. Vorlaufige Vercdhrensrichtlinie, LA f. Umweltschutz, Karlsruhe, 8 pp. Netherlands Normalisatie Instituut (1 991, September) Dutch Draft Standard-Soil: Investigation Strategy for Exploratory Suruey (1st. ed.), (UDC 628.5 16). Nothbaum, N., Scholz, R. W. (1991 a) Probenplanung und Datenanalyse bei kontaminierten Biiden - Projekthericht (erstellt im Auftrag des IWS-TU-Berlin). Bielefeld, Technical Report. Nothbaum, N., Scholz, R. W. (1991 b) Strategien hypothesengeleiteter Informationssuche im Umweltbereich, Technical Report. Scholz, R. W., Nothbaum, N. (1991) Beprobung und Flachenklasslfikation van Oberboden in den Uherschwemmungsgebie ten der Mulde beziiglich persistenter Clzlorkohlenu.assersto~fe.Bielefeld: GOE, Gesellschaft fur Organisation und Entscheidung. VDLUFA (1991) Die Untersuchung uon Boden. Methodenbuch VDLUFA. Darmstadt VDLUFAVerlag.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

18 Sampling for Trace Analysis of Lake Sediments Ursula M . Cowgill

18.1 Introduction Sampling and experimental design are of utmost importance in assuring the ultimate success of a study. It has been frequently stated (Keith et al., 1983, 1988; Hikanson, 1984) that more money and time have been wasted in environmental studies due to poor experimental design than for any other cause. It has also been pointed out by Gy (1 979) that improper experimental design and sample handling may introduce more error than is inherent in the analysis of the material. As yet no standard procedure has been established, such that two investigators, could unknown to each other, visit the same water body, take sediment samples in the same place, using similar equipment, and on chemical analysis arrive at concordant results. This is not particularly astonishing in view of the complexity of sediments. Hikanson and Jansson (1983) have noted 12 different factors that might influence the information it is possible to derive from sediment analysis. These factors include the size and depth of the water body, physical, chemical and biological characteristics of the sediment, anthropogenic factors, surface mud characteristics, currents, whether the samples are taken in the littoral or pelagic area of the water body under study, sampling devices, sample handling, number of samples collected and analyzed and finally the trustworthiness of the analytical results. The biggest single problem in the sampling of sediments to be used for trace analysis is the procurement of a representative sample. Under normal circumstances, in a lake, for example, the best place to take a sediment sample is in the deepest section where disturbance of the water column is minimal. Here, sedimentation proceeds in as orderly a fashion as is possible in nature. However, for the purpose of trace analysis, the intent of which is pollutant monitoring, the sample will normally be taken near the discharge area which usually is situated in or near the littoral zone, which often is the most disturbed and active portion of any body of water. In addition, continuous or near continuous turbulence is contributed by the discharge. Under such circumstances, a representative sample is almost impossible to obtain. Continuous sampling over time of such a situation will reveal trends. The ideal objective of any sampling program is to procure a representative undisturbed sample of the sediment to be investigated. Bottom sediments consist of a variety of particles varying in size, shape, and chemical composition which have fallen through the water column. The origin of these particles may be autochthonous such as dying algal blooms, fecal pellets, precipitated compounds, the composition of which exceeded the solubility product and thus a crystalline substance was formed, or an aged zooplankter that died. In

348

U . M . Cowgill

contrast, some particles, in fact, under most circumstances, the vast majority of particles are allochthonous in origin. That is to say, they have been transported by water, wind or ice action from their terrestrial environment. Such particles accumulate on the bottom of bodies of water at the rate of a few millimeters a year or in some areas as much as several centimeters per year. The “rule of thumb” for the world is a meter per millenium. Once these particles have sedimented to the bottom of the body of water under study, changes may occur that alter their structure as well as their chemical composition. Some of these changes are brought about by reduced oxygen concentrations or a lack of oxygen. Others, especially in the littoral zone, may be initiated by burrowing organisms. Some changes cause the release of chemical compounds from the particles with which they were associated resulting in their entrance back into the water column where they remain in the water column or enter the food chain. Thus, bottom sediments are viewed as a source of geochemical history of the basin in which the particular lake lies, as a contaminant sink as well as a source of contamination of the aquatic environment. This contribution will review sampling tools, though not exhaustively, some thoughts on experimental design and the number of samples required to obtain an adequate picture of the trace chemistry. In addition, adequate sampling of bottom sediments, surficial sampling as well as sampling at depth, mud/water sampling and pore water sampling will be discussed. The objective is to provide a basis for adequate sediment sampling for the purpose of trace analysis.

18.2 Sampling Devices This contribution is not intended to be an exhaustive review of sampling tools and techniques but rather an overview of the subject. Reviews of samplers and sampling techniques may be found in Murdoch and MacKnight (1991), Hopkins (1964), Wright et al. (1965), Bouma (1969), Sly (1969), Elliott and Tullett (1978, 1983), Robbe (1 98 I), and Hikanson and Jansson (1 983). The object of sampling is to extract a subsample of the sediment that resembles the original material as closely as possible. This subsample should be collected in such a way that in the process of extraction it should remain undisturbed. Mixing and compaction should be avoided as well as losses of water, particulate material and so forth. To attain these goals, the sampler should be lowered slowly to avoid setting up currents or causing shock waves that might disturb recently sedimented material. There are three types of sampler that are commercially available : dredges, grabs and corers. Dredges will not be discussed here since they are only used for qualitative studies as it is not possible to control the location or depth of the sampler. Thus, it is not possible to take, with any degree of certainty, samples of surface mud next to each other (Plumb, 1981). Grab samplers are used to collect surficial samplers. These samplers usually take only the surface 1 to 3 cm and are used for spatial

Sampling of' Luke Sediments

349

distribution of trace elements and surveys. Corers may be used to obtain both surficial as well as sediment column samples. These samplers extract mud with the least amount of disturbance. Such mud may on analysis provide historical information on trace element trends, may locally provide background values of trace substances and may establish trace element trends over time. Tab. 1 shows the types of samplers in relation to sediment depth sampled. Tab. 1. Sediment Depth Collected by Different Samplers under Optimal Conditions ( - 2 m of line-grained mud) Depth sampled

Sampling devices

0- 10 cm

Birge-Ekman, Ponar, mini-Ponar, mini-Shipek-light weight, small volume grabs Van Veen, Smith-McIntyre-heavy, large volume grabs Kajak-Brinkhurst, Phleger corers-single gravity corers box corers multiple corers Benthos, Alpine corers-single gravity corers Piston corers

0-30 cm 0-50cm

0-2 m Deeper than 2m

18.2.1 Grab Samplers Essential components of grab samplers are a messenger weight, a line and a sampler. Two main types exist: one type consists of a set of jaws which when lowered to the surface mud will shut as a result of the messenger weight acting as a trigger mechanism. The second type consists of a bucket which rotates into the sediment when it reaches the mud surface as a result of a trigger mechanism. There are a number of disadvantages to the use of grabs as sampling tools. The first is that the depth to which they penetrate is uncontrolable and unrealiable as this depth is determined by the physical nature of the sediments, the speed with which the sampler strikes the surface mud and the weight of the sampler. Secondly, depending on the nature and size of the particles that make up the surface mud, there is a loss of material as the sampler and the sample are drawn up through the water column. This is especially true of fine-grained sediments. In the process of lowering the sampler through the water column there is always the chance of creating currents which disturb the recently sedimented particles. There is always the danger that the sampling tool contaminates the sample. Finally, it is quite difficult to remove the sample from the grab without further disturbing it. The best way to approach these problems is to gather a large number of subsamples and amalgamate them into a single composite sample rather than to collect one sample and have it represent the whole site. The choice of sampler depends upon the desired depth, the volume of mud to be sampled, the nature of the mud to be sampled, the size of the water body to be

3 50

U . M. C'ou,gill

sampled, presence or absence of currents, the sampling platform (floating raft, boat, bridge, etc.), and the availability of lift equipment (Horowitz, 1991). Ponar and Van Veen samplers arc best for areas where large samples are desired, while Shipek, Ponar and Birge-Ekman are more commonly used in lakes and in slow-moving waters. The Peterson grab is most effective for bulk samples of mud, silt, sand and gravel (Sly, 1969). Shipek grab does poorly in cohesive sediments but is often used for soft clays, mud, silt, sand and gravel (Sly, 1969). The samples taken with this grab are often disturbed and arrive at the surface upside down from the way they were originally collected. Grabs should be closed after the sample has been collected to avoid loss of water and material and they should be stable. Generally, grabs d o well in weak currents and poorly in strong ones. Light weight samplers such as Birge-Ekman behave poorly in deep water with high currents and are much less stable than the heavier Van Veen grab. Tab. 2 shows the area sampled, the weight of the sampler sample, the volume the sampler sampled and the required lifting capacity for the Birge-Ekman, Ponar, Petersen and Van Veen grabs. Before discussing specific grab samplers, it is important to mention the comments of Hiikanson and Jannsson (1983) concerning the ideal sediment sampler. They state that an ideal sampler should permit water to pass through it as it descends through the water column so that pressure waves may be avoided when the sampler reaches the surface mud. I t should produce a minimal amount of friction, compaction and deformation of the sediment. To achieve this, the wall of the sampler must have a small thickness compared to the sample area and the inside surface of the sampler should be smooth. The cutting portion should be sharp and have a small edge angle. After sampling, thc top and the bottom of the sampler should close in situ. To record major features and any apparent stratification, it is most convenient if one side of the sampler is transparent. It should be able to take large samples, not be too heavy so that a winch is not necessary, be easy to operate so that lengthy instruction for use is unnecessary and finally this ideal sediment sampler should be adjustable in terms of weight so that it may be used on different substrata. Birgr-Ekman Grab. This grab is available in several sizes. It may be operated manually although the larger model requires a winch. It works wcll in sediments

+

~

Tab. 2.

Basic Information on VariousTypesofGrab Samplcrs (Murdochand MacKnight, 1991)

Grab

Birge-Ekman petite standard Ponar standard Petersen Van Vccii standard large

Sampled area (cm)

+ scdimciit

Wcight

Sample volumc

(kg)

(an3)

Needed lifting capacity (kg)

1 5 x 15 2 3 x 23

10- 15

3 400 13 300

I00

2 3 x 23 3 0 x 30

23

7250

35

9 450

150-250

3 5 x 70 50 x 100

30 65

18 L 15 L

150 -400 300 - 800

40

50

100

Sumpling of Luke Sediments

35 I

that are fine-grained, soft, and a mixture of silt and sand. Care must be taken to avoid sediments that contain or support high numbers of large objects such as gravel, shells, or pieces of wood as such objects may interfere with the closing of the jaws and thus result in the loss of some or all of the subsample. In very soft sediments, this sampler, due to its weight may penetrate the mud too deeply. This situation may be partly avoided by lowering the sampler slowly and releasing the messenger about the time the sampler has reached the mud surface. Once the sampler jaws have closed as a result of the messenger activating the trigger mechanism the sampler may be retreaved. If it is desirable to divide the sample into a number of subsamples, the sediment may be removed from the top of the sample. From a statistical viewpoint, it would be wiser to take a series of subsamples and amalgamate them into a composite. These may then be removed through the bottom of the sampler into a container. This then is a bulk surface sediment sample. The Ekman grab has been modified by Hikanson (1986) such that there is an automatic closing mechanism on a detachable trigger assembly, a n inside liner made of plexiglass that has a lid that closes and tight-fitting jaws. - Ponar Grub. Unlike the Birge-Ekman grab, which is unsuitable for use with coarse sediments, the Ponar is successful in extracting subsamples from such terrain as well as from muds rich in shells. This sampler is suitable for most sediment types ranging from soft, fine-grained to firm sandy muds. It does poorly in hard clays. This grab sampler is heavier than the Birge-Ekman and thus functions more efficiently in fast moving waters. The ponar is too heavy to use manually and thus it is lowered into the water by use of a winch. It consists of a pair of jaws weighted and tapered that are held open by a catch bar across the top of the sampler. This device is activated by the release of rope tension on the lifting mechanism once the sampler has reached the mud surface. During retrieval the tension on the rope keeps the jaws closed. Hikanson (1982) has modified the Ponar by lining the metal sampler with plexiglass and placing a closure mechanism on the top. - Petersen Grub. This sampler is ideal for collecting large bulk samples from hard bottomed materials such as sand, mud, gravel and hard clay. More weights may be added to increase penetration into hard clay sediments. It consists of a pair of weighted semicylindrical jaws held open by a catch bar. When the sampler has arrived a t the mud surface, the tension on the bar is released permitting the jaws to shut. For deeper penetration into hard sediment extra weights may be added to the jaws. On retrieval, the sampler may be removed and placed in an appropriate container by opening the sampler jaws. This sampler requires an electrical winch (cf. Tab. 2) as it is too heavy to operate manually. - Van Veen Grub. This sampler is among the largest of its kind and is quite heavy and funtions efficiently in deep water and strong currents. Its biggest advantage is that it has a strong closing mechanism which allows the jaws to collect relatively undisturbed sediments. This sampler can be lined with plexiglass or any other suitable material to avoid sample contamination. It is used to collect bulk samples of indurated or soft clay, mud, silt, sand, and gravel (Sly, 1969).

18.2.2 Corers Corers are samplers that take cross sections of sediment at varying depths. These instruments are used for the investigation of historical inputs of pollutants to aquatic systems, paleolimnological investigation as well as geological survey. Corers consist of a hollow metal tube, which when the object of the study is metal distribution, are lined with a plastic liner, that vary in length and diameter. As a rule, if more than 1 m of sediment is desired, a casing is necessary. The sample is extracted from the mud, and then the casing is pushed to the deepest level from which the sample was extracted. Tab. 3 shows the weight of sediment in relation to the diameter of the coring tube. The object of coring is to take an undisturbed sample of the sediment. At the end of the barrel a cutter is mounted which is used to penetrate the deeper sediment. Inside this cutter, a core catcher is placed to prevent the loss of sediment on retrieval. Once the sediment has been retrieved, the top and the bottom of the tube are identified and a clean stopper, usually rubber or neoprene, is placed in the top and in the bottom of the tube. Since the sediment will be around 4 "C when extracted, it should be kept cool but not cold until subsampling is possible. Freezing should be avoided. As a rule, core tubes should be about 0 . 3 m in length and 3.5 cm in diameter. If tubes are much longer than this, slumping of the sediment may occur and the stratigraphy of the sample will be destroyed. - H a i d Corers. These are corers that are suitable for the collection of sediment in the littoral zones of lakes. They may be purchased with metal or plastic tubes with a diameter of 3.5 to 7.5 cm (inside diameter) and extension handles on the top end for driving the corer into the mud. The weight of these devices varies from 5 to 17 kg and the extcnsion handles add another 4 to 12 kg. - Phleger Corer. This corer, which is a single gravity corer can be operated manually but the logistics of a raft or boat requires the presence of two people. Core tubes of this corer are 3.5 cm (inside diameter) and 50 cm in length. The Phleger corer does well in peat and soft to sandy sediments. Its biggest drawback Tab. 3. Estimated Weight of Dried Top 1 cm Mud Layer Subsampled from Core Liners of Different Diametcrs (Murdoch and MacKnight, 1991)

-

Weight of dry mud in top 1 cm sediment layer"

Soft fine-grained mudb (g)

3.5 5.1 6.6 10.0

0.71.52.65.9-

1.4 3.0 5.2 11.8

Firmer silty clay" (g)

1.1 - 2.2 2.3- 4.6 3.2- 7.9 8.R - 17.7

Based on 90-95%1 scdimcnt water content. Bascd on specific gravity of mud 1.5 (high organic matter content). Bascd on specific gravity of mud 2.3 (low organic matter content).

Sumpling

of

Luke Sediments

353

is that it tends to compact the sediment. If the material to be collected must be dated with lead, caesium or carbon fourteen then samples of the mud must be collected at relatively close intervals. Compaction will provide erroneous dating results. When this gravity corer arrives at the mud surface, the water pressure causes the stopper at the upper end of the coring tube to move up and sediment begins to fill the core tube. The pressure is maintained in the tube by the stopper as it slides back into position thus retaining the sediment sample for retrieval. This corer is not recommended if knowledge of stratigraphy is important and if dating of the mud is desired. For further discussion of this corer the reader is referred to Murdoch and MacKnight (1991). - Kujuk-Brinkhurst Corer. This gravity corer utilizes core tubes that are 5 cm (inside diameter) and 70cm in length. There is an automatic trigger mechanism which permits the user to select the time to close the valve which then creates a partial vacuum in the core tube. This partial vacuum assures the retention of the sediment in the core tube during retrieval. Although some authors (cf. Murdoch and MacKnight, 1991) suggest that this corer can be used manually, it is the experience of this writer that the corer + sediment is too heavy and the use of a winch is highly recommended. - Benthos Gravity Corer. The Benthos gravity corer is able to recover 3 m cores from soft fine-grained mud. It has stabilizing fins situated near the top of the instrument which assist vertical penetration into the mud. The auto-valve is open during free fall and sediment penetration. The plunger is pushed into its machined seat and thus a vacuum is developed which keeps the mud in the tube on retrieval. Corers may be purchased with inside diameters varying from 6.6 cm to 7.1 cm. The required lifting capacity varies from 750 kg to 1000 kg (Murdoch and MacKnight, 1991). - Alpine Gravity Corer. This corer is finless and thus tends to land on the sediment surface at an angle. It is winch operated. Its use is not recommended (Sly, 1969). For further and more detailed discussion on other corers, the reader is referred to reviews on the subject (Hopkins, 1964; Wright et al., 1965; Bouma, 1969; Sly, 1969; Mawhinney and Bisutti, 1987). - Box Corers. Box corers are gravity corers that collect large rectangular cores from the mud. Their main use is for geological and biological studies although such samples may be used in chemical surveys. They are versatile in that they can function in many sediment types and they can be made to extract subsamples at a variety of depths. The bottom of this corer can be designed as an Ekman type grab sampler or as a shovel-like device that slides under the box corer (Reineck sampler). These corers are quite heavy, weighing about 800 kg and require a lifting capacity of 2000 to 3000 kg. Their use requires a vessel capable of this lifting capacity as well as sufficient deck space to handle the samples. These corers can not be used in gravel-like muds or i n sediments that contain a lower layer of gravel since they can be damaged. Commercially available box corers can be 2 m by 2 m in size. For further discussion of this corer, see Mawhinney and Bisutti (1987). - Piston Corers. Piston corers, if carefully used, can extract cores of mud up to 30 m. They do the best job of removing sediment with a minimum amount of

354

I l . M . Cowgill

disturbance. Once the piston is at the mud surface, a partial vacuum is created and the sediment is slowly sucked into the core tube. Any sampling much greater than 1 in requires casing. The best description of the use of a piston corer may be found in Deevey (1965). Further discussion may be found in Wright et al. (1984). - Multiple Corers. The biggest single problem encountered with most coring is that the sample size is too small. One way around this problem is the use of multiple corers. This permits the extraction of several cores taken simultaneously. In this sampling tool (cf. Hamilton et al., 1970) usually four tubes, about 0.3 m in length each, are lowered to the mud surface and a messenger is released along a line from the surface which causes soft half balls to seal the top ends of each of the coring tubes. These tubes can be made transparent so that it is possible to view any stratigraphic changes and surface sediment is not lost during retrieval. It is possible to calculate the variance within a sampling station and all samples are extracted simultaneously. Since the tubes are open from both ends as they fall through the water column, the pressure waves set up are minimal. This instrument can not be used in gravel or organic debris. It is clear that there is no one sediment sampler that works effectively on all sediment types. Which sampler to select depends on the purpose of sampling. If it is for regulatory purposes, as a rule only the top few centimeters are of interest. It is also necessary to know how much materials is needed as this will govern the type of grab sampler needed. Tab. 4 provides an overview of the variation in surface sediments to be found in the USA. It is necessary to have some knowledge of the sediment terrain and if this terrain supports large shells, gravel of various sizes and sand or heavy clay. Somc attempt, therefore, should be made to gather all available information on the area to be sampled. It is mandatory to know the water depth of the area to be sampled and whether the current is fast or slow. Physical characteristics of the sediment may bc obtaincd by use of a sediment penetrometer (Hiikanson and Jansson, 1983). The penetrometer may have a variety of cores. The depth to which each core is able to penetrate reveals the nature of the sediments. A calibration table is supplied with each penetrometer purchased so that the depth each core has achieved may be converted to a variety of sediment types varying from very soft mud to very hard mud. Tab. 4. Range of Various Measured Variablcs of Freshwater Mud Samples Actual number of samplescollected was 102 (Suedel and Rodgers, 1991); N = number of replicates measured Minimum

Median

Maximum

N

11.9

+ 379

91.7

0.03

72.1 - 58 0.57

11.8

308 302 30x

'I/n sand

0.1 0 0 2.3

3.1 1.2 18.3 80.6

71 59.1 94.3 100

29 1 306 306 306

Particlc surface area cm'jg

45

96386

4.1 x 10"

306

Variable

' yn solids . Redox potential, mv% organic carbon Cation exchange capacity meq/ 100 g Yo clay 'Yn silt

- 40Y

Sampling of Luke Sediments

355

18.2.3 Sampling Devices for the Collection of Suspended Sediments Suspended sediment samplers may be divided into three general types: integrating samplers, pumping samplers and instantaneous samplers (Horowitz, 199I). The integrating samplers collect a water-mud mixture over time. Instantaneous samplers are usually a flow-through chamber which gather the whole water in the chamber when both ends are closed. Pumping samplers accumulate a whole water sample by pump action (Office of Water Data Coordination, 1982). The integrating samplers are usually the samplers of choice as these are thought to provide a more representative sample (Office of Water Data Coordination, 1982). However, it has been noted by some (Ongley and Blachford, 1982; Horowitz et al., 1989, 1990) that changes in suspended sediments can be large over time. In addition both horizontal and vertical changes in chemistry have been observed (Horowitz et al., 1989). No single sampling technique addresses directional as well as temporal variation. The investigator must decide what information is most crucial to his study and sample accordingly. Horowitz et al. (1989, 1990) have provided detailed discussions on the effect of sampling design, the type of sampler employed, temporal as well as directional variation and the grain size fractional contributions for suspended sediments on the subsequent trace chemical results.

18.2.4 Mud-Water Interface Samplers The best way to take a mud-water interface sampler in shallow water is to use a Naumann (1930) sampler, a device designed for this purpose. It consists of a plexiglass or glass or teflon tube that is beveled on one end. The length of the tube is at the discretion of the investigator. In the non-beveled end a neoprene stopper is fitted. The tube without the stopper is pushed slowly into the mud and when it has reached the desired depth the stopper is inserted and the sampler mud is slowly extracted from the mud. The liquid portion of the mud-water interface sample is removed with a syringe fitted with a glass needle, both fashioned from boron-free glass, if boron is one of the elements of interest. In this case a teflon sampler is desireable. Since the liquid portion of the mud-water interface is only a few millimeters thick, it is necessary to collect a large number of samples from different portions of the water body, not only to try to obtain a representative sample but also to have enough sample for analytical chemistry. These liquid subsamples are amalgamated into suitable bottles and kept cool until analysis is possible. The mud subsamples are extruded and amalgamated into suitable vessels and reserved in a cool environment until analysis is possible. This extrusion of both the liquid and solid portion of the sample may be accomplished in the field. For further discussion of this technique the reader is referred to Cowgill (1 988). The collection of mud-water interface samples from deep waters is best accomplished with a piston corer following the description of Deevey (1965).

+

356

I/. M . Coii~gill

There is a paucity of information of the chemical composition of the mud-water interface. However, it exerts a considerable influence in both shallow aerobic systems as well as deep eutrophic ones (Cowgill, 1988).

18.2.5 Sediment Pore Water Sampling To understand the physical and chemical mechanisms that are responsible for the actual chemical concentrations in mud, an accurate description of the physicochemical characteristics of both the liquid and solid portions of the mud-water interface must be provided (Forstner and Wittman, 1979). There are three methods presently in use to separate pore water or interstitial water or substances dissolved in the pore water from sediment particles (Hesslein, 1976; Mayer, 1976; Robbins and Gustinis, 1976; Brinkmann et al., 1982). It is important to realize that highly organic muds may contain in excess of 90% water but those that are basically minerogenic deposits may range from 30% to 50% in their water content. The three methods are centrifugation, pressure or vacuum filtration and dialysis. The amount of water extracted from the mud is usually no more than 25 to 50% of the total water content. The chemical composition of pore water is highly variable. The stability of its chemical composition is, for the most part, a function of proper handling of the mud prior to extraction and analysis. For example, Krom and Berner (1 980) found that the adsorption of phosphate by anoxic marine sediments is greatly increased when the samples are permitted to become aerobic. Centrifugation is the simplest way of obtaining pore water from muds. When the need for accuracy and reproducibility are low, as in the case of a survey, this is the procedure of choice. Centrifugation does a poor job of removing colloidal and fine particles from the supernatant liquid. It also encourages chemical changes such as oxidation or altered gas equilibria. Removal of colloidal particles is difficult although the larger ones can be separated by filtration. To minimize chemical changes of the mud sample, extrusion from the core tube or subsampling a large grab must be accomplished in a glove box under nitrogen gas. Samples intended for centrifugation should bc collected under nitrogen gas and their centrifuge tubes stoppered. Following such a procedure will certainly minimize chemical changes such as oxidation during centrifugation but i t will not eliminate it. It is important to realize that the speed of centrifugation will have an effect on the chemical composition of the pore water. For example, Adams et al. (1980) centrifuged muds at different speeds (7000 rpm to 19000 rpm) and found that the phosphate doubled but no change was noted in the concentration of Ca, Fe, Mn or Zn. If centrifugation is the method of choice for the extraction of pore water, then the centrifugation unit must be refrigcrated since the process of centrifugation heats the samples thus bringing about chemical changes. Filtration is the most frequently used process to obtain interstitial water. There are several dcvices in common use in laboratories as well as in situ sampling. Pressurc

Sampling of Lake Sediments

351

filtration equipment is usually operated with nitrogen gas, thus avoiding the oxidation of the mud during processing. A mud pore water squeezer (cf. Robbins and Gustinis, 1976) uses a 0.45 pm membrane filter. High pressure acts through a rubber membrane on the mud sample which is enclosed in a removable cassette. The pore water passes first through a coarse pre-filter and then through a fine membrane filter. When the system is made of stainless steel, the determination of anions, alkali metals or alkaline earths in pore water can be accomplished without contamination from the apparatus, but Fe and some transition metals cannot be (cf. Presley et al., 1967). Presley and his coworkers (1967) solved this problem by lining the gas-operated squeezer with teflon. Manheim (1972) used disposable plastic syringes fitted with screen discs and filter paper circles to extract pore water from muds. If organic compounds are of interest the plastic syringes can be replaced with stainless steel syringes. The main disadvantage of laboratory squeezers is the disturbances that occur in the mud during transport from the lake to the laboratory. Brinkman et al., 1982) described a squeezer that could be used in siru. It was designed primarily for use in shallow waters. It is difficult to use and functions poorly in depths of 5 m and greater. A number of difficulties have been noted by investigators who have obtained pore water by laboratory squeezers. For example, iron concentrations were initially high, presumably due to colloidal particles transporting iron as they passed through the filter (Klinkhammer 1980); other workers (Emerson et al., 1980) found higher amounts of ammonium ion in the initial liquid obtained from the squeezer. Higher sulfide values were detected in pore water obtained by in situ interstitial samplers than from laboratory squeezers (Hines et al., 1989). Seasonal variations have been noted (Klump and Martens, 1989) in pore water concentrations for sulfate, phosphate, total inorganic carbon and dissolved ammonium. Mayer (1976) describes an in situ procedure employing dialysis chambers or bags containing distilled water which are placed in the sediments. This system remains in the sediments until the concentration in the bag is the same as in the pore water of the sediments. The “extracted” pore water is free of colloidal particles and is very pure. In fact, this method provides the purest pore water sample of any of the methods yet described. However, there are disadvantages. To achieve equilibrium is time consuming. Most sediments in water as deep as 5 m or more remain close to 4 “C much of the year and thus attaining equilibrium takes time. In addition, a curious catfish may cause the dialysis unit some difficulties. In the centrifugation method, samples of sediment must be collected. As was described earlier, the nature of the material found at the bottom of most bodies of water is such that gathering replicate samples in the true statistical sense is not possible. Sampling with subsequent chemical analysis over time will hopefully show a trend if the object of the study is to monitor pollutants. Chemical concentrations in pore water vary as to where the sample is taken, that is to say in situ or transported to the laboratory, how the sample is taken, i.e., what device was employed, the speed at which the pore water was extracted as in the case of centrifugation, and what kinds of ions come out initially in contrast to those that come out at the end of the extraction process. Aside from all these problems it is not possible to collect replicate samples of pore water.

358

U . M . Cowgill

The amount of pore water needed to carry out chemical analysis will require the collection of large amounts of sediment. In the case of sampling for the purpose of establishing trends there may not be much recent sediment available after several years of collection.

18.3 Subsampling of Sediment Grabs and Cores The purpose of sampling will determine the frequency of subsampling. With grab samples, subsamples should be extracted as soon after collection as possible. The surface of the mud should be removed as soon after sampling as possible as it will contain contaminants from the sampler unless the metal sampler has been lined with a noncontaminating liner. Water content of 1 cm3 of mud should be determined immediately, so that all chemical determinations may be expressed on this basis. The simplest way to collect 1 cm3 of mud is to use a chemically clean porcelain spatula and push mud into the lower end of a 10 cm3 glass or plastic pipette (depending upon the chemical analysis of interest) until 1 cm3 has been measured. This 1 cm' of mud is then extruded with a clean glass or plastic rod into a preweighed porcelain crucible. The crucible samples is then weighed and placed in a 110 "C oven for 48 h and then weighed until constant weight is achieved. Sampling for water content should occur at each interval where samples are taken for chemical analysis. Sediment cores should be subsampled as soon as possible after retrieval of the cores. Should this not be possible, the cores must be stored at 4 "C in a humidity controlled room. This storage is possible for several months without any large change in sediment properties occurring. However, if it is the intent to analyze for chemical contaminants then cores should be extruded as soon as possible. The best way to extrude a core is as follows: (1) take a coring tube and cut it in half and clean it thoroughly; (2) remove the stopper from the end of the coring tube containing the mud to be sampled; (3) remove the stopper from the top of the coring tube containing the mud and replace it with a clean stopper the bottom of which is the same diameter as the mud; (4) using one of the coring rods from a piston corer, push slowly and gently, so as not to compact the sediment, against the stopper described in (3). Gradually extrude the core onto the clean halved coring tubc, pulling the core containing the mud down the halved coring tube so that finally the extruded core lies on the clean halved core tube described in (1).Once extruded, samples for water content and chemistry should be taken immediately. In the event that acid volatile sulfides and heavy metals are of interest, the complete procedure of extrusion just described must be carried out in a large glove box or a series of glove boxes evacuated with a very clean source of nitrogen. It should be pointed out that any sediment samples to be analyzed for their lead content should be extruded and subsampled in a nitrogen evacuated glove box. In the case of a grab sample, subsampling must be carried out in a nitrogen evacuated glove box. It has been found by the author that in some cities the laboratory air contained more lead than the samples being analyzed. Collected subsamples should be placed in clean glass vessels, sealed and stored at 4 "C.

+

Sumpling of Luke Sediments

359

18.4 Quality Control It is very difficult to measure sampling accuracy of sediments which are, as a rule, quite heterogeneous. It is possible to take two sets of samples in exactly the same way. The results will show variations due to sampling and subsampling techniques but the heterogeneity of the sediments will still affect the chemical results. Fig. 1 shows the distribution of some elemental concentrations in two cores taken about l 0 0 m apart in 4 m of water in Laguna de Petenxil, Guatemala (Cowgill et al., 1966). The only way that the two cores could be related to each other was through carbon fourteen dating techniques. Baudo et al. (1981) published some data form Lake Mezzola (Northern Italy) showing mean concentrations of elements for sediments and pore water. The 'YO sampling error is always greater for elemental quantities for littoral pore water where n = 32 pairs than for the littoral muds where n = 51 pairs. In some cases the 'YO sampling error is greater than the mean, as is the case for Cr in pore water (cf. Fig. 2). Williams and Pashley (1979) compared the carbon content of two cores, one taken with a piston borer and the other using a scuba diver operating a piston corer and found the results to be similar. Evans and Lasenby (1984) found similar results when they compared a Kajak-Brinkhurst corer with a diver driven one. Rutledge and Fleeger (1988) found that diver operated equipment resulted in a difference between the chemical composition of the outer portion of the core from the center portion because particles tended to concentrate in the center during core transport by the diver.

CI-core 2 S-core 3 S-core 2 K-core 3 ~

K-core 2

L

Na-core 3 c

Na-core 2

0

0.I

0.2

0.3

0.4

mg/cm*/a Fig. 1.

Chemical comparison of two cores from Laguna de Petenxil, Guatemala.

360

I/.

M . (‘owgill

Mn

Ni + t

-5 Zn w

CU

Cr

o

zoo

400

600

eoo

Fig. 2. Elcmental quantities and their YO sampling error (mud: n = 51 ; pore water: n = 32).

Concentration

Core shortening may occur when cores are collected with a gravity corer. Apparently this shortening is related to tube size. In addition when sampling soft sediments it is possible to obtain a core that is not representative of the mud (Blomqvist, 1985). Baudo et al. (1985) compared chemical results from cores taken with a gravity corer and a modified Ekman grab sampler in six locations in Lake Como (Northern Italy). They concluded that the variability in chemical results between cores extracted by the different sampling devices was no greater than the analytical error of the chemical analyses. Based on the work of these authors, it would seem that results from grab samplers and piston corers of the top 10 cm of mud give reasonably concordant results. Sample handling often affects the chemical results. Cores must never be frozen. Freezing and thawing not only destroy the stratigraphy but also tend to homogenize the chemical results. Rutledge and Fleeger (1988) found that fast freezing of cores distorts the sample, although the state that commercial freezing does not create distortion any more than the compaction due to the “drag of the corer wall upon insertion”. Some investigators (Ginsburg et a]., 1966; Bouma, 1969; Crevello et al., 1981) have suggested the use of epoxy or polyester resins as a means of preserving stratigraphy. However, this treatment will distory the chemical results. The best way to handle samples, whether they are grab samples or cores is to keep them at 4 ”C in the field and maintain them at this temperature until subsampling can occur. Finally, great effort must be exerted to avoid contaminating the core or grab sample. Samplers must be cleaned between sample sites. How they are cleaned is determined by the kind of chemical analysis intended. Most sampling devices can be fitted with liners that will avoid contamination. However, it should be borne in

Sampling of Lake Sediments

36 1

mind that water loss can occur through the walls of some liners. Under these circumstances, subsampling should occur on arrival at the laboratory. Blanks should be taken during sediment sampling. These are equipment, field and sampling blanks. The sampling blank is extremely important as the chemical composition of atmosphere can affect the chemical composition of the mud samples. The sampling blanks are merely a vessel of double distilled (glass) water exposed to the atmosphere during the entire sampling period. If sampling proceeds for days, then a new sampling blank should be exposed each day.

18.5 Statistical Considerations Sampling design is important to obtaining acceptable results. The design is governed by the purpose of sampling. If the purpose is paleolimnological, several cores taken in the deepest section of the lake will suffice. Hikanson and Jansson (1983) divided design stratigies into three general types: ( I ) A sampling design that is based on previous knowledge of the study site. This previous information should include geological conditions, hydrological conditions and some chemical and physical results. Hi'ikanson (1986) has suggested the use of a sediment penetrometer to provide information as to the character of surface sediments. (2) A design where sampling stations are randomly selected such that each sample has an equal chance of being chosen. This design is often used in areas where little is known of local conditions and background or baseline data are needed. Another kind of random design depends on a good deal of local information since the strategy depends on selecting heterogeneous areas and dividing these into local homogeneous sites wherein sampling sites are selected at random (Horowitz, 1991). (3) A sampling design based on a regular grid system. This is probably the best though most expensive approach. Many books have been written on planning and design (Watterson and Theobald, 1979; Green, 1979; Gy, 1979; Sanders et al., 1983; Keith, 1988). This contribution is not intended to provide a detailed discussion on planning and design, and therefore the reader is referred to the above noted citations. Green (1 979) has made some very useful suggestions concerning sampling. He recommends that some type of preliminary sampling be carried out to help evaluate a sampling design and develop statistical analysis. In addition, it should be verified that the sample size is sufficiently large to account for the type of particle distribution and physical aspects of the sediment. Then, based on the sample size selected, it will be possible to estimate how many replicates will be needed to obtain the kind of precision desired. For example, Baudo (1989) collected 57 samples of Lake Orta sediments and analyzed them for copper. The data were not distributed normally. They were found to be skewed and leptokuritic. Thus, the negative binomial distribution was used instead of the normal distribution. Hikanson and Jansson (1 983) have stated that six samples of surface mud taken at regular intervals over

362

U . M . Cowgill

the lake surface should approximate the mean value for the muds. Kratochvil and Taylor (1981) state that six samples should be sufficient to estimate the mean with a coefficent of variation of 100% at the Student’s t-test level corresponding to 95% confidence limits. With Baudo’s 57 samples, the coefficient of variation was 25%. If a 10% coefficient of variation is desired, 412 samples would be necessary (cf. Baudo, 1989). Green (1979) points out that variation in the efficiency of sampling from place to place will ultimately bias the chemical results. Further, he suggests that the investigator ascertain that the desired population is in fact the population being sampled. He emphasizes that the data should be tested to ascertain whether the error variation is homogeneous, normally distributed and independent of the mean. For further discussion on statistically sound sampling techniques, the reader is referred to Cochran (1 963) and Green (1 979). An excellent discussion on how to estimate the number of samples required to obtain a particular coefficient of variation at a given confidence limit is given by Baudo et al. (1990). Some of the important points are given below. Frequency distribution of the concentrations of 13 elements were measured by Muntau et al. (1 986). The frequency distributions were calculated according to procedures described by Sokal and Rohlf (1969) and Davis (1971). Of the 13 elements, nine were best described by the negative binomial distribution, one element was best described by the Poisson distribution and the remaining three followed a normal distribution. The point of this study was to emphasize that all elemental concentrations do not exhibit the same frequency distribution. Presumably the explanation for different frequency distributions is due to the association of different elemental concentrations with particle size (cf. Horowitz et al., 1990). HBkanson and Jansson (1983) present an empirical formula from which the required number of samples may be calculated. This formula requires morphometric information, to wit, the shore development, a crude estimate of bottom roughness and the area of the lake. This formula is useful only for large lakes (Evans and Dillon, 1986). Sly (1975) sampled 843 stations following a nested grid pattern as a sampling strategy. Sodium, K, Hg and mean particle size were accurately estimated with a grid spacing of 100 In. He suggested that for shallow lakes a grid spacing of <30 m would be necessary while for deeper lakes as much as 300 m would be adequate.

18.6 References Adams, D. D., Darby, D. A,, Young, R. J . (1989) in: Contaminants and Scdinient.r: Baker. R . A . (cd.), Ann Arbor: Ann Arbor Science Publishers, Vol. 2, pp. 266-282. Baudo, R. (1989) Hplrohiologiu 176/177, 441 -448. Baudo, R., Galanti, G.. Guilizzoni, P., Varini, P. G. (1981) Mern. Ist. Ital. Idrohiol. 39, 177-201. Baudo, K., Galanti, G., Guilizzoni, P., Marcngo, G., Muntau, H., Son, van M., Schramcl, P. (1985) in: Procec~/ingslnternirtionril Confirenw on H w q y Metals in the Enuironment: Athens, Scptcmber; Vol. 1, pp. 252-254.

Sampling of Lake Sediments

363

Baudo, R., Giesy, J. P., Muntau, H. (1990) Sediments: Chemistry and Toxicity ofln-Place Pollutants, Chelsea : Lewis Publishers. Blomqvist, S. (1985) Sedirnentology 32, 605-612. Bouma, A. H. (1969) Methods ,for the Study of Sedimentary Structures, New York: Wiley & Sons. Brinkman, A. G., Raaphorst, W. Van, Likklema, L. (1982) Hydrobiologia 92, 659-663. Cochran, W. G . (1963) Sampling Techniques, New York: Wilcy & Sons. Cowgill, U. M . (1988) in: Aquatic Toxicology und Environmental Fate. Suler 11, G. W., Lewis. M. A. (cds.), Philadelphia: American Society for Testing and Materials, Vol. 11, pp. 72-89. Cowgill, U . M., Hutchinson, G . E., Racek, A. A , , Gouldcn, C. E., Patrick, R., Tsukada, M . (1966) The History ofLagunade Petenxil, New Haven: Conn. Acad. Arts& Sciences. Vol. 17, pp. 1 - 126. Crevello, P. D., Rine, J. M., Lanesky, D. E. (1981) J . Sed. Petrol. 51, 658-660. Davis, R. G. (1971) Computer programming in Quantitative Biology, London: Academic Press. Deevey, E. S. (1965) in: Handbook qf Paleontological Techniques, Kummel, B., Raup, D. (eds.), San Francisco: Freeman, pp. 521 - 529. Elliott, J. M., Tullett, P. A. (1978) A Bibliography ofSunzplersfor Benthic Znoertehrates, Ambleside: Occ. Publ. Freshwater Biology Assoc., No. 4. Elliott, J. M., Tullette, P. A. (1983) A Supplement to (I Bibliography of Samplers .for Benthic Invertebrates, Ambleside: Occ. Publ. Freshwater Assoc., No. 20. Emerson, S.. Jahnke, R., Bender, M., Froelich, P., Klinkhammcr, G., Bowser. C., Setlock, G . (1980) Earth Planet. Sci. Lett. 49, 57-64. Evans. R. D., Dillon, P. J . (1986) Paper presented a the EUCHEM Cogfirence on Sampling , RIVM, Bilthoven, The Netherlands, January Strategies crnd Techniques in Environmental Anal 20 - 24. Evans, H. E.. Lasenby, D. C. (1984) Hydrobiologiu 108, 165-169. Forstner, U., Wittman, G . T. W. (1979) Metal Pollution in the Aquatic Environment, Bcrlin: Springer-Verlag. Ginsburg, R . N., Bernard, H. A., Moody, R. A,, Daigle, E. E. (1 966) J . Sed. Petrol. 36, 1 I 18 - I 12 1 Green, R. (1979) Sampling Design and Statisticul Methou’s for Environmental Biologists, New York : Wiley & Sons. Gy, P. (1979) Sampling ef Particulate Materials: Theory and Pructice, New York: Elsevier. HQkanson, L. (1982) A Modified Ponar Grab Sumpler ,for Coarse and Consolidated Sediments. Uppsala: National Swedish Environmental Protection Board. HQkanson, L. (1984) Water Resour. Res. 20, 41 -46. Hikanson, L. (1986) In/. Rev. Ges. Hydrtrhiol. 71, 719-733. Hikanson, L., Jansson, M. (1983) Principles of Lake Sedinwntology, Berlin: Springer-Verlag. Hamilton, A. L., Burton, W., Flanagan, J . F. (1970) J . Fi,sh Rex. Board Can. 27, 1867- 1869. Hesslcin. R. H. (1976) Limnol. Oceanogr. 21, 912-914. Hincs, M. E.. Knoll Meycr, S. L., Tugel, S. B. (1989) Limnol. Oceanogr. 34, 578-590. Hopkins, T. L. (1964) in: Progress in Oceanography. Sears, M. (ed.), New York: Pergamon-MacMillan, Vol. 1 I , pp. 213-256. Horowitz. A. J . (1 99 I ) A Primer on Seditnent-Trace Elivnent Cl?emistry,Chelsea: Lewis Publishers. Horowitz, A. J.. Rinella, F., Lamothe, P. Miller, T., Edwards, T., Roche, R., Rickert, D. (1989) in: Sediment and the Environment. Hadley, R., Ongley, E. (eds.), Washington, D.C.: IAHS Publication. No. 184, pp. 57-66. Horowitz, A. J., Rinella, F., Lamothc, P., Miller, T., Edwards, T., Roche, R., Rickert, D. (1990) En~iron.Sci. Technol. 24, 13 13 - 1320. Keith, L. (1988) Principles UJ’ Ent~ironmentalSumpling. Washington, D.C.: American Chemical Society. Keith, L., Crummett, W., Dcegan, J., Libby, R.. Taylor. J. Wcntler, G. (1983) Anal. Clzem. 55, 2210 -221 8. Klinkhammer, G. T. (1980) Burth Pluner Si.Lett. 49, 81 - 89. Klump, J. Val, Martens, C. S. ( I 989) Limnol. Oceanogr. 34, 559 - 577. Kratochvil, B., Taylor, J . K . (1981) A n d . Chem. 53, 924A-938A. Krom, M. D., Bcrner, R. A. (1980) Limnol. Oceanogr. 25, 327-330. Manhcim, F. T. (1972) J . Srd. Pctrol. 38. 666-670.

3 64

U. M . Cowgill

Mawhinncy, M. R., Bisutti, C. (1987) Cominon Corers and Grub Samplers: Operating Munuul, Burlington: National Water Institute. Environment Canada. Maycr, L. M. (1976) Lirn~iol.Oceimogr. 21, 909-912. Muntau, H.. Son, M. Van, Baudo, R., Schramel, P., Marengo, G., Lattanzio, A,, Amantini, L. (1986) Papcr presented at the BUC‘HEM Conference on Sumpling StratqicJs und Techniques in finnz;ironmenttrl Anrr/ysi.c, RIVM, Bilthoven, Thc Netherlands, January 20-24. Murdoch, A,, MacKnight, S. D. (1991) C‘RC‘ Himdbook qf Techniques ,for Ayuutic Sediment Smnipling, Boca Raton: CRC Press. Naumann, E. (1930) Die Bbzneng~~~~u.s.ser 9, I - 126. Office of Water Data Coordination (1982) Nutionul Handbook of’ Recommended Metkods ,fbr Wur(~r-DatuAcquisition, Reston: U.S. Geological Survey. Ongley, E., Blachford. N. (1982) Eniirun. Tdznol. Let[. 3, 219-228. Plumb, R., Jr. (1981) Technical Report EPA/CE-81-1, Environmental Laboratory, U S . Army Engineers, Waterways Experiment Station, Vicksburg, MS, USA. Prcsley, B. J., Brooks, R. R., Kappel, H. M. (1967) 1. Mur. R a . 23, 355-362. Robbe. D. (1 98 1) Pollutions Metulliques d ~ Milieu i Nuturel. Guide Methodologiquc de Leur Etude ( I Partir des Srrlimrnts. Rupport Bibliogruphique, Paris: Ministcre de I’Urbanisme et du Logemcnt, Rapport de Recherche, LPN 104. Robbins. J. A., Gustinis, L. (1976) Lirnnol. Oceunogr. 21. 905-909. Rutledgc, P. A., Fleegcr, J . W. (1988) Lirnnol. Oceunogr. 33, 274-280. Sanders, T., Ward, R., Loftis, J., Steele, T., Adrian. D., Yevjevich, V. (1983) Design o/ Netwrk.s f.r mu nit or in^ Wutor Qiiulity. Littlcton: Water Resources Publications. Sly, P. (1969) in : Procecdings oftlie 12th Conference on G r m Lukes Rcseurch. Ann Arbor: University of Michigan Press, pp. 883-898. Sly, P. G. (1975) in: I X Internir/ionul Congress qf’ Sedimentology. Nicc, France. Sokal. R. R., Rohlf, F. J. (1969) Biometry. The Principles and Pructice uf Stutistics in Biologicrrl Rrwwrcli. San Francisco: Freeman. Suedel, R. C.. Rodgcrs, J. H.. Jr. (1991) Wut. Res. BLIII.27, 101 - 109. Watterson, J., Theobald, P. (1979) Geochetnical Exploration 1978: Proceedings of’ the 7th Intcmutionul Geocliemicul Esploration Symposium,Toronto: Association of Exploration Geochemists. Williams, J . D. H., Pashley, A. E. (1979) ./. Fish. RPs. Board Can. 36, 241 -246. Wright, ti. E., Cushing, E. J., Livingstone, D. A. (1965) in: Handbook of’Puleontolugicul TechniqLlcx Kummcl, B., Raup, D. (eds.), San Francisco: Freeman, pp. 494-520. Wright, H. E., Mann, D. H. Glasser, P. H. (1984) Ecology 65, 657-670.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

19 Sampling Design for Studying the Relationships between Heavy Metals in Soils, Sediments, and Discharged Wastewaters Zueng-Sang Chen

19.1 Introduction Contamination of soils can occur when irrigated with polluted wastewater discharged from industrial parks and municipalities such as sewage effluents (Chen et al., 1992; Wang et al., 1992). To avoid agricultural soils being contaminated by polluted discharged wastewater from industries, monitoring the quality of the water must be continued. When highly contaminated water were introduced into the irrigation system, then the extent of pollution will become more widespread and will be of increasing social and environmental concern as is now happening in Asia in the last one or two decades (Chen, 1991). In Taiwan, agricultural soils polluted with different heavy metals have been significantly related to pollution sources, especially near industrial parks (Chen, 1991; Chen et al., 1992; Wang and Wang, 1992). In order to understand the distribution of the pollution regions of heavy metals, a comprehensive sampling design is needed for saving time and budget in order to explain the distribution of heavy metals in the environment. The objectives of this contribution are (1) to discuss the principles of sampling design and sampling methods and (2) to give examples for sampling design for studying the relationships between heavy metals in soils, sediments and discharged wastewaters.

19.2 Case I: Studies of Rice-Growing Soils near Chemical Plants 19.2.1 Principles of Sampling Design Pollution of irrigation water by the discharged water from chemical plants will probably cause pollution of the adjacent agricultural soils by the metals in the water. Fig. 1 shows the distribution of Cd and Pb contaminated regions of the rice growing soils in northern Taiwan. This is a famous case of Cd and Pb pollution by a chemical plant producing plastic stabilizing materials. The total area in case I study was only 13 hectares.

366

(a 1

Z . 3 . Chen Case

I area

3

\ \

Pollution region

0

500m

Fig. 1 a, b. Case I area is located in northern Taiwan (a). The area was polluted from the irrigation water discharged from the chemical plant. The pollution region is located alongside the irrigation river. The total pollution area covers about 80 hectares. The area of case I region discussed in this contribution covers only I3 ha and is shown in a black color block (b). Survey area

=

13 ha

6

5

I 4

3

2

C’ Fig. 2. Distribution of sampling locations of ricc growing soils, water, and sediments in case I area; ( 0 )sampling site for soils (n = 1 IX), (w) sampling site for irrigation water and sediments (n = 20). (-P) flow dircction of irrigation water.

Heuuy Metals in Soils, Sediments, and Wustewuters

367

The objectives of sampling design were to understand : (1) the distribution of heavy metals in the agricultural soils irrigated with the discharged water, (2) the vertical distribution of metals in the polluted rice-growing soils as a function of distance from the outfall of irrigation water.

19.2.2 Sampling Methods 19.2.2.1 Soil Sampling Three kinds of sampling methods in case I region were conducted and listed as following: (1) Sampling every field in the polluted region, the total sampling number is 1 18 (Fig. 2). The sampling depths were 0 to 15 cm and 15 to 30 cm, respectively. The soil was sampled with an 8 cm diameter auger. (2) Random sampling of polluted soils was done, especially adjacent to the irrigation river or water. The total of soil samples was only 28 (Fig. 3). The soil sampling depths were the same as in (1).

0

T e 5

0 C

Fig. 3. Locations of second random sampling in case I area; ( 0 )sampling location for soils, (-P) flow direction of irrigation water.

368

Z . 4 . Chen

(3) Sampling from 0 to 100 cm depth at every 10 cm. The location was at a distance of 6, 20, and 35 meters from the entrance of the irrigation water, respectively. Every profile has 10 soil samples in the polluted region.

19.2.2.2 Sampling Design for Water and Sediments Discharged water from the chemical plant and sediments in the river were sampled at the outfall of the chemical plant and every 70 meters from the outfall. The total sample number of irrigation waters and sediments was 20. The location of the sampling sites is shown in Fig. 2.

19.2.3 Analytical Methods The concentrations of heavy metals in soil samples and river sediments were determined by extraction with 0.1 N HCI for one hour. This extractable concentration was regarded as the quantity of metals bioavailable to plants in Taiwan (EPA-ROC, 1991). For extraction, 100 mL of 0.1 N HCI were added to 10 g of soil and shaken for one hour at 180 rpm, then filter with Whatman No. 42 filter paper. Concentrations of heavy metals (Cd and Pb) were determined with atomic absorption spectroscopy (Hitachi 180-30 type). Water samples were filtered with Whatman No. 42 filter paper and concentrations of heavy metals determined as described above.

19.2.4 The Distribution of Heavy Metals in Soils, Discharged Water, and Sediments The concentrations of heavy metals (Cd, Pb, Cu, Zn, Cr, and Ni) in agricultural soils of case study I obtained with different sampling methods are shown in Tab. 1. The sample distribution with different concentrations of heavy metals in surface soils (0 - 15 cm) are shown in Tab. 2. Case I region was a cadmium polluted area. Results indicate that 0.1 N HCI extractable cadmium in the surface soil (0- 15 cm) in case I region ranged from 1.12 to 148 mg/kg. The mean and standard deviation of the Cd concentration is 16.35 mg/kg and 22.93 mg/kg, respectively. The distribution of extractable Cd in case I region also indicates that the area adjacent to the irrigation river had very high Cd concentrations which were even higher than 50 mg/kg. Thc distribution of three levels of Cd concentrations ( < 3, 3 to 10, and > 10 mg/kg) of agricultural surface soils are shown in Fig. 4.

Heavy Metals in Soils, Sediments, and Wastewaters

369

Tab. 1. The Concentrations (mg/kg) of Heavy Metals in Soils (0- 15 cm) of Case I Area Sampling number"

n

=

118

n = 28

Element

Mean f Standard deviation

Cd Pb cu Zn Cr Ni

16.35 f 22.93 25.78 k 22.55 11.93 f 47.46 42.71 & 54.06 1.02 f 1.02 2.40 f 1.75

1.12 6.34 3.06 0.39 NDb 0.39

Cd Pb cu Zn Cr Ni

12.36 f 25.45 23.36 f 21.36 15.68 f 35.87 36.15 k 46.37 1.38 k 1.56 1.83 f 1.68

2.34 130 7.54 118 6.75 315 1.26 325 N D ~ 5.45 0.46 8.46

Rangc

-------

148 126 509 361 6.30 9.39

n = I 18 for sampling every field in the polluted region; n = 28 for random sampling of polluted soils. Non-detectable.

The concentrations of 0.1 N HCl extractable Cd in discharged water and sediments in case I region are shown in Tab. 3. The results indicate that the Cd concentrations in discharged water ranged from 2.7 to 13.5mg/L, and concentrations varied according to the season. Tab. 3 also indicates that the Cd concentrations in the sediments ranged from 120 to 560 mg/kg. The vertical distributions of Cd in polluted soils are shown in Fig. 5. The Cd concentrations in the surficial 20 cm of soil at the 6 m, 20 m, and 35 m distance from the outfall of the irrigation water were 26.18, 36.42, and 8.78 mg/kg, respectively. The Cd concentrations in soil layers below 20 cm depth were less than 0.5 mg/kg (Fig. 5). The results in Fig. 5 indicate that (1) the highest Cd concentraTab. 2. Proportions of Differing Levels of Metals in Soils (0- 15 cm) of Case I Area Percentage of total samplcs (n = 118)

Element Cd

10 30

Pb

20 40

Zn

Cu

--

30 50 > 50

-- 12040 > 120 30 - 80 80 - 120 21

12 6 1 14 6 1

> 120

12 12 4

100 > 100

8 1

-

t

t

Fig. 4. Distribution of thrcc lcvcls of cadmium conccntrations in the case I area. Very high concentrations of cadmium ;( 10 mg/kg) are mainly located alongside thc irrig,ation rivcr.

Tab. 3. The Concentrations of Cadmium and Lcad in Irrigation Water (mg/L) and Sediments (mg/kg) of Case I Area Location number

Sampling number

Mcan

STD"

Cd b

7 7

3.5 & 2.1 7.8 f 3.2

Scdiments 1- 6 7-13 14-20

6 I I

32.5 _+ 12.1 165 & 65 240 f 80

I-

STD: standard deviation.

Pb

Cd

Pb

**+ 0.81.2

3.5 - 13.5 2.7- 5.4 5.3 - 12.5

U. 1 - 3.6

650 _+ 350 210 & 150 325 250

175 - 560 I20 - 213 158-325

1 30 1 240 65- 370 85- 560

1.K

0

7- 13 14-20

Range

1.2 1.4

1.0

+

0 , 1 2.4 ~

0.1 - 2.7

~

Heavy Metals in Soils, Sediments, and Wastewaters

Fig.5a -d.

371

Three-dimensional distribution of HC1-extractable Cu, Zn, Cd, and P b in case 1 region.

tions were usually highest in the surficial20 cm of soil, and (2) that Cd in the water can be transported and accumulated in soil to about the 2 0 m distance from the irrigation outfall, and decreases to lower concentrations in the soil surface of the polluted soils.

19.2.5 The Relationship between Heavy Metals in Soils, Discharged Water, and Sediments In case 1 region, the distribution of cadmium in agricultural soils was significantly related to the Cd concentrations in the polluted river and sediments. If the farmers used the polluted water for irrigation, the rice-growing soil could be polluted by it.

312

Z . - S . Clien

In areas close to the irrigation river and as far out as 3 0 m from the river, soils could accumulate very high concentrations of Cd in the surficial 20 cm. From the sampling design, we can conclude that the concentrations of heavy metals in the soils were related to the flow pathway, direction, and distance of polluted irrigation water from the chemical plant.

19.3 Case 11: Studies of Rice-Growing Soils near an Industrial Park 19.3.1 Principles of Sampling Design As in case I region, it was presumed that the irrigation water was polluted by an industrial park. Thus, agricultural soils adjacent to it or receiving the discharged water for irrigation could be highly polluted by heavy metals. Fig. 6 shows the

Fig. 6. Distribution of sampling locations of rice growing soils, water, and sediments in case I1 arca, Shiang-Shang Industrial Park; (0) sampling site for soils (n = 28), ( 0 ) sampling site for irrigation watcr and sediments (n = Is), (+) flow direction of irrigation water, (A)sampling site for background samples (n = S ) , (m) chemical plants of the industrial park.

Heavy Metals in Soils, Sediments, and Wastewuter:y

373

distribution of sampling locations of rice-growing soils, irrigation water, and sediments in case I1 region contaminated by the discharged water from Shiang-Shang Industrial Park in northern Taiwan. This is also a famous industrial park in northern Taiwan comprising several chemical plants. Facilities include heavy metal coating, treatment of leather, production of glasses, dyeing of clothes, production of stabilizing materials for plastics, etc. The survey area covered about 112 hectares. The objectives of sampling design were: (1) to survey the distribution of heavy metals in agricultural soils, discharged water, and sediments, (2) to study the relationship between the concentration of heavy metals in soils, water, and sediments in case I1 region.

19.3.2 Sampling Methods 19.3.2.1 Soil Sampling Adjacent to the river polluted by the discharged water from the industrial park, a 4 ha area (20Ox200m) was proposed as a sampling site. In this sampling area, 10 soil samples were randomly collected and homogenized to one representative soil sample. The total number of representative soil samples was 28. There were only 5 samples for the control area (unpolluted region), as shown in Fig. 6 . The sampling depths were 0 to 15 cm and 15 to 30 cm, respectively. The soil was sampled with an 8 cm diameter auger.

19.3.2.2 Sampling Design of Water and Sediments Discharged water from the Shiang-Shang Industrial Park and sediments in the river were sampled at locations where the river just changes its flow direction (see Fig. 6). These are the sections of the river where heavy metals will be accumulated in the sediments.

19.3.3 Analytical Methods The concentrations of heavy metals in soil samples and sediments were determined by extraction with 0.1 N HC1 for one hour and this concentration was regarded as the quantity of plant bioavailable metals (EPA-ROC, 1991).For extraction, 100 mL 0.1 N HCl were added to 10 g of soil samples or sediments and shaken for one hour with 180 rpm, then filter with Whatman No. 42 filter paper, and concentrations of metals (Cd, Cr, Cu, Ni, Pb, and Zn) were determined with flame atomic absorption spectroscopy (Hitachi 180-30 type).

314

Z.-S. Chen

The water samples of the discharged water were filtered with Whatman No. 42 filter paper, and concentrations of metals (Cd, Cr, Cu, Ni, Pb, and Zn) were determined as well with flame atomic absorption spectroscopy (Hitachi 180-30 type).

19.3.4 The Distribution of Heavy Metals in Soils, Discharged Water, and Sediments Tab. 4 shows the concentrations of heavy metals in agricultural soils in case I1 rcgion (Shiang-Shang Industrial Park). The results indicate that the 0.1 N HCI extractable metals in the surface soils (0- 15 cm) ranged from 11.3 to 60.3 mg/kg for Cr, 22.3 to 158 mg/kg for Cu, 11.1 to 122 mg/kg for Ni, and 35.0 to 214 mg/kg for Zn, respectively. Most of thc soils in the survey area showed much higher concentrations of Cr, Cu, Ni, and Zn in the surface soils than the background concentrations in these soils. Background concentrations of 0.1 N HCI extractable Cr, Cu, Ni, and Zn in uncontaminated soils were less than 5, 20, 10, and 25 mg/kg, respectively. Eleven sampling sites (44 ha) of agricultural soils in Fig. 6 adjacent to the discharged water show that the concentrations of heavy metals in agricultural soils were very high, with concentrations higher than 60, 125, 60, and 80 mg/kg, for Cr, Cu, Ni, and Zn, respectively. These also indicate that virtually the whole sampling area of 44 ha was considerably contaminated by the discharged water and sediments from the industrial park. Tab. 4. The Concentrations (mg/kg) of Heavy Metals in Soils of Case I1 Area Element"

Surface soil (0- 15 cm) (n = 26)' Conc. in high and very high conc. level =

Subsurface soil (1 5 - 30 cm) (n = 26)'

Sample (YO) Sample (Yo) in high in very conc. high levcl conc. level

Conc. in high and very high conc. level ~~

Cd Cr CU Ni Pb Zn As

Hg

0.4 1 11.3 60.3 22.3 158 11.1 122 24.1 42.8 35.0 214 0.46 0.66

----

4 15

50 69 19 15

0 12

~

0 4 8 4 0 42 0

14.6 21.2 31.2 12.1 63.4

0

-

--

34.1

Sample (YO) Sample (YO) in very in high conc. high level conc. level

-

140

~

0 4 19

46 0 35 0 0

~

0 0 0 0 0 4 0 0

As and Hg: total contents; Cd, Cr, Cu, Ni, Pb, Zn are 0.1 N HC1 extractable. Sample number: 26; every sample represents 4 ha (200 m x 200 m). High concentration level ofheavymetals:0.40-10, 11-16, 21-100, 11-100, 16-120,26-80, 16-60, and 0.4-20 mg/kg, for Cd, Cr, Cu, Ni, Pb, Zn, As and Hg, respectively. Very high concentration level of heavy metals: > 10, > 16, > 100, > 100, > 120, >80, > 60, and > 20 ing/kg, for Cd, Cr, Cu, Ni, Pb, Zn, As and Hg, respcctively.

Heavy Metals in Soils,Sediments, and Wastewaters

315

Tab. 5. The Concentrations (mg/L) of Heavy Metals in Irrigation Water of Case 11 Area

Element

Cd Cr

cu Ni Pb Zn As Hg

Range (n = 18);l

Sample (YO)of conc. in water higher than the criteria of irrigation water

---

0 16 6 44 0 0 6 0

ND ND 1.60 N D 0.20 0.13 2.89 ND 1.22 ND 0.30 1.95 ND

Sample number. Non-detectable.

The concentrations of 0.1 N HCl extractable metals in discharged water in case

I1 region are shown in Tab. 5. The results indicate that the concentrations of Cr, Cu, Ni, and Zn in discharged water ranged from N D (non-detectable) to 1.60 mg/L for Cr, N D to 0.20 mg/L for Cu, 0.13 to 2.89 mg/L for Ni, and N D to 1.22 mg/L for Zn, respectively. The concentrations of Cd, Hg, and Pb in the discharged water were considered within the normal range. Tab. 5 also indicates that the concentrations of Cr and Ni were significantly higher than those set for the irrigation water quality proposed by the Government of the Republic of China. The concentrations of 0.1 N HC1 extractable metals in sediments of the river in case I1 region are shown in Tab. 6. The results indicate that the concentrations of metals in the sediments ranged from 0.36 to 24.8 mg/kg for Cd, 40.7 to 787 mg/kg for Cr, 21.7 to 2250 mg/kg for Cu, 57.3 to 3590 mg/kg for Ni, 8.9 to 178 mg/kg for Pb, and 85 to 8710 mg/kg for Zn, respectively. Tab. 6 also indicates that the proportion of samples with very high concentrations of metals correspond to 16 to 50% of the total sediment samples. The critical concentrations of metals in the sediments were 15.4 mg/kg for Cd, 166 mg/kg for Cr, 1160 mg/kg for Cu, 482 mg/kg for Ni, 124 mg/kg for Pb, and 1090 mg/kg for Zn (see Tab. 6). Tab. 6. The Concentrations (mg/kg) of Heavy Metals in River Sediments of Case 11 Area Element

Cd Cr

cu

Ni Pb Zn As Hg

Range (n = 18)b 0.36 40.1 21.7 57.3 8.9 85 1.50 0.28

-----

Samples with very high concentrations Range

24.8 787 2250 3590 178 8710 14.3 1.84

15.4 166 1160 482 124 1090

----

24.8 787 2250 3590 178 8710

Sample (YO) 22 50 28 44 16 50 0 0

As and Hg: total contents; Cd, Cr, Cu, Ni, P b and Zn are 0.1 N HCI extractable. I, Sample number.

316

Z . - S . Chef1

19.3.5 The Relationship between Heavy Metals in Soils, Discharged Water, and Sediments Fig. 7 shows the distribution of heavy metals with high concentrations in agricultural soils and the course of the irrigation system in case I1 region. The concentrations of Cu, Cr, Ni, and Zn were significantly higher than those in the agricultural soils in the neighboring region, regarded as the background concentration. The concentrations of heavy metals in the agricultural soils along the irrigation river were higher than 125 mg/kg for Cu, higher than 122 mg/kg for Ni, and higher than 82.8 mg/kg for Zn, respectively. The proportion of samples with very high concentrations of Zn ( > 80 mg/kg) comprised more than 40%. The following relationships between heavy metals in soils, discharged water, and sediments were observed: ( I ) The high levels of Cr, Cu, Ni, and Zn in the contaminated discharged water were reflected also by their high levels of accumulation in soil and sediments.

Fig. 7. Distribution of heavy metal concentrations in the rice-growing soils in case I1 area. Very high concentrations of heavy metals in soils (@) are mainly located alongside of thc irrigation river, the concentrations of HCI-extractable Cr, Ni, Cu, and Zn in soils are higher than 16, 100, 100, and 80mg/kg, respectively. Medium high concentrations of heavy metals in soils (D) are mainly located at a long distance from the river. the concentration of HCI-extractable Cr, Ni, Cu, and Zn are 1 1 - 16, 11 - 100, 21 - 100, and 26-80 mg/kg, respectively; (A)sampling locations of background samples (n = 9, (u) chemical plants of the industrial park.

Heavy Metals in Soils, Sediments, and Wastewaters

377

(2) Soils adjacent to the irrigation river discharged high levels of metals, indicating the effect of the contaminated irrigation water by lateral movement. (3) The high concentrations of heavy metals, Cr, Cu, Ni, and Zn, in irrigation water were caused by the discharged water of the industrial park. (4) There were significant relationships between heavy metals (Cr, Cu, Ni, and Zn) in agricultural soils, irrigation water, and sediments.

19.4 Conclusions Two rice-growing regions where soils were suspected of being contaminated by heavy metals, were selected to study the relationships between heavy metal concentrations in soils, discharged water, and sediments. Sampling design and sampling methods in these two regions were described. In areas close to the irrigation river and as far out as 30 m from the river, soils could accumulate very high concentrations of Cd in the surficial 20 cm. From the sampling design, we can conclude that the concentrations of heavy metals in soils were related to the flow pathway, direction, and distance of polluted irrigation water from the chemical plant. Results also indicate that the concentration of Cr and Ni in case I1 study were significantly higher than those set for the irrigation water quality proposed by the Government of the Republic of China. The high levels of Cr, Cu, Ni, and Zn in the contaminated discharged water were reflected also by their high levels of accumulation in soil and sediments. Acknowledgements. The author thanks the Environmental Protection Administration of Republic of China (ROC) and the Council of Agriculture of ROC for their financial supports in 1988 and 1991. The author is thankful to Dr. Domy C. Adriano, Professor of Savannah River Ecology Laboratory, The University of Georgia, USA, for his comments and review of this manuscript. The author also thanks to Mr. T. C. Lin, Mr. J. C. Liu, Miss D. N. Huang, and Mr. H. H. Liau for their assistance in sampling of agricultural soils, discharged water, and sediments in the two case study regions and in chemical analyses, and to Mr. J. 1. Hsu and J. C. Liu for their preparation of the tables and figures.

19.5 References [I] Chen, Z. S. (1988) A survey of heavy metal concentration of soils and rice grain in northern Taiwan. Project report of Council of Agriculture. Executive Yuan, ROC. [2] Chen, Z. S. (1991) Cadmium and lead contamination of soils near plastic stabilizing materials producing plants in northern Taiwan. Water, Air, and Soil Pollution 57-58, 745-7754, [3] Chen, Z. S. (1992) Metal contamination of flood soils, rice plants, and surface waters in Asia, in: Biugeuchemistry of' Trace Metals, Adriano, D. C. (ed.). Lewis Publishers, pp. 85- 108.

378

Z.-S. Chen

[4] Chen, Z. S., Lee, D. Y. (1993) The relation between pollution source and heavy metal concentrations in soils, in : Proceedings of the fourth Symposium of' Soil Pollution Preventioli andRernediution, pp. 163 - 176, Tai, K . P. (ed.), published by Energy & Resources Laboratories, Industrial Tcchnology, Hsinchu, Taiwan, ROC. [5] Chen, Z. S., Lec, D. Y., Huang, D. N., Liau, S. H. (1992) The relationships between heavy metals in soils, waters, and sediments: Case study in Shiang-Shun Industrial park. Project reports o f EPA-ROC in 1992. [6] Cheng, S. I+'., Lee, S. M . (1992) The relationships between heavy metals in soils, waters, and sediments: CUSP study in An-Nan Industrial purk. Project reports of EPA-ROC in 1992. [7] EPA-ROC ( 1 991) The standurd metliod,s,for determination of heavy metals in soils and plants. National Institute of Environmental Analysis of EPA-ROC. Taipei, Taiwan, ROC. [8] Lcc, C. D., Ling, H . T. (1983) The suroey of'heuvy metals in soils. Project report of Taiwan Water Pollution Protection Institute, Taichung, Taiwan, ROC. [9] Wang, M . C., Wang, Y. P. (1992) Effect of nitrogen and phosphorus contents in irrigation water from Ta-Chia River on soil properties and rice growth. J . Chinese Agric. Chem. Soc.. 30, 1-13. [lo] Wang, Y. P., Wang, M. K., Liu, C. L. (1992) The relationships between heavy metals in soils, watcrs, and scdimcnts: Cuse study in Chung-Mvu Industrial park. Project reports of EPA-ROC in 1992.

D. Plants and Animals

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

20 Sampling of Plants for Environmental Trace Analysis in Terrestrial, Semiterrestrial and Aquatic Environments Wilfried H . 0. Ernst

20.1 Introduction When the atomic absorption spectrometry developed to a rapidly operating and sufficiently sensitive instrument nearly 30 years ago, an enormous amount of plant material was possible to be analyzed. The International Biological Program has stimulated inventories of chemical elements in ecosystems (cf. Duvigneaud and Denayer-De Smet, 1964; Bormann and Likens, 1976; Ellenberg et al., 1986). Obviously due to great gap on element uptake, accumulation and translocation in wild plants, this program was started without any directory, how, when and what to sample. In 1971, a survey of methods applied in the research of the German Solling Project was published, however, without any directory for sampling plant material for analytical purposes (Ellenberg, 1971). With increasing progress in instrumentation for multielement analysis (Lieth and Markert, 1988; Markert, 1992) standard reference material (Griepink et al., 1983) was established to achieve comparable data of chemical laboratories. Even for this reference material it is not known how it was sampled. In the meantime, various options have developed for appropriate sampling methodologies, i.e., twig sampling in biogeochemical prospecting (Warren, 1980), leaf sampling for seasonal dynamics of elements in plants (Ernst, 1975), leaf age gradients within a plant (Ernst, 1982), element allocation in plant organs (Ernst, 1974), leaf sampling of a few tree species for regional comparison of environmental conditions (Zimmermann and Plankenhorn, 1986; Wagner, 1990). The main question to be answered will be: Can we elaborate a few standardized sampling methods and strategies, which will enable a n interlaboratory comparison?

20.2 What Do we Want to Know? Plant species can explore all three environmental compartments, i.e., the atmosphere, the hydrosphere and the pedo(litho)sphere, or they can be restricted to only one sphere. The objectives of the analysis has to be clearly established because it will demand different sampling and preparation procedures. Plant samples which are intended to establish the concentration of a chemical element and its physiological

382

W. H . 0. Ernst

and ecotoxicological consequences for a plant’s function and survival have to be sampled at random and cleaned from all remnants of the three environmental compartments (Otte et al., 1989). In the case of environmental monitoring and identification of an emission source, sampling has to be carried out in transects running from the suspected emission source to the background area (Ernst and Leloup, 1987). Atmospherically dispersed pollutants may be trapped by hairs or glutinous leaf surfaces so that species selection may be based on morphological characters. In food web studies sampling will depend on the type of the heterotrophic organism. Grazing mammals and caterpillars will ingest the plant material with all adherent soil particles and particulate fall-out (Joosse and van Vliet, 1982); sucking aphids will only take phloem sap (Ernst et al., 1990). Therefore, each herbivore typc demands another sampling procedure. Plants affected by parasites such as mildews and rusts may change the concentration of chemical elements (Ernst, 1983); therefore, a random sampling procedure will not be possible in ecosystems with a frequent occurrence of such parasites.

20.3 Are Trace Elements Relevant Objectives? Plants as other organisms demand a certain amount of elements of the periodic table for their metabolic processes, which enable growth, survival and reproduction, and for the support of biologic structures. The evolution of the uptake systems of chemical elements in plants has not resulted in a tailor-made uptake control, perhaps due to the metabolic costs or the chemical impossibility, e.g., the similarity of phosphate and arsenate (Meharg and Macnair, 1990). The moderate specificity of the element uptake systems (EUS) has as consequence that all plants can contain all chemical elements of the periodic table (Brooks, 1972) if these elements are present in a plant-available form. The lack of a high specificity of the EUS has advantages for the evolution of heterotrophic organisms. Jodium is an excellent example. Higher plants d o not use jodium in their metabolism ; mammalia inclusive humans, however, cannot live without jodium because this element is essential for the functioning of thc thyroidea. As a result of the evolution of metabolic processes, certain elements are demanded in high amounts, i.e., carbon (C), oxygen (0) and hydrogen (H) for all carbohydrates, and together with nitrogen (N) for all proteins; others are necessary in moderate amounts, i.e., phosphorous (P) for metabolic activations and energy conservation, potassium (K) for hydrature, sulfur (S) for some amino acids, peptides and sulfolipids, magnesium (Mg) for chlorophyll and calcium (Ca) for fine turning of the hydrature and biomembrane stability (cf. Marschner, 1986). All thcse elements (C, H, 0, N, P, K, S, Ca, Mg) are therefore categorized as mujor essentiul elements or mucronutrients. These macronutrients are out of the scope of this contribution and the topic of this book.

PLints f o r Trace Analysis

383

A group of other elements are only necessary in small to very small amounts in plants, sometimes only known for one process, i.e., nickel (Ni) for the activation of the enzyme urease. This group of elements, defined as minor essential elements or micronutrients,includes Fe, Mn, Zn, Cu, Mo, Ni, B, Na and C1. The sophistication of instruments for trace elements and of the purification procedure of chemicals, used in experimental approvals of the essentiality of an element will certainly let detect metabolic functions of elements beyond the above-mentioned groups of micronutrients. Due to the high diversity of biological organisms and the moderate specificity of the uptake system certain plant species or plant groups of a higher taxonomic unit have evolved a demand for some additional element. The amount demanded may range from high, i.e., equivalent to major essential elements as silicium (Si) in the structure of diatoms, Equisetaceae, Cyperaceae, and Poaceae, to low, as equivalent to minor essential elements in the stingle hairs of Urticaceae. Other elements are only specific for one group and for one process: cobalt in nitrogen fixating bacteria living in symbiosis with Fabaceae (Dilworth et al., 1979). Some of these and other elements are often defined as benejfi’cial(Marschner, 1986) because they can stimulate plant growth, i.e., titanium (Pais, 1983), although their role in the metabolism is not understood. All other elements of the periodic table are non-essential to plants. The are often called trace metals (Adriano, 1992), trace substances (Adriano, 1992) or trace elements (Stiles, 1946). Unfortunately, none of the authors gives a rational definition of “trace”. Often the inadequate use of the term “trace” provokes biologically awful misunderstanding. Adriano (1 992) looks to trace metals as “important environrnental contaminant that affect all the ecosystem components”. Either he excludes all micronutrients - what he does not - or he overlooks that each element - also N, P, K - injures biological systems when a certain concentration is surpassed. When the term “trace” should indicate the presence of a very small concentration of an element in biota, the high biological diversity of element accumulation will be contradictious. A lot of plant species can hyperaccumulate an element in one of its plant part, i.e., more than 1000 pg element g - ’ dry wt (Brooks et al., 1977). As an example, the concentration of lead in leaves is low in most plants (1 pg g-’ dry wt; Markert, 1992), but it can increase above 10000 pg g - I , as in leaves of Minuartia vernu growing on ore outcrops (Ernst, 1974). Taxon-specific accumulation is well known for several elements, such as zinc in the families of Betulaceae and Salicaceae or nickel in the taxon Alyssum. Certain taxa are able to incorporate an element into organic compounds, i t . , V in the amavadin of the Fly agaric Amanita muscariu (Kneiffel and Bayer, 1973) or F in fluoracetate of the F-accumulating Dichupetulum species (Meyer et al., 1992; O’Hagan et al., 1993). In other taxa an element can be bound in such a way to an organic compound that it will not affect the performance of adapted plant species. Examples are Se in seleno-amino acids of Se-accumulating plants (Peterson and Butler, 1967), and A1 in phenolic complexes of tea (Nagata et al., 1992). Therefore I will treat “trace” elements as the group of chemical elements being micronutrients and non-essential elements for plants. A good analysis of plant material, however, will also incorporate a part of the major nutrients, because

384

W. H . 0 . Ernst

the general knowledge about uptake, accumulation and retranslocation has such a good background (Baumeister and Ernst, 1978; Marschner, 1986), that ratios of N/P, P/K, K/Ca can give a realistic impression on a plant’s physiological status.

20.4 The Exploration of Environmental Compartments The diversity of plant species makes it possible that different portions of the environment in space and time can be exploited. Terrestrial higher plants are normally exposed to the hydrosphere, atmosphere and pedosphere; aquatic plants may only have access to elements present in the hydrosphere; epiphytic plants may be exposed only to elements of the atmosphere. Exposure to one environmental compartment may allow to restrict the sampling size due to a certain homogeneity of the exposed plants.

20.4.1 The Hygro-and Hydrophytes Plants which only explore the hydrosphere without contacts to the pedosphere and the atmosphere are planktonic algae, some aquatic mosses and some submersed aquatic angiosperms (hygrophytes). Although a lot of studies on the uptake of chemical elements are performed with unicellular algae like Chlorcdlu and Euglcwu (De Filippis et al., 1987;Weber, 1981; Kessler, 1986; Gekeler et al., 1988), it will be impossible to collect a monospecific, homogeneous sample of planktonic algae for chemical analysis. Nevertheless, some of these algae may be excellent reference material, because a periodic or permanent exposure of these algae to a surplus of a chemical element will result in the selection of specific resistant populations. Therefore, not the chemical analysis, but the biological resistance test will give excellent results of the environmental quality (Stokes et al., 1973; Butler et al., 1980; Ahlf and Weber, 1981; Foster, 1982). In the few possibilities of sufficient amounts it will be necessary for an adequate data interpretation of element concentrations in uni- to paucimulticellular algae to determine the surface/volume ratio, because many elements will not be taken up, but only adsorbed to the cell surface (Christlieb and Werner, 1980). A lot of other organisms in the aquatic environment makes also contact with the pedosphere enabling the fixation of the organism a t a certain site for a prolonged period of time. Element concentrations of these benthic organisms may be the result of the access to chemical elements of their substrates and the surrounding waters. Filamentous benthic algae such as Cladophora, Stigeoclonium and Vaucheria have only little contact with the substrate so that samples of these algae may give an integration of element exposure to the surrounding water over a half year to year period (Keeney et al., 1976). The drawback of these algae is the strong adhesion of fine soil particles from the surrounding water, which are difficult to be removed.

Plants for Trace Analysis

385

A similar sampling problem will arise with the epiphytic algae and protozoa on these benthic algae. Although it has been stated that the element concentration of this algal “aufwuchs” is the same as its supporting host (Prosi, 198l), a generalization of this statement may fully underestimate the uptake and accumulation specifity of algae. The development of populations resistant to a surplus of zinc (Harding and Whitton, 1976), phosphorus and nitrogen (De Vries et al., 1985) may interfer with uptake and accumulation of chemical elements, as it is known from Cu-tolerant and non-tolerant populations of the marine fouling alga, Ectocurpus siliculous (Hall, 1981). In marine ecosystems analysis of macroalgae, e.g., Fucus serratus, F. spiralis and F. vesiculosus, Ascopliyllum nodosum and Ulva lactuca have widely been used to study the concentrations of chemical elements (for a review, see Phillips, 1977; Ernst, 1987). Gradient sampling can easily be performed, but particulate contamination is a significant problem in sample preparation (Barnett and Ashcroft, 1985). Aquatic plants with roots can take up chemical elements from the sediment and the surrounding water. Water sediments, however, are often characterized by steep oxygen gradients, which affects element speciation and bioavailability in a distance of a few millimeters. Due to strong polar element transport through the leaves of aquatic plants, photosynthetically active leaves can precipitate huge amounts of calcium carbonate on the upper leaf surface (Steemann Nielsen 1947; Baumeister and Ernst, 1978). These precipitates have to be removed during sampling, because they also contain coprecipitates of heavy metals. Another problem with aquatic plants is the colonization of leaves and stems by a lot of aquatic algae, the so-called “aufwuchs”. It is very difficult to remove part of this “aufwuchs”, especially diatoms and unicellular green and bluegreen algae. Sampling for a database of metal concentrations in aquatic plants demands a thorough cleaning procedure and a microscopial control of the material prior to analysis; especially leaves of Potamogeton species are difficult to handle. In addition, adsorbed elements have to be removed during sampling by short-term exchange processes. Plants floating on water surfaces like Lemna, Azolla, Nuphar and Nymphaea can take up chemical elements from the hydrosphere and the atmosphere. During sampling, two pollution sources have to be avoided: the “aufwuchs” and the adsorption of elements on the leaf surface surrounded by water, and the contamination by faeces on the leaf surface exposed to the atmosphere. In addition, wind can remove Lemna species from the sites where they have taken up the chemical elements, within a few hours so that sampling of floating water plants have carefully to register their site “stability”.

20.4.2 The Epiphytes Epiphytes on terrestrial higher plants have only access to chemical elements which are supplied to them by wet and dry deposition. Sampling of epiphytic lichens and

386

W. H . 0. Ernst

epiphytic mosses for chemical analysis has a long history (Ferry et al., 1973). Due to the high cation exchange capacity (Clymo, 1963) most cations are adsorbed to the cell wall, so that washing procedures can remove at least a part of the adsorbed elements. Strong precipitation just before and during sampling, especially precipitation with a low pH, may affect the element concentrations (Brown and Brown, 1990). To my knowledge, there is no detailed study of the impact of the various factors, although the sampling procedure has to consider plant height, lichen exposition, etc. ; these details, however, are not mentioned in recent publications (Herzig et al., 1990). Without washing, the plant sample remains contaminated by particulate deposition. Another aspect of sampling epiphytic lichens and mosses is their exposition to precipitation. U s n w species hanging from a twig under an open canopy will collect chemical elements from throughfall more than from crown drip, whereas those under a closed canopy will rely on crown drip. Epiphytes near the lower part of a stem will be exposed to the chemical elements of the stem flow. Due to the great differences of element concentrations between throughfall, crown drip and stem flow (Ellenberg et al., 1986) sampling of these epiphytes has to register these conditions.

20.4.3 Soil Exploring Plants Exposure to chemical elements in the soil is conditioned by soil moisture. With the exception of some algae living a few millimeters below dune sand, all organisms being completely restricted to the pedosphere are heterotrophic. Sampling of these heterotrophs, mostly slime molds, bacteria and fungi, is very difficult, if not impossible. Their interaction with the substrate is so intense that they cannot be dissected without a high degree of sample contamination. There are only some exceptions like the hyphae of fungi condensed to rhizomorphs (Stark, 1972) and the sporophores of fungi (Allen and Steinnes, 1978; Mutsch et al., 1979; Hinneri, 1975; Tyler, 1982; Dietl, 1987; Gast et al., 1988). Due to the rapid uptake of chemical elements from the litter layer into the hyphae and the translocation to the sporophore, fruit-bodies can be used for the localization of particulate and wet deposition, as clearly demonstrated after the Chernobyl disaster (Ernst and van Rooij, 1987). Although sporophores of fungi are exposed to the atmosphere for some days, all chemical elements are derived from the pedosphere. Sampling can be carried out a t random, in relation to suspected host plants in the case of mycorrhizal fungi and as transects in the case of monitoring the impact of a point emission source on ecosystems. Litter and soil particles can be removed from the sporophores without essential loss (Allen and Steinnes, 1978; Hinneri, 1975); separation of stipes and pilei is often advised, but it will depend on the research scope (Hinneri, 1975). One aspect is unsufficiently recognized : The chemical composition of the pileus (cap) will depend on the amount of basidiospores already released. Therefore, during sampling the maturity of the cap is an essential factor to be considered.

Plants for Traee Analysis

387

Roots of higher plants explore the pedosphere, often related to the various soil horizons. Sampling of roots is one of the most difficult tasks in plant analysis of chemical elements. First of all, most higher plant roots have a strong association with mycorrhizal fungi, either vesicular-arbuscular mycorrhizal (VAM) fungi and/or with ectomycorrhizal (EM) fungi (Cooke, 1979; Allen, 1991). Due to the seasonal variations of the infection degree with VAM fungi (Ietswaart et al., 1992; Veenendaal et al., 1992), the difficulty to collect sufficient hyphae for chemical analysis and the inability of a pure culture of VAM fungi, all samples of roots associated with VAM fungi will be a mixture of fungi and angiosperm tissue; the biomass of the fungal component is estimated between 1 and 10%. In the case of ectomycorrhizal roots, EM fungi can be grown in monoculture without host plants and have shown a very species-specific and population-specific uptake pattern of chemical elements (McCreight and Schroeder, 1982; Colpaert and Van Assche, 1992), resulting in great differences of the element concentration in the host tree (Colpaert and van Assche, 1993). During the life history of a tree there is an age-dependent succession of ectomycorrhizal fungi (Fleming, 1985). Therefore, ideal sampling procedures of roots will establish the degree of mycorrhizal infestation and the involved fungus species. However, I have to agree that such an ideal sampling is extremely time-consuming and it demands an intensive cooperation with fungi taxonomists. Independent of the mycorrhizal fungi, root sampling gives some more problems. Root hairs have a very intimate relationship with soil particles, so that it is impossible to remove soil particles without losing root hairs and fine roots. The root surface itself, i.e., the rhizoplane, is a well-known site for interactions between free-living microorganisms and plants. The number of cells of microorganisms in the rhizoplane are estimated up to 120 x lo9 cells cm-3 root (Paul and Clark, 1989). Also these organisms can only be removed with a loss of root material. Furtheron, removal of root tissue during sampling and sample preparation will have a strong impact on element concentrations, because the various root tissues accumulate the various elements to a different degree (Ernst, 1974). As mentioned earlier, many chemical elements are adsorbed to exchange sites of the root cell wall. If a difference will be made between adsorbed and cell-internal element concentrations, exchange of adsorbed elements has to be carried out immediately after sampling. Effective exchange procedures, however, will load the root cell wall with another ion. Therefore, the aim of the data collection will select the exchange solution applied, e g , Pb(NO,), for removal of adsorbed heavy metals others than lead (De Vos et al., 1992)or rinsing thoroughly with demineralized water for removal of adsorbed sodium and potassium. In wet ecosystems welladapted plants have aerenchyma which ensure not only oxygen supply to the root cells but also initiate an oxydation zone around the root building up an iron plaque (Otte et al., 1989). This iron plaque can be removed by treatments with dithionitecitrate-bicarbonate. In every case its presence has to be registered at sampling. All in all, sampling of roots will remain a difficult operation. Chemical analysis of root material will result in data which are inherently contaminated. As long as all sampling procedures are well described, “contaminated data” are better than no data.

388

W. H. 0 . Ermt

20.5 Plant Parts and Life History Representative sampling demands the recognition of the individual, which can be a difficult task in clonal plants. Individuals of these plants may cover surfaces from some to over hundred m2. e.g., Urticn dioica, Chamnenerion angustifolium, PhrugmitPs australis, Hedern helix and Hippophac rhamnoides (Ernst, 1990). When the stolons run over environmental gradients as in Agrostis stoloniferu and in Cure.u arennria, they may have access to completely different concentrations of chemical elements so that local uptake and transport through the stolon (Tietema, 1981) may be mixed. In plants with a well developed dioecy, it is necessary to sample both sexes. A combination of dioecy and clonal life form, e.g., Hippophue rhnmnoides, may give problems for representative sampling because only a few individuals make up a population. Plants are composed of various plant organs with different physiological functions and element demands. Therefore, each sampling procedure will depend on the research scope, so that only some general rules can be established. At a fixed time in the life of an individual each plant organ is characterized by concentrations of chemical elements, which are the result of uptake, translocation to and from other organs and accumulation potential in relation to the supply by the environmental compartments and to biological interactions, partially governed by its genetic make-up. Dynamics of elements is part of the physiological processes inherent to biodiversity. Therefore, accounts on mean concentrations of chemical elements of plants for a whole country are excellent examples of pseudoscience (Angelone and Bini, 1992) and demonstrate the misunderstanding of the principles of biodiversity, plant physiology, plant genetics and ecology. When the dynamics of chemical elements is the scope of the investigation, plant organs have to be collected during the growing season or a t certain developmental stages on a previously designed pattern. In the case of small plants, it may be necessary to sample a complete individual or a group of individuals, the amount of biomass to be sampled for at least three replicates of chemical analysis will be determined by the sensitivity of the analytical instrument for those chemical elements with an estimated low concentration. Samples of 300 mg dry mass may be sufficient. In the case of medium-sized plants, it may desirable to take plant organs at random. Plants with a high biomass of each or several plant organs and with a plant height exceeding that of the collector demand special designs and instruments for random sampling like the telescope knife as described by Muller and Wagner (see Lieth and Markert, 1988). Whereas in the case of small individuals or low biomass of sampled plant organs the whole sample can be used for analysis, samples with a biomass above the analytical demand will create problems of homogenization (see Markert, 1993). Critical examinations of damage by heterotrophs and the presence of fungi is necessary (see Sect. 20.4). In all sampling procedures, it should be obvious that the impact of sampling on the population has to be near zero. As soon as populations of the same species at various sites or ecosystems will be compared, sampling can be concentrated on one plant organ, and in the case

Plunts for Trace Analysis

389

of leaves, at one well-defined position of the leaves on the plant. Such a research scope makes it necessary, to enlarge the number of sampled individuals to ten or more per site, dependent on the visible genetic variation. If data sets show a high variability within sites, even other techniques such as isozyme analysis (Cuguen et al., 1985) can be applied. Relatively long-living plant organs such as the needles of coniferous trees and the annual increment of wood (cf. Ernst, 1990) can be sampled by age classes. Age gradients are often overlooked in shoots of rushes, where the top can be two years old and the shoot base only a few weeks, even if the shoot gives the impression of an evenly aged organ (Ernst, 1991). The developmental stage of the plant, however, has to be registered carefully . When the objectives of the research is the maximum of element load which will be recycled in the ecosystem, then deciduous plant parts just before shedding will be the sampling target. With a few exceptions like F and B, a surplus of an element taken up from the environment is mostly accumulated at highest concentration in a species-specific pattern in senescent plant organs (Ernst, 1990). Ageing may not occur simultaneously in all plant organs of the same plant; therefore, samples of the oldest leaf will give the most relevant information, as shown for leaf rosettes of Armeria nzaritima, Plantago lanceolatu and Thlaspi caluminave (Ernst, 1974). Loads of non-essential elements are generally very small in seeds, although the process hampering element translocation from the shoot to the seed is physiologically not well understood. Sampling of seeds has the advantage of a mostly clean material, especially seeds enclosed in capsules or in fleshy or dry fruits, but it has the disadvantage that the small seed mass of most plants demands the collection of a seed pool, sometimes of a great part of the population, thus endangering population recruitment. The preference of a plant species, a genus or even a family for the uptake and enrichment of specific elements, such as silicium by Equisetaceae and others (see Sect. 20.3) may give matrix problems in analysis. Therefore each sampling procedure should take into account the sampling of a “model” species. In each sampling protocol the following parameters have to be noted:

( I ) environmental conditions of soil, water table if high, temperature, last day of precipitation; (2) biological conditions of the plant such as developmental stage, presence of parasites, saprophytes and/or herbivores, health conditions of the plant or the plant part (visible necrosis, chlorosis, malformation, damage by herbivores) ; (3) obvious contaminations with soil particles, which is quite frequent in episodically or regularly flooded ecosystems like salt marshes and river banks. In the case of such contaminations analysis of Ti and A1 can later help to control the efficiency of the cleaning procedure. Chemical elements in plants are mostly derived from the pedosphere and/or hydrosphere. Therefore, a meaningful design of plant sampling has to be related to soil and/or water sampling. A general sampling strategy for plant materials is presented in Tab. 1.

390

W . H . 0. Ernst

Tab. 1. Sampling Strategy for Plant Material Sampling

Definition of the objectives Ecosystem budgets

Community interaction (food web)

Population biology

Ecophysiology

Environmental monitoring and emission source identification

Sampling strategy

Random

Selected

Random/ selected

Random selected

I . Selected for emission source 2. Random for general changes

Sampled species

Dominant species in various ecosystem Strata

Interacting autotrophic plants and hctcrotrophs

Species of interest; genotype selection

Species of interest; genotypc selection

1. Same species along a trailsect 2. Sevcral species common in the area

Sampled organ

All organs

At least affected organ

Selection due to objectives

Selection due to objectives

Mostly leaves

Sampling time

Once a year

During the interacting period

Throughout the year

Throughout the growing season

Mostly a the end of the growing season (maximum accumulation)

General attention Special attention: None

Identification of individuals, dcvclopment (age) stage Impact on non-affected organ

Allocation changes during development

Allocation changes during development

Comparable age of the organ; comparable exposition; comparable environmental conditions except elements of thc monitoring or identification purpose

20.6 Conclusion A sampling strategy of plants has to be based on a well-formulated scope. The objective of the data analysis dcterniines the sampling period, the samplcd plant organ and the sampled biomass. The high dynamics of chemical elements in plants during growth and development demands a protocol with a lot of biological details. In the ideal situation the state of mycorrhization and the involved fungal species have to be determined, because finally all plant data will be related to data of thc

Plants ,for Truce Analysis

39 1

surrounding environment. Except for food web studies, “aufwuchs” on leaves of aquatic plants and the phylloplane organisms on terrestrial plants will cause severe biological contamination of plant samples.

20.7 References Adriano, D. C. (ed.) (1992) Biogeochemistry of’ Truce Metuls. Boca Raton, Florida: Lewis Publ. Ahlf, W., Weber, A. A. (1981) A simple monitoring technique to determine the heavy metal load of algae in aquatic ecosystems. Environ. Technol. Letters 2, 3 17 - 322. Allen, R . 0..Steinnes, E. (1978) Concentrations of some potentially toxic metals and other trace elements in wild mushrooms from Norway. Chemospkere 4, 371 -378. Allen, M. F. (1991) The Ecology of Mycorrhizae. Cambridge: Cambridge University Press. Angelone, M., Bini, C. (1992) Trace elements concentrations in soils and plants of Western Europe, in: Biogeochemistry yj’Trace Metals; Adriano, D. C. (ed.). Boca Raton, Florida: Lewis Publishers, pp. 19-60. Barnett, B. E., Ashcroft, C. R. (1985) Heavy metals in Fucus uesiculosus in the Humber estuary. Environ. Pollut. B9, 193-213. Baumeister, W., Ernst. W. H. 0. ( I 978) Miniwdstoffe und Pflanzenwachstum. Stuttgart: Gustav Fischer Verlag. Hormann, F. H., Likens, G. E. (1979) Pattern and Process in a Forested Ecosystem. New York: Springer Verlag. Brooks, R. R. (1972) Ceohotany and Bioguochi>mistry in Mineral Exploration. New York: Harper & Row. Brooks, R. R., Lee, J., Reeves. R. D., Jaffre, T. (1977) Detection of nickeliferous rooks by analysis of herbarium specimens of indicator plants. J . Geochem. E-ypl. 7, 49. Brown, D. H., Brown, R. M . (1990) Reproducibility of sampling for element analysis using bryophytes, in: Elment Concentration Cudasters in Ecosystems, Lieth, H., Markert, B. (eds.). Weinheim: VCH Verlagsgesellschaft, pp. 5 5 - 62. , M . M. (1980), Copper tolerance in the green alga, Chlorella Butler, M., Haskew, A. E. .I.Young, vulguris. Plant Cell Emiron. 3, 1 I9 - 126. Christlieb, Th., Wcber, A. (1980) Dic Bedeutung der ZelloberflCche fur die Sorption von Blei durch cine coccale Grunalge. Emiron. Technol. Letters 1, 3 1 1 - 3 18. Clynio, R. S. (1963) Ion exchange in Sphagnum and its relation to bog ecology. Ann. Bat. 27, 309- 324. Cooke, W. B. (1979) The Ecology c?f’Fiingi. Roca Raton, Florida: CRC Press. Colpaert, J . V., Van Assche, J . A. (1992) The effects of cadmium and the cadmium-zinc interaction on the axenic growth of ectomycorrhizal fungi. Plant Soil 145, 237-243. Colpaert, J. V., Van Assche. J. A. ( I 993) The effects ofcadmium on ectomycorrhizal Pinus syluestris L. NCN’Plzytol. 133, 325 -333. Cuguen, J . , Thiebaut, B., N’Tsiba, F., Barrierc, G. (1985) Enzymatic variability of beechstands (Fagus sylvaticu L.) on three scales in Europe, in: Genetic Diflerentiation and Dispersal in Plants: Jacquard, P., Heim, G., Antonovics. J. (eds.). Berlin: Springer Verlag, pp. 17-39. De Filippis, L. F., Hampp, R.. Ziegler, H. (1981) The cffect of sublethal concentrations of zinc, cadmium and mercury on Euglena. Growth and pigments. Z . Pjlanzenphysiol. 101, 37 -47. De Vos, C. H. R., Vonk, M. J., Vooijs, R., Schat, H. (1992) Glutathione depletion due to copper-induced phytochelatin synthesis causes oxidative stress in Silene cucuhalus. Plant Physiol. 98, 853 - 858. De Vries, P. J . R., De Smet, S. J. M., van der Heide, J. (1985) Effccts of phosphorus and nitrogen enrichment on the yield of some strains of Stigeoclonium Kutz. (Chlorophyceae). Freshwater B i d . 15. 95- 103. Dietl, G. ( I 987) Abhingigkeit dcr Schwcrmetallaufnahme hoherer Pilze von der Substratzusammensetzung und von Standortsfaktorcn. Bihl. Mycol., Vol. 110, Cramer, Berlin.

392

W. H. 0.Ernsf

Dilworth, M. J., Robson, A. D., Chatel, D. L. (1979) Cobalt and nitrogen fixation in Lupinus ungustifolius L. 11. Nodule formation and functions. New Phytol. 83, 63 - 70. Duvigneaud, P., Denayer-De Smet (1964) Le cycle des elkments biogenes dans l’ecosysteme forEt (For?ts temperees caduesfoliees). Lqjeunia 28, 1 148. Ellenberg, 1 1 . (ed.) (197 1 ) Integroted Experimental Ec0/11g~~. MethoLis and Results of‘ G o s j ~ w t Research in the Gerniun So//ing Proyect. Ecoiogicdl Studies. Berlin: Springer Verlag. Vol. 2. Ellcnbcrg, H., Maycr, R., Schaucrmann, J. (1986) Okosystemjorschung. Ergehnisse des Sollingprojekts 1966- 1986. Stuttgart: Ulmer Verlag. Frrnst, W. H. 0. (1974) S c h ~ ~ ~ ~ r m r ~ f a l l o ~ ~der g ~ ~Erde. t r r t iStuttgart: on G. Fischer Verlag. Ernst, W. H. 0. Okophysiologic von Pflanzcn in Kiistenduncn Europas in cincm Gradientcn von dcr Nordsec zum Mittclmccr, in: Berichte der Rrinhold-Tuxen-Gt~sellschaft,Pott, R. (cd.). Gottingen: Goltze Verlag, Vol. 3, pp. 157- 172. Ernst, W. H. 0. (1975) Variation in thc mineral contents of lcavcs of trees in miombo woodland in South Central Africa. J . Ecol. 63, 801 -808. Ernst, W. H. 0 (1982) Schwermetallpflanzen, in: Pjlunzenijkokogie und Mineral.st~~~ivc,t.hsc.l, H . Kinzcl (cd.). Stuttgart: Ulmer Verlag, pp. 472-506. Ernst. W. H. 0. (1983) Population biology and mineral nutrition of Anemoiw nc~n7uru.s~with emphasis on its parasitic fungi. Flora 173, 335-348. Ernst, W. H. 0..Leloup, S. (1987) Perennial herbs as monitor for moderate levels of metal fall-out. Chemosphere 16, 233 - 238. Ernst, W. H. 0. (1987) Mctal fluxcs to coastal ccosystcms and the rcsponsc of coastal vegetation, in: Vqetation hetwec‘n 1,undand Sca, Huiskes, A. H. L., Blom, C. W. P. M., Rozema. J. (eds.). Dordrecht: Junk Publishers, pp. 302-310. Ernst, W. H. O., Van Rooij, L. F. (1987) 134’137Cs fall-out from Chcrnobyl in Dutch forest, in: H L ~ metals J in the Environment, Lindberg, S . E., Hutchinson, T. C. (cds.). Edinburgh: CEP Consultants, pp. 284-286. Ernst, W. H. 0. (1990) Element allocation and (re)translocation in plants and its impact on representative sampling, in: Elemenr Concentration Cadasters in Ecosystems: Lieth, H ,, Markerl. B. (eds.). Weinheim: VCH Verlagsgesellschaft, pp. 17-40. Ernst. W. H. O., Schat, H., Verkleij, J. A. C. (1990) Evolutionary biology of metal resistance in Silene oulgaris. Evol. trends plants. 4. 45 - 51. Ferry, B. W., Baddeley, M. S., Hawksworth, D. L. (1973) Air Pollution ond Licliens. London: Athlone Press. Fleming, L. V. (1985) Experimental study of sequences of cctomycorrhizal fungi on birch (Betulir sp.) seedling root systems. Soil Biol. Bioc/iem. 17, 591 -600. Foster, P. L. (1982) Metal resistance of Chlorophyta from rivers polluted by heavy metals. Freshwater B i d . 12, 41 -61. Cast, C. H., Jansen. E., Bierling, J., Haanstrd, L. (1988) Heavy metals in mushrooms and their relationship with soil characteristics. Chemosphere 17, 789 - 799. Gckclcr, W., Grill, E., Winnackcr, E. L., Zenk, M. H. (1988) Algae sequester heavy metals via synthesis of phytochelatin complexes. Arch. Microhiol. 150, 197-202. Griepink, B., Muntau, H., Colinet, E. (1963) Certifications of the contents of cadmium, copper. manganese, mercury, lead and zinc in two plant materials of aquatic origins and in olive leaves. Fresenius Z . Anal. Chem. 315, 193 - 196. Hall, A. (I98 I ) Copper accumulation in coppcr-tolcrant and non-tolcrant populations of the marinc fouling alga, Ectocarpus siliculosirs (Dillw.) Lynbyc. Bot. Mar. 24, 223 - 228. Harding, J. P. C., Whitton, B. A. (1976) Resistance to zinc of Stigeor.lonium tenuc’ in the field in the laboratory. Br. ph~vcol.J . 11, 417-426. Hcrzig, R., Urcch, M., Licbcndorfcr, L., Ammann, K., Guecheva, M., Landolt, W. (1990) Lichens as biological indicators of air pollution in Switzerland: passive biomonitoring as apart of an integrated measuring system for monitoring air pollution, in: Element Concenrrution Cadusters in Ecos,ystems, Lieth, H., Markcrt, B. (eds.). Weinheim: VCH Verlagsgesellschaft, pp. 55 - 62. I Iinneri, S. (1975) Mincral clements of macrofungi i n oak-rich forests on Lensholm Island, Inncr Archipelago of SW-Finland. Ann. Bot. Fenn. 12, 135- 140. ~

Plants ,for Trace Analysis

393

Ietswaart, J. H., Griffioen, W. A. J., Ernst, W. H. 0. (1993) Seasonality of VAM infection in three populations of Agrostis cupillaris (Graminede) on soil with or without heavy metal enrichment. Plunr Soil 139, 67-73. Joosse, E. N. G., Van Vliet, L. H. H. (1982) Impact of blast-furnace plant emission on a dune ecosystem. Bull. Emiron. Cuntum. Toxicol. 29, 279 - 284. Keeney, W. L., Breck, W. G., Van Loon, G. W., Page, J. A. (1976) The determination of trace metals in Cludop//ora glomerirtu as a potential biological monitor. Watrr Research 10, 981 -984. Kessler, E. (1986) Limits of growth of tive Clzlurellu species in the presence of toxic heavy metals. Arch. Hydrohid. Suppl. 13 (I), 123- 128. Kneiffcl, H., Bayer, E. (1993) Strukturermittlung der Vanadiumverbindung des Fliegenpilzes, Amavadin. Angew. Chem. 85, 542- 543. Lieth, H. H. F., Markert, B. A. (1988) Aujstellung und Austwtung ijkosystrmnrer E k v w n f Kr)nzentrutions-Kataste~.Berlin. Springer Verlag. Markert, B. (1992) Multi-elemcnt analysis in plant materials. Analytical tools and biological questions, in: Biogrochemistry uf'Trucc1Metals. Boca Raton Florida, Lewis Publ., pp. 401 -428. Markert, B. (1993) Instrumental analysis of plants, in: Plants as Biomonitors. Indicatorsfor Henuy Metals in t l i ~Tcwt~striulEnuironment: Markert. B. (cd.). Weinheim: VCH Vcrlagsgesellschaft, pp. 65- 103. Markert, B. (1993) Instrumentelle Multielrrnc~ntunalpse uon Pflunzenprohen. Weinheim: VCH Verlagsgcsellschaft. Marschner, H . (1986) Minerd Nutrition 1 4 ' Higher Plants. London: Academic Press. McCreight, J. D., Schroeder, D. B. (1982) Inhibition of growth of nine ectomycorrhizal fungi by cadmium, lead and nickel in iiitro. Enuiron. Exp. But. 22, 1 - 7 . Mcharg, A. A,, Macnair, M. R. (1990) An altered phosphate uptake system in arsenate-tolerant Holclt.7 lutlutus L. neb^ Phytol. 116, 29 - 35. Meyer, J . J. M., Grobbelaar, N., Vleggaar, R., Louw, A. J. (1992) Fluoroacetylcoenzyme. A hydrolase-like activity in Dichupetahtm cymosum. J . P l m i Phpsiol. 139, 369- 372. Mutsch. F., Horak, O., Kinzel, H. (1979) Spurcnelemente in hoheren Pilzen. Z . Pflunzenpliysiol. 94, 1 10. Nagata, T., Hayatsu, M., Kosugc, N. (1992) Idcntification of aluminium forms in tea lcaves by "AI N M R . Phytochemistrp 31, 1215- 1218, O'Hagan, D., Perry, R., Loek, J. M., Meyer, J. J. M., Dasaradhi, L., Hamilton, J. T. G, Harper, D. B. ( I 993) High lcvels of monofluoroacctate in Diclzapetalum hraunii. Phytochernistry 33, 1043- 1045. Otte, M. L., Rozema, J., Koster, L., Haarsme, M. S., Broekman, R. A. (1989) Iron plague on roots of ,4ster tvipolium L.: interaction with zinc uptake. New Phytol. 111- 309-317. Pais. I. (1983) The biological importance of titanium. Plunt Nutrition 6, 3- 131. Paul, E. A., Clark, F. E. (1989) Soil microbiology and Biochemistry. San Diego: Academic Press. Peterson, P. J., Butler, G. W. (1967) Significance of selenocystathionine in an Australian selenium accumulating plant Neptuniu arnple.xicau1i.s. Nature 219, 599 - 600. Phillips, D. J. H. (1977) The use of biological indicator organisms to monitor trace metal pollution in marine and estuarine environments - a review. Enuiron. Pollut. 13, 282-317. Prosi, F. (1981) Heavy metals in aquatic organisms, in: Metal Pollution in rhe Aquacic Enuironment. Forstner, K., Wittmann, G. T. W. (eds.), Berlin: Springer Verlag, pp. 271 -323. Stark, N. (1977) Nutrient cycling pathways and litter fungi. Bio Science 22, 355-360. Stccmann Nielscn, E. (1947) Photosynthesis of aquatic plants with special reference to the carbon sources. Dunsk Bot. Ark. 12, 1-71. Stiles, W. (1964) Truce Elmnents in Plunts und Aniinuls. Cambridge: Cambridge Univ. Press. Stokes, P. M., Hutchinson, T. C., Krauter, K . (1973) Heavy-metal tolerance in algae isolated from contaminatcd lakes near Sudbury, Ontario. Cun. J . Bot. 51, 2155-2168. Tietema, T. (1991) Ecoykysiology uf'the Sund Sedge. Curex arennria L. Dissertation University of Utrecht. Tyler. G. (1980) Mctals in sporophores of basidiomycetes. Trans. Br. mycol. Suc. 74, 41 -49. ~

394

W . H . 0. Ernst

Veenendaal, E. M., Monnaapula, S. C . , Gilika, T., Magole, J. L. (1992) Vesiculardrbuscular mycorrhizal infection of grass seedlings in a degraded semiarid savanna in Botswana. Neic Phvtof. 121, 477-485. Wagner, G. (1990) Variability of element concentrations in tree leaves depending on sampling parameters, in: Element Concentration Cadasters in Ecosystems. Lieth, H., Markert, B. (eds.). Weinheim: VCH Verlagsgesellschaft, pp. 41 - 54. Warrcn. H. V. (1980) Biogeochemistry, trace elements, and mineral explorations, in: Applied Soil Truce Elements, B. E. Davies (ed.). Chichester: John Wiley. Wcber, A. (1981) An uncomplicated screening test to evaluate toxicity of environmental hazardous compounds in water. Environ. Tecknol. Letters 2, 323 -328. Zimmerman, R. D., Plankenhorn, W. E. (1986) Methodik der Blattprobenentnahme an der Rotbuche unter emissionsokologischem Aspekt. AlIg. Forstz. 41, 33 - 35.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

On the Samp ,,ig of Vascu ar Plants for Monitoring of Heavy Metal Pollution 3

Rumiunu Djingovu und Ivelin Kuleff

21.1 Introduction Sampling of environmental materials for chemical analysis has lately received serious attention since it was recognized that incorrect proceduces may introduce an error exceeding orders of magnitude the analytical error thus rendering the subsequent analysis meaningless (e.g., Sansoni, 1986; Markert and Steinbeck, 1988; Markert, 1988). While sampling procedures for soil and air pollution monitoring have been developed, discussed and standardized, special attention to plants has not been paid for a long time because it was considered that sampling strategies valid in forestry and agriculture may be utilized directly (Chapman, 1976). Nowadays it is already clear that due to the specificity of monitoring the respective sampling procedures for plants should fulfill special requirements (e.g., Ernst, 1990; Markert, 1993). A first step in any analytical strategy is the definition of the purpose for which the analysis is performed (Ku, 1979), and on this basis a sampling procedure is to be developed. If biomonitoring of heavy metal pollution is a final aim in plant analysis then some very specific points should be considered. According to Markert (1993) biomonitoring means quantification of the pollution which can be obtained by several means: -

comparison of the polluted areas; comparison of long-time series; comparison with normal values; comparison with direct measurements.

That is, in all cases the samples should be taken in a way that permits meaningful comparison of the analytical results according to time and/or place. This undoubtedly introduces except the general sampling requirements concerning: avoidance of contamination and losses; quantity of field and laboratory samples; - representative division of the sampling area. -

A number of additional factors are to be considered. Among them the most important to receive attention in the case of plant biomonitoring are: -

concentration differences at biological levels; sampling period; sampling pretreatment.

396

R . nlmgovtr uml 1. K u k f f

I n recent years, a number of excellent contributions have been published discussing different aspects of the above mentioned factors (e.g., Martin and Couphtrey, 1982; Fleming et al., 1986; Camerlynck and Kiekens, 1986; Ernst, 1990; Wagner, 1990; Markert, 1993) and proposing standardized sampling proccdures for some plant species for monitoring purposes (e.g., Wagner, 1987; Knabc, 1984). The present contribution aims to stress some problems of sampling for biomonitoring studies using vascular plants for the assessment of heavy metal pollution and to summarize some of the results obtained in this field.

21.2 Concentration Differences at Biological Levels Concentration fluctuations in the chemical composition of plants are due to genetic variability, different climatic and epidaphic conditions etc. which determine the differences among species, populations and stands (Markert, 1993). These factors are valid for both, higher and lower plants. Vascular plants, in comparison to bryophytes and lichens, have a well developed root and transport system. Therefore, they have a mixed mechanism of accepting nutrient and polluting elements from the surrounding environment which increases the importance of the processes of localization, transport and relocali7ation of substances in the plant. Therefore, the concentration differences at individual and compartment levels are much more important and should be considered with due attention. At equal microclimatic and edaphic conditions the most important factor for individual differences is the stage of development of the plant (namely its agc) which determines among other things the degree of exposure, the direction and the level of transportation and localimtion. Individual differences for some elements (mainly nutrients) have been reported to vary about 20% for Digifulispurpurea and Senecio syluaticus (Ernst, 1990), about 2 5 O h for Tiirusucutn officnalc (both, for nutrient and heavy metals) (Djingova and Kuleff, unpublished data), higher values have been also established in some bushes (Markert, 1987b, 1993). To overcome the influence of the individual variations two approaches have been in use. One is to define the number of individuals to be collected in order to mask the interindividual fluctuations and obtain a representative sample and the other is to define exactly the sampling parameters with respect lo age, height, position and exposure, the second approach being better developed in the case of trees. On the basis of typical individual elemental fluctuations between l0-50% it is recommended to collect samples from at least 10 but preferably 25 plants for chemical analysis (Markert, 1987b, 1993; Ernst, 1990). In the case of tree species excellent standardized sampling procedures have been developed for Popultis nigra (Wagner, 1987) and Pirca ahies (Knabe, 1984) specifying the height (4-6 m for Popufus nigra) and the age (7 year-old branches

Sumpling of Vusculur Plunts

397

for Picea abies), the age of the twigs (new-grown long sprouts from each branch for Populus nigra and 1 year old twigs for Piceu abies), and the position of the leaves/needles on the branch that are finally to be sampled. In this way the age and height factors are strongly reduced and comparative studies are possible. Additionally, to account for exposure and shading, sampling from solitary trees, clumps or lines of trees is standardized. Although it is considered that deposition of pollutants may raise with height (Wagner, 1990) the investigations of Zimmermann (1989) proved that in the case of Fugu,r siluutica L. the highest concentrations of heavy metals (Pb, Cd, Zn) are found in the lower inner parts of the trees. This difference in the observations may in our opinion, be due either to interspecies differences or to the fact that the investigations of Zimmermann (1989) were performed in areas not directly affected by pollution and that the presence of heavy metals in the trees (Fagus silvutica L.) is due only to the available quantities in the soil (not air deposition) and that the localization of the elements in the lower part is explained by the intensive water use. In two aspects, the inhomogeneous distribution of elements in the different organs of a single plant is very important for vascular plants. Firstly, the analysis of organs may give valuable information about the degree of exposure (Tyler et al., 1989) and the ways of pollutants entering the plant (Kabata-Pendias et al., 1991). Secondly, on this basis the most suitable “critical” organ of a plant may be defined for use in monitoring studies. Of course, in biomonitoring studies using tree leaves have been favored with the correct assumption that they can be obtained high above the ground, not subjected to soil contamination and splashing, and are exposed to direct atmospheric pollution (Martin and Couphtrey, 1982). But still there are investigations of the heavy metal distribution in other tree organs such as roots, twigs, seeds. Of course, roots of older trees are seldomly considered since in polluted areas the major part of the tree’s root system is likely to be in the less contaminated soil horizons (Martin and Couphtrey, 1982). Monitoring using tree bark will not be discussed here. An excellent review is to be found in (Walkenhorst et al., 1993). The results from the investigations indicate that there is certain difference in the behavior of deciduous and conifer tree species with respect to the distribution among twigs and leaves (needles) (Steubing and Haneke, 1993). Steubing and Haneke (1993) investigated the distribution of mainly U but also of Ce, Co and Au in Pinus syluestris, A h u s glutinosa, Carpinus betulus and Quercus robur, grown on uranium-rich soils and conclude that for conifers the elements are located in the twigs during the first two years. Sheppard et al. (1985) investigated the distribution of practically the same elements in Scots pine grown in waste-side soils and proved the accumulation of the elements in the roots. The study of Wotton et al. (1986) on Picea hunksiana and Piceu nzurinu in areas subjected to smelter air pollution with Ni, Cu, Fe, Pb and Cd established enrichment and localization of the elements also in the roots (seedling roots) and in the cones but not in seedling shoots, the concentrations decreasing with the distance from the smelter. Wagner (1990) reported higher concentrations of As i n spruce needles with respect to shoot axes, lower Pb, Co and Ni and similar Cd and Zn concentrations.

398

R. Djingorcr und I. K u k f f ’

Thus it seems that with conifers, no matter whether soil or air pollution is investigated, most of the heavy metals are enriched and localized in the roots and twigs and lower concenrations are detected in the needles. This assuniption is indirectly supported by the works of Krivan and Schaldach (1986) who demonstrated that the content of “washed” needles (after removal of all mechanically sticking particles) is not very high in polluted areas. On the other hand needles seem to be an excellent “passive” collector of air pollutants (to be discussed later). Rather contradictory to this assumption is the observation of Cousen (1990) who reports highest Cs-I37 contents i n young needles (grown a year after the Chernobyl accident) but as he remarks seemingly afterwards this effect of initial localization in young needles is to be repaired. For deciduous trees and bushes Steubing and Haneke ( I 993) report localization of U in the leaves. Martin and Couphtrey (1982) discuss that the concentration of Cd is higher in the leaves of Corylus uvellunu, but that of Pb is significantly highcr in small twigs while Zn and Cu are evenly distributed. Investigating organ distributions within different vascular plants from a birch forest area near a Cu smelter Lobersli and Steinnes (1988) prove that Betulnpuhescens is a good indicator of environmental pollution. Cu and Zn in the leaves and Cu. Zn, Cd and Pb in the twigs are decreasing with the distance from a smelter (displaying monitoring properties). Studying Betulu species as well Taha and Greim (1 979) demonstrated higher concentrations of Co, Ga and Au in the lcavcs and of As, Bi, Sb and La in the twigs and even distributions of Cd and Fe. Thus it seems that Pb, and probably As and Sb, are located in the twigs of deciduous trees and shrubs while there is strong evidence that U, Co and Cd are found in higher concentrations in the leaves, irrespective of the type of pollution. With grasses and herbage, the investigations of different organs concern the distributions of elements in roots, stems, leaves and seeds. Thornton (1983) studying the influence of soils heavily contaminated with Cd and Zn found highest concentrations in the roots of the grass Holcus lunutus and a decreasing concentration gradient from root to stem, leaf and head. At the same time he cstablished high Cd contents in the above-ground parts of Turuxacwn officinulr growing on the same soil. Kanias and Philianos (1 978, 1979) found highest conccntrations of Br, Th, Ba, Cs, Rb, Eu, C1 in petiols, of Sc, As, Zn, Cr, Fc, Hf, and La in the rhizomes and only of Au in the leaves of Hellehorus cyclophyllu,~ BIOSS. The petioles selectively accumulated Br, their content being 4 times higher than in the soil. In a detailed investigation of R1ziizanthu.s angustijulius Ernst (1 990) studied thc distribution of ninc nutrient elements in roots, stalk, leaves, calix, petals, capsule and head and established highest values of P, Na, Ca, Mg, Mn and Zn in the leaves, of P and N in the seeds and only of Fe in the roots. The investigations of herbage are mainly directed towards studying the distribution between roots and aerial parts aiming to discover the ways pollutants are entering the plant. In this respect the pot experiments of Aspiazu and Romero (1986) with Loliurn rnicltiflorum grown on perlite with different additions of heavy mctals are very intcresting. They proved that using smaller additions all elements (Zn, Co, Ni, Cd,

Sampling qf Vascular Plants

399

As, Pb) were accumulated in the roots, but with higher additions top uptake increased and, except for Pb and Cu, the elements were translocated and stored in the above-ground parts. Analyzing several weed species collected near the Metallochimia Factory (near Budapest) Kovacs et al. ( 1993) established highest concentrations of Zn in the leaves, of As and Pb in the roots of Ambrosia elutior, Atriplox tuturicu and Convolvulus urvense. The concentrations of Cd and Ni generally were similar in leaves and roots. In the case of LoZium perenne roots accumulated Cd, Cr, Cu, Fe, Pb and Zn. The presented accumulation capacities of the investigated weed species proved that the root systems of Lolium perenne, Pluntrrgo lunceolutu and Turuxucum offieinale are very effective accumulators of heavy metals. In several papers, Kabata-Pendias et al. (1 989, 1990, 1991) present results on the distribution of Cd, Zn, Cu, Cr, Fe, Mn, Ni and Pb in tops and roots of Turaxucum oflicinule that prove the enrichment of Cd, Zn and Pb in the tops. On this basis the authors come to the conclusion that increased levels are due to airborne contamination. Studying the elemental fluctuations among different organs of Turaxucum officinule we have investigated samples both from background areas (situated in the Rhodopes, the Balkan mountains and Rila mountains in Bulgaria and one polder region in the Netherlands), and from industrially polluted areas (in Bulgaria and in the Netherlands). The results obtained in the background regions are presented in Fig. 1. Higher concentrations of Br, Ca, Cd, Sb, Se are established in the leaves whereas the roots are enriched in Au, Co, Cr, Fe, Pb, Cu, Mn and Hg. Ba and K have highest concentrations in the stems, As in the blossoms, while Na, R b and Zn are evenly distributed. In polluted areas (Fig. 2) the comparison between the concentrations in the organs prove that the critical organ of Turaxucum qfficinule for As, Ba, Br, K, Hg, Sb, Se is the leaf while Au, Co, Cr, Cu, Fe, Mn and Pb are definitely higher in the roots. Zn and Cd have similar values in leaves and roots while in all cases the other organs show lower concentrations. The CR values (ratio between plant concentration and the available content in the soil) prove that As, Br, Pb, Cd and Cu are accumulated in the plant in quantities higher than available in the soil (Djingova and Kuleff, unpublished data). On the basis of these results we have made the assumption that Turuxucum oflicinale is reflecting not only soil but air pollution as well, the leaves accumulating more As, Br, Sb and Se and the roots more Co, Cr, Cu, Fe, Mn and Pb. Zn and Cd are evenly distributed in the roots and leaves. Although the conclusions are similar to those made by Kabata-Pendias et al. (1989, 1990, 1991) the observations concerning Pb, Zn and Cd are rather different since we have not established higher concentrations in the tops with respect to the roots of the plant. We confirm, however, the contribution of aerial pollution to the concentration in the plant, on the basis of the higher content in the plant than the available in the soil and on the basis of the investigations we performed earlier (Djingova et al., 1986) on the mathematical models describing the behavior of the element concentration in Turuxucum officinule in dependence on the distance from a pollution source. The results of the different authors concerning distributions in herbage lead to the more general conclusion that at normal levels, or in slightly polluted areas,

400

5 Root/Leaf

0Blossom/Leaf

Stem/Leaf Seed/Letlf

4 0 CI

I-

a CK

3

z 0 u

I-

a 0:

2

IZ

W

0

z 0 0

1

0

A s Au Ba B r Ca Cd Co C r Cu Fe Hg K MnNa Pb R b Sb Zn Fig. 1. Element distributions i n diffcrcnt organs of Thruxuciwi of7kinale from background regions.

most o f the heavy metals are accumulated in the roots, though the above-ground parts also react to the pollution. (In somc cases, e.g., in Turuxucum oJi:cinale (Kuleff and Djingova, 1984; Djingova et al., 1986; Kabata-Pendias et al., 1989, 1990, 1991) this reaction is very strong.) There are indications that under heavily pollutcd conditions (soil or air) the top uptake increases (Azpiazu and Romero, 1986) and most o f the elements are transferred to the above-ground parts. Of sourse, all these conclusions are valid in the case of washed samples.

5 Root/Leaf

Stem/Leaf

Blossom/Leaf

4 0 U

I-

4

c r 3 Z

0 U

I-

a 2

CK I-

z W

0

z 0 1 0

0 As Au Ba Br Ca Cd Co Cr Cu Fe Hg K MnNa Pb R b Sb Se Z n Fig. 2.

Element distributions i n diffcrent organs of Trrrusircum officinule from polluted regions.

21.3 Seasonal Variations in the Heavy Metal Content of Plants Most investigations on the seasonal changes of the heavy metal contents in vascular plants performed until 1980 are summarized by Martin and Couphtrey (1982). Starting from the earliest investigations of Hunter (1953) and Mitchell (1957) the results seem to be species and element dependent and obviously each case should

402

R. Djingoon mid I. Kukf]'

be studied separately. Nevertheless, some distinct trends can be detected for the different groups of plants. Seasonal changes in the elemental contents of deciduous trees and shrubs have been an object of investigation in a number of papers (e.g., Fidora, 1972; Ahrens, 1964; Smith, 1978; Guha and Mitchell, 1966; Martin and Couphtrey, 1982; Markerl, 1987a, b, 1993; Kovacs et al., 1982; Wagner, 1987; Deu and Kreeb, 1993). If the results are summarized according to Markert (1993) seasonal changes in the elcmcnt concentrations exceed the fluctuations due to site-by-site differences and to the analytical error. (Markert (1 989) established differences of up to 159% within 2 months for Mg, Ca, Sr, Ba and Zn in Vucciniun?myrtillus). In many cases, however, such drastic changes have not been observed. Capannesi et al. (1993) conclude that the seasonal changes of Pb in Quercus ilex as well as of Eu, Sb, As and Hg are far less important than site-by-site variations. Most of the results from the studies of the seasonal behavior of heavy metals (Pb, Cd, Cu, Fe, etc.) in Mulus purnilu, Pyrus comnzunis, Prunus cerusifkru (Deu and Krebs, 1993). Aesculus hippocastanurn (Guha and Mitchell, 1966), Celtis orientalis, Tilia tomentosa (Kovacs et al., 1982) and Populus nigru (Wagner, 1987), lead to the assumption that in the case of deciduous trees heavy metals increase with the season from May to October reaching maximum concentrations directly before leaf fall. There are some contradictory results for specific elements and/or species but generally maximum is detected in September - October and a possible minimum during the preceding summer months. Therefore, as Wagner ( 1 990) concluded, sampling of decidious plants is best to be done in early autumn shortly before scenescence. Studying seasonal changes in conifers Wyttenbach and Tobler (1988) divided the 20 determined elements in 4 groups depending on their behavior. Ahrens (1964) reports remobilization of trace elements in one-year old needles to ensure the growth of new twigs. Ernst (1990) mentions that plants with long-lived needles e.g., Picca ahirs and Pinus syluestris demonstrate a continuous age gradient. Anyhow, the changes of pollutants in washed needles are not so well expressed (Krivan et al., 1987b) as was established for decidious trees. The investigations of grasses prove a quite different behavior. Thornton (1983), Martin and Couphtrey (1982), Matthews and Thornton (1980), Mitchell and Reith (1966) Lindsay and Brokhout ( 1 9781, Wilkins ( 1978), Roberts and Goodman (1974), and Turney et al. (1972) established lowest concentrations of heavy metals in pasture herbage and grass species in spring and highest in winter. However, for non-grass species like Taruxucum qfjicinale (Matthews and Thronton, 1980; Djingova and Kuleff, unpublished data) another behavior was established which is similar to that of Loliumperenne (Crump, 1980),Krimyn grass (Beavington, 1976). Bermuda grass (Brown, 1983), rye grass and robur (Bacso et al., 1984) and rye grass (Fleming, 1986). In all these cases the seasonal changes of Zn, Cd, Ni, Sb, Cu, Pb, etc. show maximum contents in autumn and relatively high contents in spring. Thus in the general case of monitoring heavy metal pollution using vascular plants it seems that sampling of common pasture grass should be performed in winter (of course this may be possible in areas without a stable snow cover), of decidious trees in carly autumn, while for non-grass herbage spring is also possible. With conifers cumulative effects over the years may be expected.

Sumplrng of Vuscukur Piuntr

403

These final conclusions made on the basis of the reviewed literature, are absolutely correct but an additional question should be posed. Is it possible to standardize the exact time period of sampling in the case of comparative studies, including regions with various altitudes and latitudes? Due to climatic variations, differences in the stage of development of the plant are to be expected. This question is directly connected with the proper organization of global monitoring studies using vascular plants. Therefore, it is far more important to standardize the stage of development of the plant instead of the exact season. We applied this approach in the biomonitoring studies with Turuxucum officinule (Kuleff and Djingova, 1984; Djingova et al., 1986; Djingova and Kuleff, 1993). The samples were collected in 4 European countries (Bulgaria, Greece, Germany and The Netherlands) from regions situated from below sea level (The Netherlands) up to 2000 meters above sea level (Rila mountains, Bulgaria). Investigations proved that it is not possible to standardize an exact time period of sampling because when Turuxucum oJficinale was flowering in Thessaloniki (Greece) and Sofia (Bulgaria) it was not yet to be found in the mountains. Therefore flowering was standardized as a stage of growth and samples were collected in the period April - July.

21.4 Sample Pretreatment It is already accepted that in the case of food chain studies cleaning plant material prior to analytical procedures should be avoided while in physiological investigations it is an obligatory step. Cleaning may be performed in several ways (Markert, 1993), either dry or wet brushing or wiping or washing in different media. In biomonitoring studies final conclusions about the necessity of preliminary cleaning of plant inaterial have not been reached yet and a lot of investigations are being done with unwashed samples or after using various washing procedures which makes comparative studies rather difficult if not totally impossible. The cleaning of plant material in biomonitoring is usually discussed according to the following aspects: (1) Avoidance of soil contamination. This is the main aim of pretreatment of plant samples when analysis must provide information about heavy metal pollution, and grass and herbage are used as monitors. However, it is an important step in the treatment of tree leaves as well, especially when regions situated in the more southern parts of the Northern Hemisphere are monitored. The soil dust content in the atmosphere is very high during summer and autumn months and precautions should be taken to avoid an influence on the monitoring results. (2) Establishing a difference between aerial and soil pollution. (3) Reduction of the effect of drastic climatic changes.

Regarding the second aspect it is usually considered that atmospheric pollution is reflected by plants by foliar deposition, which more or less is physically sticking

404

R.Djiugovn mu’1. KulcIff’

of particles onto the leaf surface. Therefore, washing (or mechanical treatment) is used to assess (or discard) that part of pollution which is attributed to the atmosphere. It is, of course, not always true that atmospheric pollution is reflected by plants only as deposition. Due to climatic changes (draught periods and rainfalls) a lot of the deposited elements are transformcd to plant available forms and either enter the plants directly through the leaf stomata or are leached in available form to the soil and taken up by the plant via the roots. Additionally, there are a lot of processes with vascular plants including the allocation and transportation of the elements within the plant that further contradict this simplified assumption. Tab. 1 presents a brief overview of the washing procedures more often used in biomonitoring studies. It is clear that each treatment leads to different extents of removal of particulate matter. The situation is further complicated by the fact that for the various plant types different washing is proposed and even if one and the same procedure is applied it may lead to contradictory results. The most aggravating Tab. 1. Sample Pretreatment in Biomonitoring with Vascular Plants Type of treatment Mechanical treatment 1. Wiping with cloth

2. Blowing with nitrogen jet Washing I . Organic solvent and water rinse

2. Rinsing with distilled (tap) water

3. Tap (distilled) water and detergent 4. Acid solutions

5. Complexing agents

Effect

Reference

N o changes in Ca, Na, Cu, Mn; changes in Al, Fe, Si, Ti, Pb Cleaning off all particlcs

Buchauer, 1973; Camerlynck et al., 1980; Hall et al., 1975; Wallacc ct al., 1980 Capaunesi et al., 1993

Removal of Al, Pe, Sc, As, Sb; negligiblechanges ofCa, CI, Mg, Mn, Zn Removal 10- 80% of Pb, smaller changcs for othcr heavy mctals

Hartmann and Baechmann, 1988; Krivan et al., 1986; Wyttenbach et al., 1985 Aidid, 1988; Bacso et al., 1984; Camerlynck et al., 1986; Carlsoii et al., 1976; Deu and Kreeb, 1993; Ernst, 1972, 1978; Fidora, 1972; Fytianos et al., 1986; Hagcmcyciet al., 1985; Lerche and Brccklc. 1974; Martin and Couphtrcy, 1982; Schuck and Locke, 1970; Wagncr, 1987; Wissmath, 1982 Camerlynck et al., 1986; Kowalenko, 1984

Changes in macro- and micro-nutrients and in Sc, Cs, Zn, Fc Solubilizatioii and removal of small particles; reproducible removal ;leaching of P, Mg, K ; spccial conditions for Cd

Cainerlynck et al., 1986; Hagcmeyerct al., 1985; Kcller and Preis, 1967; Little, 1973; Lindberg and Harris, 1981 ; Wallacc et al., 1980; Wood and Bormann, 1975; Yainada ct al., 1964 Martiiicz et al., 1971 ; Pricbe et al.. 1981

Sumplinl: cf Vusculur Plunts

40 5

observation is that the same treatment applied to the same plant by different authors often gives different results. Nevertheless, we will try to present (as far as possible) the status of washing pretreatment of plants in biomonitoring systematically. With conifers these problems have been studied and clarified rather well. Investigating the major and trace element contents of Picea abies Wittenbach et al. (1985) used a cleaning procedure of an 1 min. treatment with chloroform and rinsing with water. A considerable amount of the waxy material was removed from the needles, one forth of which was attributed to mineral substances. The elements were divided in four groups according to the fraction in the removed material. The highest content was established for the dust and soil forming elements Al, Fe, V, Sc, As and Sb (about 80%). The second group included Br, Co and Na (with 45-50% in the removed surface fraction) followed by Ca, C1, Mg, Mn and Zn ( 1 - l0Y0). The smallest content (below 1%) was established for K. On this basis the authors conclude that washing with chloroform does not remove elements from the interior of the needles (negligible K) and that the values obtained without washing are often more representative for the dust than for the plant itself, and therefore are useless which is especially true for trace elements. For older needles the situation is even worse. In a series of papers Krivan et al. (1986, 1987a, b, 1989) devote their attention to the different washing procedures of needles. Comparison is performed between unwashed needles, and treated with water or chloroform or by the stripping technique. The result proved that for the elements B, Ba, Ca, Co, Cr, K, Mg, Mn, N, Rb, S and Zn the difference between treated and untreated needles is insignificant. Treatment with water removes Al, Pb, Sb, Th, Fe and Sc to a different extent. The effect is deepened by using chloroform and the results are more reproducible. Hartmann and Baechmann (1988) investigated the same problem using similar washing procedures (treatment of needles with chloroform for I min, and with water for 2.5, 10 and 20 min). The elements were divided in 3 groups: -

-

K, Rb and Mn were more efficiently removed by water; Ca and Zn were similarly removed by water and chloroform; Pu and Fe were efficiently removed only by chloroform.

The most interesting result of all these experiments is that needles react quantitatively to atmospheric pollution and that the deposited layer on the surface is representative. It was proved that the wax layer works as a collector of atmospheric deposits (Krivan, 1986) while after its removal the content of Cu, Fe, Pb, Sb, etc. in the tissue remains relatively constant in needles of different age, and even (for some of the elements) in needles from different sites and soils. Therefore, it is quite obligatory in the case of physiological and toxicological investigations with conifers to analyze separately the deposited and the tissue fraction. Since most of the experiments prove that water treatment of the needles is not very effective. (Even Krivan and Schafer (1 989) reported that heavy raining did not cause differences in the deposited matter.) Treatment with chloroform is the choice in these cases. As far as washing for biomonitoring is concerned on the basis of the results reported by Krivan and Schaldach (1 986), Krivan and Schifer (1989) and Wyttenbach et al. (1985) it seems that such severe treatment is not obligatory.

406

R. D J ~ I I ~ trnd OUU I. Kirleff

A number of cleaning techniques have been used i n the analysis of dcciduous tree leaves and bushes. Rinsing with distilled water (Schodeller, 1967; Schuck and Locke, 1970; Lerche and Breckle, 1974; Carlson et al., 1976; Ernst, 1972, 1978; Fidora, 1972; Hagemeyer et al., 1986; Wissmath, 1982; Wagncr, 1987; Deu and Kreeb, 1993; Aidid, 1988), mechanical treatment for removal of adhering particles (Buchauer, 1973; Hall et al., 1975; Wallace 1980), treatment with acid solutions (Keller and Preis, 1967; Little, 1973; Lindberg and Harris, 1981; Wallace, 1980; Yamada etal., 1964; Wood and Bormann, 1975) and with complexing agents (Priebe et al., 1981; Martinez et al., 1971) have been tested. The behavior of Pb during the different washing procedures has been intensely investigated and the majority of authors report a 50- 80% decrease in concentration even after simple rinsing with water (Martin and Couphtrey, 1982; Carlson et al., 1976; Ernst, 1972, 1978; Wissmath, 1982)but lower values also have been established in a number of cases (Goodman and Roberts, 1971 ; Leh, 1965, Pfeilsticker and Markard, 1975; Wagner, 1987; Deu and Kreeb, 1993). The changes of other clcments, heavy metals and nutrients have been shown to be not that drastic as of P b after cleaning. Kowalenko (1984) found differences in the analysis of washed and unwashed filbert orchard leaves for macro- and micronutrients, except for K and Ca. Since washing reduced leaf weight, contamination was attributed to particulate matter, washing including scrubbing with a soft brush using phosphate free soap, followed by two rinses and a final rinse with deionized water. Kowalenko (1 984) concluded that washing filbert leaves prior to analysis is essential for proper interpretation of Concentration data. Wagner (1987) analyzing Populus nigru leaves applied three 30 s rinses, the last with deionized water and established removal of Pb up to 40% and no significant changes for Zn and Cd. Finally Wagncr (1987) analyzed unwashed leaves. Hagemeyer et al. (1985) found only a minor effect in lowering the adsorbed fraction of Cd off the leaves after rinsing with demineralized water at pH 5.5. Cadmium was most efficiently removed from the leaves of Fugus silvutica L. eithcr at low pH (3.5) or high pH (9.0). Deu and Kreeb (1993) established a significant reduction of Fe in orchard tree leaves (Mulus purnilu, Pyrus communis and Prunus cerusiferu) after washing with tap water. Lead and copper were reduced only in Prunus cerasifkra while Mn and Zn were unaltered in all thrce specics. After comparison of washed and unwashed lcavcs from trccs in polluted and unpolluted areas the authors concluded that metals are easily deposited on the leaves of Prunus cerusiferu but adherence is weak and most probably metals are not adsorbed by the leaves, while the behavior of Mu1u.s puwiila is opposite. Aidid ( 1 988) investigated the difference between washed and unwashed leaves for 17 elements in two families of tropical plants. Washing included 3 rinses with doubly destilled water followed by soaking in doubly destilled water for 30 min. Thc cffcct of washing was more significant for Sb, Zn, Cs and Fe but not for toxic elements as As, Hg, Au, Br, etc. The cleaning was chosen as to resemble the way Malaysian housewives prepare vegetables for cooking.

Snmpling

of

Vnsculuv Plant.\

401

Obviously washing only with water is not very effective in removing heavy metals deposited on the surface of leaves of deciduous trees. Little (1973), however, demonstrated that washing with dilute nitric acid is advantageous in solubilization of small particles and that acidification leads to reproducible results. Yamada et al. (1964) presumed that heavy metals may be held on the ion-binding sites on the cuticle and therefore may be removed only by acidified agents. Wallace et al. (1980) compared the effects of: -

wipe with cloth; wipe plus detergent and acid (0.1 N HCI) wash; acid wash; wipe and acid wash, followed by washing with deionized water; a control untreated group of orange leaves.

Ca, Na and Cu and Mn were not changed by any procedure. P, Mg and K were leached by severe washing. Al, Fe, Si, Ti and Pb were removed to a different extent by all washing procedures but most reproducible results were obtained after wiping with acid. An example of quite a different technique of surface cleaning is the investigation of Capannesi et al. (1993) of 23 elements in Quercus ilex which proved (by microscopic determination) that blowing with a strong nitrogen jet is sufficient to clean off all adhering particles. This brief review demonstrates that cleaning deciduous tree leaves from adhering particles and deposited material is rather complicated and obviously dependent on the type and structure of the leaves, the type and form of the pollutants, etc. This is why the effect of one and the same procedure may be different when applied to various tree leaves. It seems that weak acid wiping is a promising cleaning technique in these cases. Attention may be paid to the proposed nitrogen jet blowing (Capannesi et al., 1993). The analysis of unwashed herbage material is further complicated not only by the foliar deposition problems already discussed, but mostly by the possibility of soil contamination. It is well known (though often disregarded) that due to rain splash or especially in areas with high soil concenrations of heavy metals soil contamination of herbage samples can be very problematic (Martin and Couphtrey, 1982). Usually, to assess the possible soil contamination, the determination of elements with a high soll/plant ratio is recommended (Mitchell, 1960; Fleming et al., 1986). As such elements Ti, Al, Fe, Cr and Co are being recognized (e.g., Berrow, 1988). In most investigations, however, only elements as Pb, Cd, Zn, Cu are determined and no assessment of the soil/plant ratio and of contamination is performed. According to Klocke and Riebartsch (1964) simple washing with water removes more than 80% of Pb from herbage. However, Garber (1970) established only 40% removal of Pb in this way. On the other hand Bacso et al. (1984) report a decrease of Pb and Br below 10% (with maximum 20% in single cases) after 5 min of shaking in deionized water of Lolium perenne, Trifolium, Taruxucum oficinale, Polygonurn Liviculure, Quercus rohur and Plutanus. Investigating the contamination of roadside vegetation with Pb, Cd and Zn, Fytianos et al. (1985) report a decrease of up to 60% in Zn and up to 40% in Pb

R. Qjingova and 1. KuIcfi

408

after simplc water rinsing of grass samples. N o significant changes in the Cd content were detected. The difference of washed and unwashed samples is strongest nearest to the road and diminishes with the distance from its edge. Cammerlynck et al. (1986) make the following recommendations for smaller leaves: washing with deionized water and detergent or only with destilled water; wiping with dry or damp cloth; - dipping for 30 s in 0.01 N HCI;

-

-

and for larger leaves: -

wiping with dry or damp cloth; dipping in 0.01 N HCl.

A comparison of the results from the analysis of Turaxacum offkinale leaves washed with tap and destilled water (for I min); dipped for 30 s in dilute nitric acid (pH = 3); - untreated;

-

-

is presented in Fig. 3.

r

50 ffl I-

-

unwa#hed

z 3

w

-

w a s h e d w i t h water

with d i l u t e d acid

t

40

> Id J

w

30

[2

z z

0

20

I-

d

rr: I-

z

10

w 0

z 0 0

Ca

Fig. 3.

Cd

Cu

Fe

K

Mn

Na

Pb

Effects of diffcrcnt washing procedures on Turcc.uucum officinule leaves.

Sampling of Vusculur Plunts

409

The samples were collected in the city of Sofia (Bulgaria). Washing with water brings reduction in the contents of Pb and Mn (about 25%) and practically no changes in the contents of Cu, Cd, K, Na and Ca. The acid treatment reduces strongly the contents of all determined elements with the exception of Cd. The significant decrease of K and Ca in this case means that most probably there is leaching from the interior of the leaves and not only removal of foliar deposition and contamination. The washing of samples is, in our opinion, closely related to a problem that has been mentioned several times in this contribution, namely the changes that may appear due to intensive and/or prolonged rainfalls. These changes have been acknowledged and discussed sparingly (Rentschler, 1977; Martin and Couphtrey, 1982). The effect of rain on the heavy metal concentrations of plants used in biomonitoring studies has two aspects. Firstly, it washes off part of the deposited metals from the surface of the leaves. Cawse (1980) reports mean water solubilities of Zn, Se and Br of 80% due to rain water, of A1 and Fe of 25%. On the other hand, as Lindberg and Harris (1981) point out, rain waters are enriched with heavy metals originating from deposited particles. Especially acid rain (Heinrichs and Mayer, 1980) plays an important role in the cycling of elements by transforming them into soluble forms and increasing their availability to plants. This, however, is rather a long-term effect. The immediate result from rainfalls in sampling may be generally detected as a decrease in the content of some elements due to washing off. As a result of this, comparison of the results for samples collected after draught and raining periods different in duration is not possible or very risky. In practice this has been investigated by Ho and Tai (1979) who determined the lead content in the leaves of Alocasia odora and Mikania guaco in dependence on rainfalls during a four months period. Two weeks of dry period preceded the first sampling, after only 13 mm of rain the content of Pb dropped 3 times and resumed its initial content after 6 days. At the end of spring after heavy rainfalls the lead content reached stable values which was assumed to be the lead content of the plants themselves or the so called “non-washable” fraction that corresponds to the portion of lead that cannot be washed with distilled water (Motto et al., 1970; Page, 1971; Little and Wiffen, 1977) or by simulated rain (Carlson et al., 1976). Thus Ho and Tai (1979) accept that the effect of rain is equal to the effect of water washing and simultaneously recommended sampling after at least one week of drought perioid (but not washing of the samples). The data from our experiments with Taraxacum qficinale collected in Sofia (Bulgaria) after 3 days of heavy rainfall are presented in Fig. 4, in comparison to samples collected the day before the rainfall after a 3 weeks dry period. Part of the latter are washed by tap and distilled water and the rest are analyzed without washing. It is obvious that the values for all elements are very similar between the water washed samples and those collected after the rainfalls (three days later). In both cases the values for Pb and to a certain extent Mn and Fe are lower than in the unwashed material, while Cd, Ca, K, Cu and Na are stable. Therefore, washing of herbage with tap and distilled water seems to reduce soil contamination and rain effects while acid treatment most probably leaches elements from the inside of the leaves or generally as Benton Jones (1971) pointed out “rinsing

R . Dlingoi:~utid I. Kulrff

410 50 v)

=

0after rain

washed with water

unwashed

kb.-

z 2 W

40

-

>

c a

-I W

30

[L

z M

z 0 20 3

F Q [L

t

ia W

0

z 0 0

0 Ca

Fig. 4.

Cd

Cu

Fe

K

Mn

Na

Pb

Cornparisoil of washing and rain cffccts on thc analytical rcsults or TU~USUCUIIZ leaves.

in running water will remove most attached substances. The washing procedure should not be prolonged or the plant material allowed to ‘stand’ in either washing or rinsing bath”.

21.5 Discussion The aim of this necessarily incomplete survey has been to stress the complexity and controversity of the problems concerning representative sampling of vascular plants for biomonitoring purposes. We have only discussed problems concerning differences (i.e., sources of error) due to individual and compartment biological levels, time of sampling and pretreatment of the samples (which in our opinions is definitely a sampling step since in most cases it should be done immediately after sampling in the fields and its being done or not changes the meaning of results), all of which are in a way connected to a very important problem in biomonitoring studies, namely what type ofhcavy metal pollution can be determined by using vascular plants - aerial or soil or both‘? As was mentioned already in the text the usual approach is to accept that deposition onto the leaf surface is representative of aerial pollution and the

Sumpling of' Vasculur Plunts

41 1

concentrations inside the plant of the soil conditions. If this assumption is accepted then sampling may turn out not to be such a difficult task after all. Generally, then it might be recommended to collect leaves (they are mostly subjected to deposition), just before scenescence (in most cases the deposited particulate matter is highest at that time) and to apply severe removal of all deposited particles to be analyzed for aerial pollution. However, such an approach may be contradicted by two arguments. Firstly, there is no case where continuous air pollution has not lead to simultaneously high soil pollution. (Elevated concentrations in soil without respective air pollution are well known but this is more in the line of geobotanical and geochemical studies.) Combining this fact with the mixed mechanism of accepting nutrient and polluting elements of vascular plants it becomes quite clear that in most cases the above mentioned division of aerial/soil pollution reflected by plants is too gross a simplification. Additionally, if rainfalls and other metereological conditions are considered the question becomes even more complicated. Therefore, we have tried to stress in the text that leaves are not always the organs collecting highest concentrations of pollutants and for biomonitoring other organs may be more useful in some cases. But if, for the sake of argument, it is accepted that the division into aerial and soil pollution is correctly represented by on/in plant concentrations (such cases have been reported) then ofcourse a question may be posed. Is it worth using plants as passive collectors of aerial pollution? Is it not much simpler and representative to use directly fallout sampling procedures and analysis? The main advantage of biomonitoring over direct methods for monitoring is that it gives an excellent, representative and relatively cheaper opportunity for assessment of the conditions of the ecosystem and prognostication for its future development, and for risk assessment. Additionally, it enables relatively cheap large-scale screening. All this means that data should be collected for what is in and fixed on the plant but these data should be reproducible, representative for the ecosystem and comparable both within the ecosystem and outside it, which leads to the big question of the pretreatment. While it is already possible to give some general recommendations concerning the minimum number of individuals to be sampled, the plant organs (for certain species), the season of sampling, etc. it is yet extremely difficult (if not impossible) to recommend a general procedure for cleaning. Though the purpose of cleaning in biomonitoring has been divided and subdivided into different points in fact, in our opinion, it is to remove contamination and loosly adhering particles so as to get reproducible, representative and comparable results. Therefore, cleaning should lead to the removal of direct soil contaminations in grasses and herbage and of soil-dust contaminations on tree leaves both of which are dependent on the type and state of soil, local metereological conditions and are easily changable by rainfalls. Evidence for such contaminations are higher and changable after water treatment concentrations of elements like Sc, Th, Al, Fe etc. Pretreatment should more or less copy the effect of metereological conditions leading to changes of the deposited fraction and to make results independent of the local factors. This does not simplify very much the question but it gives a direction in which to search for unification.

412

R. Djingoau and I . Kukff‘

The view we are trying to present here is that in order to prcscrvc the main advantages of biomonitoring over direct methods, namely a relatively quick, cheap and most of all representative and comparable evaluation of the state of ecosystems, pretreatincnt should be carried out as to remove contamination and losely adhering particles more or less modelling the action of meteorological factors. (This is not valid for physiological and food-chain investigations.) In many cases, this may turn out to be only water treatment or gas blowing but to come to a conclusion in this respect it is obligatory to investigate parallel to the usually studied heavy metals the behavior during washing of elements like La, Sc, Sm, Th, A1 etc. In practice this means that for a monitoring investigation i t is not sufficient to determine only the content of Pb, Cd, Hg, Zn etc. but also of “neutral” elements as the rare earths, Sc, etc. Their content in/on the plant is dependent on the type of soil, and the dust content of air. This undoubtedly implies a new approach for biomonitoring studies. Obviously the analytical methods routinely used up to now (e.g., AAS) do not provide such a possibility and combinations with other analytical techniques should be looked for. Only after that, after correct evaluation of what part of soil particles and dust are reliably removed by the different media off different plant types a more definite conclusion about the most appropriate pretreatment of plants may be achieved.

21.6 References Ahrcns, E. (1 964) Allg. Forst- 1 1 . Jugdztg. 135, 8 - 16. Aidid, S. (1988) J. Rariioanul. Nucl. Chenr. 120, 335-344. Azpiazu, M., F. Romcro (1986) Water Air and Soil Polliit. 28, 1-20, Bacso, J., M. Kis-Varga. P. Kovacs, G. Kalinka (1984) J . RadioanaI. Nucl. Chern. 81, 59-65. Beavington, F. (1976) Plant and Soil 45, 283 -286. Benton Joncs, J. Jr., K.Large, D. Pfleiderer, H. Klosky (1975)CropsandSoilMu~azine23, 15- 18. Berrow, M. (1988) A i i u l j t . Proc. 25, 116-118. Berrow, M., J. Burridge (1984) in: Metalli, in dw Urnwelr. E. Merian (ed.). VCH, Weinheim, pp. 125-133. Brown, K., J . Thomas, J. Slowey (1983) Wuter Air and Soil Pollut. 20, 431 -446. Huchauer, M. (1973) Enuiron. Sci. Techno/. 7, 13 1. Camerlynck, K., L. Kiekens (1986) in: Sunipling Prohlernsfor the Chemicul Anulysis ($Sludge, Soil and PIants. A. Gomez, R. Leschber, P. L’Hcrmitc (cds.). Elscvier Appl. Sci. Publ., London, p. 45-51. Capannesi, G., A . Cecchi, G . Ciavola, A. Scdda (1993) J . Rudioonol. Nucl. Chem. 167,309-320. Carlson, R., F. Bazzaz. J. Struckel (1976) En‘nairon. Sci. Techno/. 10, 139- 142. Cawse, P. (1977) Proc. Con$ Inorgunic pollution und Agriculturc, London, p. 22 -46. Chapman, S. (1976) Methods in PIunt Ecdogy, Blackwcll Sci. Publ., Oxford. Cousen, G. (1990) in: Element Concrntration Cudusters in Ecosystems. H. Lieth, B. Markert (eds.). VCII. Weinheim, p. 303. Djingova. R., I . Kulefl‘ (1993) in: P l m t s us Biumonitors. B. Markcrt (cd.). VCH, Weinheim, p. 435 - 460. Djingova. R., I. Kulcfi; 1. Penev, B. Saiisoni (1986) Scicwce Totul Enuiron. 50, 197-208. Dcu, M., K. Kreeb (1993) in: Pluntsus Biomonitors. B. Markert (ed.). VCH, Weinheim, p. 577- 592. Ernst, W. (1972) Ber. Deutsch. Bot. Ces. 85, 295-300. Ernst, W . (197X) Chcm. Rtlsch. 31, 32-35.

Sampling .f' Vascular Plants

41 3

Ernst, W. (1990) in: Element Concentration Cudasters in Ecosystems. H. Lieth, B. Markert (eds.). VCH, Weinheim, p. 17-40. Fleming, G., H. Tunney, E. O'Riordan (1972) in: Sampling Problems f . r the Chemical Anulysis of Sludge, Soil and Plants. A. Gomez, R. Leschber, P. L'Hermite (eds.). Elsevier Appl. Sci. Publ., London, p. 6 - 17. Fidora, B. (1972) Ber. Deutsch. B o f . Ges. 85, 219-227. Fytianos, K., G. Vassilikiotis, V. Samauidu (1985) C'henzosphcw~14, 271 -277. Garber, K. ( I 970) Landw. For.schung 25, 59. Goodman, G., T. Roberts (1971) Nature 231, 287-292. Guha, M., R. Mitchell (1966) Plunt and.roi1 24, 90- 112. Hagemeycr, J., H. Kahle, W. Breckle (1985) Water, Air and Soil Pollut. 29, 347-359. Hall, C., M. Hughes, N . Lepp, G. Dollard (1975) Proc. Symp. Heavj~Metals in the Environment, Toronto, Vol. 2, p. 221 - 245. Hartmann, G., K. Baechmann (1988) J . Environ. Ruciinactivity 8, 21 -36. Heinrichs, H., R. Mayer (1980) J . Environ. Qual. 9, 11 1 - 118. Ho, Y . , K. Tai (1979) Bull. Environ. Contam. Toxicol. 23, 658-660. Hunter, J . (1953) J . Sci. Food Agric. 4, 10-20. Kabata-Pendias, A , , S. Dudka (1989) h'nviron. Geochem. Health 11, 19-24. Kabata-Pendias, A . , S. Dudka (1990) in : Element Concentration Cadasters in Ecosystems. H. Licth, B. Markert (eds.). VCH, Weinheim, p. 265. Kabata-Pendias, A , , S. Dudka (1991) Emiron. Geochem. H d t h 13, 108- 113. Kanias, G., S. Philianos (1978) J . Radioanal. Chem. 46, 87-93. Kanias, G., S. Philianos ( 1 979) J . Radiouiial. Chem. 52, 389 - 397. Keller, T., H. Preis (1967) Schweiz. Z . Forstwesen 118, 143. Klocke, A., K. Riebartsch (1964) NaturM,issenscIzuften 52, 367 -368. Knabe, W. (1984) AZF39, 447-484. Kovacs, M., 1. Opanszky, P. Klincsek, J. Podani (1982) in: Monitoring of Air Pollutanfs by Plants: Methou's and Problems. L. Steubing, H. Jaeger (eds.). Junk Publ., The Hague, p. 149- 154. Kovacs, M., G. Turcsanyi, K. Penksta, L. Kaszab, P. Szoke (1993) in: Plants as Biomonitors. B. Markert (ed.). VCH, Weinheim, p. 495-506. Krivan, V., F. Schaefer (1989) Fr. Z . Anal. Chem. 333, 726. Krivan, V., G. Schaldach (1986) Fr. Z. Anal. Chem. 324, I58 - 167. Krivan, V., R. Hausbeck, G. Schaldach (1987a) Fr. Z. Anal. Chem. 327, 19. Krivan, V., G. Schaldach, R. Hausbeck (l987b) Natuwissenschuften 72, 242-244. Ku, H. (1979) N B S Special Publ. 519, 1-6. Kuleff, I., R. Djingova (1984) Water Air and Soil Pollut. 21, 77-85. Leh, H . (1966) Verbraucherdienst 11965, lO/li, 53-57. Lerche, H., S.-W. Breckle (1974) Angeit'. Botanik 48, 309. Lindberg, S., R. Harris (1981) Water, Air undSoil Pollul. 16, 13-21. Lindsay, S., T. Bookhout (1976) Bull. Environ. Contam. Toxicol. 19, 360-364. Little, P. (1973) Environ. Pollut. 5 , 159- 163. Little, P., R. Wiffen (1977) Atmos. Environ. 11, 437. Lobersli, E.-M., E. Steinnes (1988) Water, Air und Soil Pollut. 37, 25-39. Markert, B (1987a) in : Jahrestugung der Gesellschafi ,fur. Okologie. Gottingen. Markert, B. (1987b) in: 4th CoN. Atomsjiektromr(risch~Spurenanalytik. B. Welz (ed.). Bodcnseewerk, Perkin Elmer, Ueberlingen, p. 385 - 392. Markert, B. (1988) Arzgew. Botanik 62, 343-353. Markert, B. (1989) Fr. Z. Anal. Chem. 333, 1 I - 14. Markert, B. (1993) in: Plants us Biomonitors. B. Markert (ed.). VCH, Weinheim, p. 65-104. Markert, B., R. Steinbeck (1988) Fr. Z . Anal. Cheni. 331, 616-619. Martin, M., P. Couphtrey (1982) Biological Monitoring of Heavy Metal Pollution, Appl. Sci. Publ.. London, New York. Martinez, J., M. Nathany, V. Dharmarajan (1971) Nature 233, 564. Mathews, H., I. Thronton (1980) in: Trace Substances in Enviroiimental Health. (D. Hemphill (ed.). University of Missouri, Columbia XIV, 478 -488. Mitchell, R. (1960) J . Sci. Food Agric. 11, 553-560.

414

R. Djingorvi

ciritl

I . Kult.ff

Mitchell, R., J. Reith (1966) J. Sci. Food Agric. 17, 437-440. Mitchell, R.. J . Keith. I . Johnston (1957) J . Sci. Food Agric. 8, 51 - 58. Motlo. H., I<. Dailies, D. Chilko, C. Motto (1970) Entiiron. Sci. Tidinol. 14, 231 -237. Page. A,, T. Ganje. M. Joshi (1971) H i / g d k i 41, 1. Pkilslicker, K., C. Markard ( I 975) Z. I , r h ~ m Unters. . ForscA. 158. 129 - 136. Priebe, A., H. Klein, H . Jaeger (1981) Mitt. forst/. Bimde.si~er.suclrsctnsr. W i m 137, 319. Kentschler. F. (1977) Proc,. 4th. In?. C'ongrc,.rs Clrtrti A i r , Tokyo, 99- 102. Kobcrts, T.. G. Goodm;rn ( I 974) in: Trow Suhsfuwces in E~i,oiron/nc.ii/ulHeultli, Missouri, University of Missouri 7. 105- 116. Sansoni, B. (1986) Fr. Z.Aw/. C l i m . 323, 573-600. Schuck, E., J. Lockc (1970) Environ. Sri. Teclrnol. 4, 324. Sheppard. M.. D. Tibault, S. Sheppard (1985) Wcrtcr, Air and Soii Poihrt. 26. 85-94. Smith, W., B. Staskawicz, R . liarkon (1978) Trrrns. Brit. M y c d . Soc. 70, 29-33. Steuhing. L., 3 . llaneke (1993) in: P h t s 0s Bion?onifor.s. B. Markert (cd.). VCH, Weinheim, 11. 155- 166. Suchodoller. A. (1967) Bcr. .Sc/nwiz. B o f . Ges. 77, 266. Tnha, M., L. Greini (1979) GKSS 79/E/50. Thornton, I . ( 1983) in : Applietl ~ } ? z ~ i r ~ ) F i ~ n Geocltc,misfry. r,/t/(~/ I. Thornton (ed.). Academic Press, p. 231. Tunney, H., G . Flemming, D. O'Sullivan. J. Molloy (1972) J . Agric. Rcs. I I , 85-92. Tyler, G., A.-M. Balsberg Pahlsson, G. Bcngtsson. E. Baath, L. Travnik (1984) W n i ~ rAir , ~rnrlSoil P o l l ~ t .17. 189-215. Wagner. G. (19x7) PlrD Tit University of Saarbriicken. Wagner, G. (1990) i n : E l m e n t ConccnfrufiowCcidustws in Euiqstcrn.s. t l . Lieth, 9. Markert (eds.). VCH, Weinheim, p. 41 -54. Walkenhorst, A., J . Hagcmeyer, S. Breckle (1993) in: P1nnt.v (7s B i o ~ ) n i t o r sH. . Markert (ed.). VCI-1, Wcinheim, p. 523 - 540. Wallace, A., J. Kinnear, 3 . Cha, E. Roinney (1980) J . Pkunt Nzrtrifion 2 , 1-9. Wilkins, C. (1978) Emiron. Polhit. IS, 23-30. Wissmath. P. (1982) U m w h I . 28-29. Wood, T., F. Borinanu (1975) Atnhio 4. 169-171. Wotton, D., D. Jones. S. Floyd Phillips (19x6) Writer, Air rind Soil Pollrii. 31, 349-358. Wyttcnhach. A,. L. Tobler (198X) T r w s 2, 52-64. Wyttenhxh, A.. S. Bajo, L. Tobler, T. Kellcr (1985) Plant ri~idSoil85,313-325. Yamada. Y., M . Bukovac, S . Wittwer (1964) Pfunr f%J'.S;O/. 39, 978-982. Zimnicrmann, K.-D. (1989) AFZ 11.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

22 Samp ,,ig of Terricolous Lichen and Moss Species for Trace Element Analysis, with Special Reference to Bioindication of Air Pollution Zoltan Tuba, Zsolt Csintulun, Zoltun Nugy, Kulnzun Szente, and Zoltun Tukacs

22.1 Introduction The significance of trace elements in plants can be looked at from different viewpoints : partly, because their role in plant physiological processes is being recognized and also because their contribution to environmental pollution is increasing. Amounts of several elements - which so far occurred as trace elements in plants - have increased drastically in the environment due to environmental pollution. As a consequence, their concentrations in vegetation have also changed notably, at least when considering certain groups of plants, e.g., lichens and mosses. Because of their specific ability to accumulate trace elements these species are particularly suitable to indicate environmental loads (Nash, 1989; Meenks and Tuba, 1992). Therefore, lichens and mosses are used for bioindication purposes very frequently, because of their several advantageous features in detecting and tracking pollution, first of all air pollution (Garty, 1993; Meenks et al., 1991). At the same time it is obvious that knowledge of all the factors engendering the real situation is a prerequisite for application of lichens and mosses for bioindication (e.g., via element accumulation) to be reliable. The number of factors to be taken into account is strongly dependent on the bioindication method employed. The two principal methods in cryptogam bioindication are:

(1) The “in situ” or passive method (using lichens or mosses growing in the survey area, e.g., Andersen et al., 1978; Farkas et al., 1985; Miikinen, 1987; Bruning and Kreeb, 1993), (2) The so-called indirect or transplantation method (lichen thalli or moss cushions are transplanted and exposed at the survey field for a certain period of time, e.g., Pilegaard, 1979; one new application of terricolous lichens and mosses, e.g., Tuba and Csintalan 1993a, 1993b). The opportunity of standardization of the latter transplantation method largely decreases the number of the above mentioned factors, the effect of which can hardly be interpreted. In the latter case, the remaining crucial points in cryptogam bioindication are “only” the test plants, the way of transplantation, the circumstances of exposition and sampling. Considering the first three of these points comprehensive information is available already for both lichens and mosses used for bioindication purpose by

416

Z . Tuhrr et uI.

measuring element accumulation (Meenks et al., 1991 ; Pilegaard, 1979; Tuba and Csintalan, 1993a, 1993b). However, several questions abount sampling of terricolous lichens and mosses are still not answered, yet a few new problems are emerging. This situation is not relieved by the fact that general principles of sampling for higher plants (Markert, 1993) are mostly valid for terricolous lichens and mosses, too. The principal aspect is that the sampled parts of the thalli and cushions must represent the entire thalli or cushions for any of the investigated parameters. This condition can be fulfilled by taking into account the distribution of the parameters within the thalli and cushions. Therefore, knowledge of the characteristic distribution of the investigated parameters (e.g., trace element content) in the thalli and cushions is necessary for elaborating an acceptable sampling procedure. However, information on trace element distributions in lichens and mosses are mainly available in relatron to the age and vertical position (Bargagli et al., 1987; Mikinen, 1987). Sampling problems - considering cryptogams - are different, in some aspects, from those of higher plants (Nimis et al., 1993). These emerge from distributional differences of the elements in cryptogainic plants which originate from their peculiar features, differing in many respects from those of higher plants. Here, we refer to further new aspects. The most important is that horizontal growth and spreading of the thalli and cushions largely exceeds vertical growth, resulting in their particular physiognomical characteristics, namely:

( I ) variabilities of trace element contents in parts of the thalli or cushions differ from those in the whole thlilli or cushions, (2) horizontal variabilities of the trace element contents within thalli or cushions, (3) variabilities of trace element contents between different thalli or cushions. Therefore, characteristic information on the distribution patterns mentioned above (in addition to the information on vertical distribution) are of importance when sampling for quantitative analysis not only of the trace elements but of any parameter. In the present study we investigated the vertical and horizontal distribution of 2 meso-elements (Al, Fe) and 10 trace elements (Cd, Co, Cr, Cu, Mn, Ni, Pb, Ti, V, Zn) in the thalli of two terricolous lichens, Cladonia conuolutu and Cladoonru furcatu, and in the cushions of the terricolous moss, Tortula ruralis. Variabilities of the trace element contents in parts of thalli or cushions compared to whole thalli or cushions, and variabilities bctween different thalli or cushions are discussed. Distributions of trace element contents in the soil substrate (transplanted together with the lichens and mosses) in comparison to the thalli and cushions were also investigated. In a survey directed exclusively to determine the trace element contents adsorbed on and accumulated within the plants, one of the principal points is to establish the method by which the mechanically unseparable trace element containing pollutants adsorbed at the surface can be cleaned off from the samples.

Sampling qf Terricolous Lichen und Moss Species

41 7

To elucidate this particular point trace element contents in the lichen and moss samples were assayed after surficial cleaning of the samples by washing for different periods.

22.2 Materials and Methods 22.2.1 The Species of Investigation Cludoniu furcatu (Huds.) Schrad. ssp. jurcatu Terricolous, desiccation-tolerant fruticose type lichen; contact between thalli and soil is rather loose; podetia of 15-20 mm mean length, hollow; diameter of the thalli groups between 5 and 7 centimeters. Cludoniu convoluta (Lam.) P. Cout. Terricolous, desiccation-tolerant foliose type lichen; thalli contact with soil is stronger; size of the thalli between 5 and 8 centimeters, of thalli groups between 30 and 40 centimeters in diameter. Tortulu ruralis (Hedw.) Geartn. et al. ssp. ruralis Terricolous, desiccation-tolerant ectohydric moss; upper parts of the shoots with more and lower parts with less leaves of 4- 12 mm mean length are clearly distinguishable; size of the cushions between 10 and 12 centimeters in diameter. The nomenclature follows Verseghy (1991) and Orban (1991).

22.2.2 The Original Habitat The species investigated were originally grown in a stand of perennial dry sandy grassland (Festucetum vaginatae danubiale) in Hungary, near Vricrritot (1 9" 14' E, 47'30 N , 130 m a d . ) . This area is exposed only to the general background pollution level in Hungary. Consequently, it has been used as a control site for several air pollution studies in Hungary (e.g., Meenks et al., 1991; Tuba and Csintalan, 1992, 1993b).

22.2.3 The Cryptogam Transplantation Technique Complete lichen thalli and moss cushions were planted together with a 6 cm section of their original soil substrate into wood exposure boxes (size: 1 5 x 2 5 cm), cautiously avoiding any mechanical damage. The bottom of the exposure boxes was perforated to ensure drainage. The exposure boxes containing the transplants were fixed to the the soil surface at each site.

418

Z.Tuha c t ul

22.2.4 Exposition Different locations with various air pollution levels where the investigated cryptogamic species were exposed are described in the table legends as well a s the other conditions of exposition. The original grassland was used as the control site.

22.2.5 Sampling Trace element content distributions in the upper (younger) and lower (elder) parts Upper, thinner and lower, thicker parts (10- 12 mm length) of podetia of C.,furcutii were separated for this purpose. In the case of C. conuolutu the outer (young) thallus parts were separated from the inner (elder) thallus parts. Shoots of T. rurulis were also divided into upper (average length: 4 - 5 mm) and lower (average length: 6-7 mm) parts. Four replicates were taken parallely from the three species.

Horizontal pattern of trace element contents within the lichen thalli and moss cushions Upper and lower parts of C.,furc.atu and T. ruriilis (see above) were used by taking four replicate samples horizontally along the longitudinal axes of the thalli group and the moss cushion.

Trace element content distributions in parts of thalli and moss cushions compared to the whole thalli and cushions, respectively Four replicate samples were taken horizontally from the thalli of C. convoluta and the cushions of T. ruralis along two perpendicular axes. The remaining 88% of the whole thalli (C. convolufa) and 95% of the whole cushions ( T . ruralis) (on dry weight basis) can be considered as identical with the whole thalli or cushions. These remaining parts were mechanically crushed, milled and mixed before taking four replicate samples from them.

Variability of the trace element contents of lichen thalli groups and moss cushions Samples were taken from 3 thalli groups and 3 moss cushions exposed together at the control site. Average contents of trace elements (n = 4) of three different thalli and cushions were compared.

Trace element contents in the soil substrates of the test plants Three replicates were taken from the soil substratcs of C. cnnuoliita thalli and T. r u d i s cushions in two ways: (1) i n full depth (6 cm) a t three different positions of the soil covered by the test plants (sampling unit diameter: 3 cm), (2) soil substrate was mechanically mixed prior to taking the three samples of the same volumes a s in (1).

Sampling of' Terricolous Lichen and Moss Species

419

The effect of cleaning by water Four replicates of dry-cleaned C . convoluta and T. rurulis samples were washed with deionized water on a polythene filter for 0, 5, 30 and 60 seconds, respectively.

22.2.6 Preparation of the Samples for Measuring Trace Element Contents The samples were cleaned from extraneous materials (e.g., soil fragments, dead plant parts, etc.) in their dry state. Soil pieces remaining on the test plants following the first cleaning procedure were removed mechanically. Soil samples were cleaned by sieving.

22.2.7 Element Analysis Samples of 200 mg dry weight were measured into teflon bombs and digested after addition of 2 mL concentrated HNO, and 2 mL concentrated H,O, under pressure at 130 "C for 45 minutes. The water-clear contents of the bombs were filtered through No. 42 Whatman filter paper and brought to a volume of 10 mL with distilled water. The solutions were analyzed by ICP-AES for 27 elements, out of which the changes in amounts of 2 meso-elements (Al, Fe) and 10 trace elements (Cd, Co, Cr, Cu, Mn, Ni, Pb, Ti, V, Zn) are discussed.

22.2.8 Data Analysis Coefficient of variation (given in percentage) was used to characterize and compare the variability of the trace element contents in the samples. Means were compared by Student's t-test (Campbell, 1989).

22.3 Results and Discussion 22.3.1 Vertical Distribution of Trace Elements within the Various Lichen and Moss Parts Trace element contents in the lower, elder podetium parts of C..furcatn were significantly higher than in the young, apical parts for all the elements assayed, except for Ni (Tab. 1).

upper part lower part

upper part lower part

Podetia

CV

2.7 8.5

CV

29.1 19.7

Al

436.4 *1096.5

Ni

2.609 3.601

8.9 4.2

CV

Pb

5.932 *15.065

11.8 10.9

CV

521.4 *1476.0

Fe

4.4 6.5

10.825 *30.712

2.456 4.959

v

cv

Ti

co

0.3416 *0.7297

cv 28.3 27.4

0.3082 0.3860

Cd

Metal content (pg . gt’) and coefficient of variation (4’0)

4.7 8.2

CV

17.5 12.8

CV

CV

22.3 2.5

39.342 43.912

10.8 11.6

cv

Zn

0.8122 *2.3911

Cr

cv 2.9 3.2

1961.5 1502.5

4.0 8.7

cv

K

28.728 *48.697

5.0 3.3 CV

Mn

CV

122.614 8.4 *203.437 *5.1

Total

31.263 *52.985

cu

Contents of Meso- and Trace Elements, Total Amounts of Trace Elements, K Contents, and Coefficients of Variation (CV; n = 4) in the Upper and Lower Parts of the Podetia of Cludoniu,firrcatu Statistically significant differences (P < 0.05) are indicated by *. Transplanted thalli of C.&rrcutu were exposed for 10 weeks (from May 15, 1991 to August I . 1991) in the surrounding of a power station and an oil refinery; direction: E. distance: 2.1 km from the power station; the oil refinery station is situated 5.2 km south from the power station (Hungary, Szazhalombatta. 27 km south of Budapest, 18‘55’E. 47’20”. 118 m a.s.1.)

Tab. 1.

a

?-

N h

5-

2

h

$ 3

San?pling of Terricolous Lichen und Moss Species

42 1

In C. convoluta four elements (Al, Fe, Zn, K) were found in higher amounts in the outer, younger than in the inner, elder parts of the thalli but the differences were not statistically significant (Tab. 2). Considering the other elements, the inner thalli parts contained statistically higher amounts than the outer parts regarding 4 elements (Cd, Cr, Cu, Pb) while the differences were not significant for the remaining elements. Trace element contents in the lower, elder parts of T. ruralis shoots were significantly higher than those in the younger upper parts for 6 elements (Co, Cr, Fe, Mn, Pb, Ti) (Tab. 3 ) . The Cd content was higher in the upper than in the lower parts, but this difference was not statistically significant. Total trace element contents (sum of all the assayed trace elements) were also significantly higher in the lower than in the upper parts of C. furcata and T. ruralis (see Tabs. 1 and 3). Therefore, it seems necessary to separate the upper and lower parts when sampling for trace element assays. There are enough literature data referring to the considerable differences between th trace element contents of the different - vertically separated - parts of the lichen thalli and moss shoots. Heavy metals, for example, also accumulate in the lower, basal parts of lichen podetia in higher amounts than in the young apical segments (Pakarinen et al., 1978; Brown, 1987). Heavy metal accumulation is more expressed in the inner, elder than in the marginal parts of foliose lichens (Bargagli eta]., 1987; Shutte, 1977). Several heavy metals have been shown to accumulate in higher amounts in the lower, elder segments of other mosses, too (Makinen, 1987; Pakarinen and Tolonen, 1976). Distributions of total and single element contents between the different ion exchanging and storage compartments (cell wall, intracellular) is not known - and this question was not addressed by the present study. However, the significantly lower K' contents in the elder parts - which, on the other hand, in both species contained significantly higher amounts of trace elements than the younger parts (see Tabs. 1 and 2) - suggest that considerable fraction of the assayed elements - and among these, mainly the heavy metals are localized intracellularly. This can be accounted for the decrease in K' content, related to K + efflux induced by intracellular uptake of heavy metal ions (Puckett, 1976). This phenomenon has been observed several times in lichens and mosses (Goyal and Seaward, 1982; Burton et al., 1981; Brown and Beckett, 1985). Hence, it can be assumed that a higher portion of heavy metals is accumulated intracellularly in the elder than in the younger parts. In the course of the present survey whole thalli of C.,furcata and cushions of T. ruralis have been exposed to air pollution for the same period. In spite of this, different rates of accumulation in the young, upper and elder, lower parts might be a consequence of differences in their exposition periods, since the upper parts have a continuous but slow growth rate in both cryptogamic species. As a consequence, the lifetime of the apical segments of the upper parts was shorter than the duration of the exposition period. However, in spite of the very low growth rate of lichens and mosses there must be other processes which contribute to the different accumulation rates of upper and lower parts. The higher accumulation rate in the lower, elder parts may also be related to the downward ion transport from the upper parts.

Fe

197.1 237.7

Pb

7.435 *5.462

CV

9.83 14.73

CV

17.35 9.08

Al

185.1 204.4

Ni

1.102 1.080

2.998 3.652

10.59 18.32

1.136 0.879

v

cv

Ti

CV

9.43 17.12

0.0296 0.0182

18.10 8.45

0.3267 *0.1594

14.56 11.15

co

Cd

CV

cv

Metal content (pg . g-') and coefficient of variation (%)

31.410 24.220

8.27 11.45

14.95 11.11

CV

20.72 17.28

0.3659 *0.1538 Zn

cv

Cr

CV

42.22 39.78

CV

48.751 61.526

Total

7.298 *5.381

cu

8.42 9.97

CV

15.87 8.50

CV

cv

10.35 16.45

CV

2222.0 13.57 1918.0 10.05

K

9.43 7.76

Mn

upper part lower part

upper part lower part

Cushions

~~ ~~

Pb

31.795 *36.007

CV

16.8 6.2

Ni

5.137 5.640

17.4 12.5

CV

14.1 11.7

1333.7 *1695.0

15.1 12.2

1073.4 1329.7

CV

Fe

CV

A1

6.856 *9.433

Ti

0.6572 0.6073

Cd

7.574 8.979

17.2 11.9

CV

v cv 9.5 12.2

4.4 13.1

CV

0.7652 *1.0169

co

19.4 15.9

cv

Metal content (pg . g-') and coefficient of variation (%)

~

CV

16.3 13.3

60.490 61.272

21.1 18.5

cv

Zn

1.7696 *2.6507

Cr

199.682 222.016

Total

8.237 9.839

cu

15.5 9.1

CV

15.2 27.7

CV

2494.2 *1769.2

K

76.402 *86.573

Mn

16.4 15.6

cv

15.3 12.7

cv

Tab. 3. Contents of Meso- and Trace Elements, Total Amounts of Trace Elements, K Contents, and Coeffcients of Variation (CV; n = 4) in the Upper and Lower Parts of Shoots of Tortula ruralis Statistically significant differences (P < 0.05) are indicated by *. Transplanted cushions of T. ruralis were exposed for 10 weeks (from May 15, 1991 to August I , 1991) near (5 m) a motor road with high traffic loads (motor road No. 6. Hungary, Szizhalombatta, 27 km south of Budapest)

outer part inner part

outer part inner part

Thalli

Tab. 2. Contents of Meso- and Trace Elements, Total Amounts of Trace Elements, K Contents. and Coefficients of Variation (CV; n = 4) in the Upper and Lower Parts of the Thalli of Cladonia conoolutu Statistically significant differences (P < 0.05) are indicated by *. Transplanted thalli of C. conoolutu were exposed for 10 weeks (from May 15. 1991 to August I , 1991) at the control site (Hungary, Vacratot, 28 km north-east of Budapest)

n

-

5.

2

?J

gN

Sumpling of' Terricolous Lichen and Moss Species

423

Differences in photosynthetic and, consequently, metabolic activities between the upper and lower parts (Tuba, 1985; Tuba et al., 1993c) may also influence the rate of element accumulation and may partially contribute to the different trace element (heavy metal) contents in the lower and upper parts. Differences in thickness of the tissues may be another factor, for tissues in the lower parts are thicker than those in the upper parts. Moreover, the possibility of a higher rate of trace element accumulation in the lower parts due to an uptake from soil is not supported by literature data. In addition, a prerequisite for this to occur is a strong contact with the soil, which can be excluded in the present study, since we found the highest contents of trace elements in the basal parts of C.,furcata, the species characterized by the loosest soil contact. Furthermore, trace element contents were not significantly higher for all the investigated elements in the lower than in the upper parts of the rhizoidal species T. rurulis, associated more tightly to the soil. Another question is whether the trace elements deposited on the cryptogam plants were washed off and out from the upper and lower parts by rains, especially acid rains (Boonpragob and Nash, 1990). Therefore, consideration of the vertical element distribution pattern is of outstanding importance when sampling for trace elements.

22.3.2 Horizontal Distribution of Trace Elements within the Thalli of C.fuvcata and the Cushions of T. ruralis Horizontal distribution of the trace elements in the thalli and cushions sampled along a horizontal transect (see Sect. 22.2) was characterized by using the coefficient of variation (CV) considering the upper and lower parts separately. The variabilities of the trace element contents in the samples taken horizontally from the thalli and cushions can be considered as small in both the upper and the lower parts. At the same time there are differences in the variability of the element content distributions between the two species and between the lower and upper parts. Coefficients of variation for the contents of the assayed elements ranged between 2.5 and 29.1% for the thalli groups of C.,furcura (see Tab. 1). Differences in horizontal distribution of the element contents in the upper and lower podetia parts are the same for all but two elements (Zn, Ni), for which horizontal variability is higher in the upper than in the lower podetia parts. Within the thalli groups the highest horizontal variability was shown for Cd (27 - 28%). Horizontal variability of element content was rather low - below 9% - for another 5 elements (Cu, Ti, Al, Mn, Pb), while it was between 10-22% for the remaining 3 elements (Cr, Fe, Co). Horizontal variabilities of the trace element contents in T. ruralis were intermediate for most of the elements with CV values ranging between 10-20% (Tab. 3). Values were higher in the case of Cu for the lower, elder parts of the moss shoots, and rather low for Ni (6.2%) and for Co (4.4%) in the elder and younger moss parts, respectively.

424

Z . Tuho e t ul.

The contents of 3 elements (Cd, Co, Ni) had higher and all the others had lower horizontal variabilities in the lichen thalli compared to the moss cushions. This difference was largest in the case of Cu, while CV values were about equal for Fe. Further general trends considering the variabilities in the trace element contents in the lichen and the moss were not found. So, we can not confirm, that variabilities - as characterized by the coefficient of variation - for Fe would be the highest and for Zn the lowest in the lichen thalli (Garty et al., 1979; Garty and Ammann, 1987). These findings were interpreted as Fe was deposited in the form of large-size particles, while Zn occurred in a finer, smaller and better dispersed form. It can be assumed that these elements might be associated with particles of various sizes, since the association of heavy metals to the particles may depend on the type of the pollution source and the composition of the pollutant. Summarizing, horizontal variabilities of the contents of the assayed trace elements were more extreme in the thalli of C.,furcatu, while they were nearly the same for most of the elements in the cushions of T. ruralis. Variabilities of trace element contents within the lichen thalli groups and moss cushions did not exceed 30%) when sampling horizontally and were considerably lower in a lot of cases. Therefore, three to four samples proved to be able to represent satisfactorily the horizontal distribution of the trace element contents in the thalli and cushions.

22.3.3 Distribution of Trace Element Contents in Parts of the Lichen Thalli and Moss Cushions Compared to Whole Thalli and Cushions, Respectively After sub-sampling (see Sect. 22.2), the remaining larger sample parts (88 - 95%) of the mechanically crushed and mixed thalli and cushions (representing the whole) were used to relate the element contents in the whole thalli and cushions to the respective parts. We considered this 88 - 95% fraction as a part large enough to represent the whole satisfactorily. The differences were not significant for any of the elements in the case of C. convoluta (Tab. 4), while in T. ruralis contents of some trace elements (Al, Cr, Ni, Pb) in the sampled parts differed significantly from those in the whole cushion (Tab. 5). Namely, significant differences were found in T. ruralis between thc lower parts and the whole in two cases (Cr, Pb) and between the upper parts and the whole in three cases (Al, Ni, Pb). Variabilities of the contents of trace elements in the cushion parts were considerably lower for Cu and Cd and similarly higher for Co and Zn as compared to the variabilities for these elements in the whole cushion. In the other cases the variabilities of the trace element contents were approximately the same (Cr, Fe, Ni, Ti) or differed only slightly (Al, Mn, Pb) in the cushion parts as cornpared to the whole cushion.

9.1 19.7

CV

Mn

21.670 21.110

5.7 18.7

CV

573.1 530.7

Al

2.742 5.478

Ni

705.3 633.6

Fe

42.4 74.8

CV

~

2.8 25.0

CV

20.753 19.175

Pb

0.3721 0.3379

Cd

11.8 11.3

CV

14.3 18.0

cv

Metal content (pg g - ') and coefficient of variation (Yo)

10.629 9.387

Ti

0.2911 0.2904

co

5.2 20.1

cv

20.7 20.3

CV

2.543 2.090

v

0.6974 0.8283

Cr

19.9 18.6

CV

27.0 23.7

cv

39.027 41.025

Zn

3.654 3.654

cu

10.2 16.3

CV

14.3 20.3

CV

1550.5 1692.7

K

102.378 103.375

Total

8.6 10.8

cv

8.3 18.9

CV

upper parts lower parts whole cushions

upper parts lower parts whole cushions

Cushions

CV

15.1 12.2

9.2

CV

15.3 12.7

8.4

Al

1073.4 1329.7

1009.9

Mn

76.402 86.573

86.970

4.838

5.137 5.640

Ni

1390.5

1333.7 1695.0

Fe

9.1

16.8 6.2

CV

13.8

14.1 11.7

CV

48.125

31.795 36.007

Pb

0.6205

0.6572 0.6073

Cd

6.4

17.4 12.5

CV

7.6

19.4 15.9

cv

Metal content (pg g-I) and coefficient of variation (%)

8.005

6.856 9.433

Ti

0.9211

0.7652 1.1069

co

11.1

11.900

7.574 8.979

v

cv 9.5 12.2

2.6024

1.7696 2.6507

Cr

25.5

4.4 13.1

CV

7.9

17.2 11.9

CV

19.3

21.1 18.5

cv

77.925

60.490 61.272

Zn

8.926

8.237 9.839

cu

25.6

16.3 13.3

CV

4.7

15.2 27.7

CV

1953.2

2494.2 1769.2

K

250.831

199.682 222.016

Total

1.9

16.4 15.6

cv

7.2

15.5 9.1

CV

Tab. 5. Contents of Meso- and Trace Elements, Total Amounts of Trace Elements, K Contents, and Coefficients of Variation (CV; n = 4) in Cushion Parts (upper and lower) and Whole Cushions of Tortulu ruralis Transplanted cushions of T. ruralis were exposed for I0 weeks (from May 15, 1991 to August I , 1991) near ( 5 m) a motor road with high traffic loads (motor road No. 6, Hungary, Szazhalombatta, 27 km south of Budapest)

parts of thalli whole thalli

parts of thalli whole thalli

Thalli

Tab. 4. Contents of Meso- and Trace Elements. Total Amounts of Trace Elements, K Contents. and Coefficients of Variation (CV; n = 4) in Whole Thalli and Different Parts of Thalli of Cludonza convoluta Transplanted thalli of C. conuolutu were exposed for 10 weeks (from May 15. 1991 to August 1, 1991) near ( 5 m) a motor road with high traffic loads (motor road No. 6, Hungary, Szazhalombatta, 27 km south of Budapest)

0:

2

426

Z.Tuhu ef a1

22.3.4 Variability of the Trace Element Contents in the Lichen Thalli and Moss Cushions The variabilities in lichen thallus or moss cushion parts of nearly the same age cannot deviate considerably from those for the miiled mixture representing the whole thallus group or cushion. When comparing the thallus parts and the whole thalli of C. c'onoolura, the variabilities of the trace element contents were very similar for all elements, except for Co, Cr, Cu and Zn, for which they were significantly lower in the thallus parts than in the whole thalli (Tab. 6). In T. ruralis, variabilities of trace element contents in the moss cushion parts were higher than in the whole cushion for Ti and considerably lower for Fe, Cd, Co, Cu, Ni and Zn (Tab. 7). Variabilities of the trace element contents for the latter elements were lower between different moss cushions than within one cushion. The variabilities for the other elements assayed were nearly the same comparing the values within and between cushions. Interestingly, the variability of trace element contents - as characterized by the coefficient of variation - was higher for the whole thalli of C. conooluta than for the sampled parts of the thalli. This might be due to a slightly uneven mixing of the upper and lower parts during crushing - compared to the parts, where the typical vertical distribution was not disturbed.

22.3.5 Trace Element Contents in the Soil Substrate of the Lichen Thalli and Moss Cushions Information on the trace element contents of the soil substrate of the test plants (the lichen thalli and the moss cushions, respectively) may become an important question in ccrtain cases, e.g., when investigating the relationship between the element contents in soil and cryptogam plants. Moreover, this information is also necessary when using terricolous lichens and mosses for transplantational bioindicatlon purposes. Namely among the cryptogamic species those are more suitable for bioindication of air pollution via element accumulation, which adsorb and/or accumulate the higher portion of the trace elements on their surface and/or in their tissues, while at the same time very little of these elements is left to pass through into the soil beneath. These requirements are fulfilled by those terricolous lichens and mosses, which - like C. c'onuoluta and T. ruralis - provide full shielding of the soil beneath, due to their dense cover. In the present study, the minor changes in trace element contents of the soil substrates did not follow the accumulation pattern of the lichen and moss indicators in any of the cases, except for V in the lichen and V and Ni in the moss (Tab. 8;

Ni

cv

14.54 4.77 8.63

4.60

Mn

12.37 13.56 12.98

12.97

1.041

1.043 1.147 0.932

462.8

10.53

380.4

403.0 489.3 496.0

Fe

7.47 3.07 5.41

cv

334.2 400.7 406.2

Al

10.29

19.20 9.77 13.60

cv

11.21

8.48 4.14 6.21

cv

11.966

13.285 11.458 11.155

Pb

0.2219

0.2275 0.1983 0.2401

Cd

9.63

12.94 2.18 5.89

cv

9.67

11.63 7.59 14.30

cv

Metal content (pg . g-') and coefficient of variation (Yo)

7.034

6.295 7.225 7.582

Ti

0.0908

0.1062 0.0924 0.0739

co

9.44

1.439

1.386 1.459 1.473

V

cv 9.64 5.89 5.29

0.7166

0.7154 0.7304 0.7039

Cr

17.84

64.81 73.17 56.56

cv

3.23

2.67 12.55 2.45

cv

1.86

12.34 22.15 18.28

cv

26.732

27.186 25.618 27.393

Zn

3.528

3.480 3.567 3.538

cu

3.63

10.65 5.43 23.01

cv

1.26

15.44 13.90 14.71

cv

means for three cushions

3

1 2

1 2 3 means for three cushions

Cushions

32.17

12.03

9.27 13.46 17.40 2.333

2.381 2.374 2.243

Ni

cv

Mn

30.25 29.64 36.62

1334.4

14.19

569.4

1337.8 1326.7 1338.9

Fe

10.54 17.26 17.04

cv

547.8 501.6 658.8

A1

3.34

5.51 22.41 12.29

cv

0.51

22.66 15.28 18.74

cv

12.187

12.510 10.779 13.272

Pb

1.9579

1.9435 1.9080 2.0221

Cd

10.48

9.053

8.131 8.043 10.985

Ti

cv 9.64 12.14 15.25

0.4695

0.4510 0.4473 0.5101

co

2.98

20.77 9.85 17.70

cv

Metal content (pg . g- ') and coefficient of variation (YO)

18.49

6.92 11.59 10.79

cv

7.51

16.40 17.57 13.88

cv

1.376

1.280 1.186 1.662

V

4.9484

4.2273 6.1083 4.5098

Cr

18.30

48.632

47.795 52.923 45.178

Zn

cv 4.82 19.27 17.92

9.356

10.360 9.097 8.612

cu

20.50

18.29 26.61 16.02

cv

8.10

20.01 16.51 11.19

cv

9.65

17.83 14.93 15.99

cv

Tab. 7. Contents of Meso- and Trace Elements and Coefficients of Variation (CV; n = 4) in Three Different Cushions (1, 2, 3) of Tortula ruralis Transplanted cushions of T . ruralis were exposed for 10 weeks (from May 15, 1991 to August I . 1991) at the control site (Hungary, Vacritot, 28 km north-east of Budapest)

1 2 3 means for three thalli

2 3 means for three thalli

1

Thallus

Tab. 6. Contents of Meso- and Trace Elements and Coefficients ofvariation (CV; n = 4) in Three Different Thalli ( I . 2.3) of c'ludoniu concolutu Transplanted thalli of C. concoluru were exposed for 10 weeks (from May 15, 1991 to August I . 1991) at the control site (Hungary. Vacritot, 28 km north-east of Budapest, 19 14'E. 47-30", 130 m a.s.1.)

P N 4

&

c

h

e.

-.

3 4 5 6

MOL

Dkv

Soil Control

3 4 5 6

MOL

Control Dkv

-.

NE, 35 km S. 5 km Okm E, 2km SW, 3 k m E. 9 km N. 4km

-.

NE. 35 km S, 5 km 0 km E, 2 km SW, 3 km E, 9 km N. 4 km

1134.0 1338.3 1239.3 1250.6 1279.3 1188.2 1030.2

1465.7 1517.7 4900.7 4346,5 2049.7 1712.3 1538.3

1.50 21.44 10.93 9.55 2.45 20.93 28.37

16.88 6.75 23.60 22.56 12.50 26.58 0.38

2076.6 2387.0 2263.6 2240.3 2265.3 2154.0 1927.6

5.24 22.03 10.46 10.64 3.79 20.42 25.79

2331.0 9.28 2307.6 8.09 5905.0 9.36 6153.5 21.34 2968.6 11.44 2768.3 22.90 2347.3 6.03

NE. 35 km 6505 2232 842.9 25.51 S, 5 km 7096 1264 831.7 6.79 Okm 1466 5 37 66 1851.5 36.01 E, 2 k m 1361 4 3548 1739.0 40.90 SW, 3 km 1025 7 43 68 986.8 12.36 E, 9 km 841 3 2465 1179.2 37.66 N 4 km 727 8 5 57 752.0 17.06

Tortula ruralis

3 4 5 6

Dkv MOL

Control

Cladonru convoluta 7.32 16.78 18.84 32.45 22.04 12.62 13.85

0.6406 0.9700 4.7235 6.0980 1.2316 2.4523 0.1831

Ni

CV Pb

28.97 4.575 68.92 20.683 29.09 9.948 37.38 38.240 31.05 11.986 19.00 21.047 0.00 7.300 54.50 19.273 51.99 7.022 60.09 20.307 16.10 6.552 81-15 23.120 13.00 3.815 44.36 18.607

CV 13.04 7.17 19.75 14.16 17.47 23.55 9.00

CV

1.7242 3.9533 2.2678 4.1107 2.3900 2.1545 1.8673

9.53 88.91 18.74 35.61 17.51 19.52 18.50

3.492 4.304 3.912 4.186 3.515 3.290 3.181

15.49 26.85 5.12 28.67 3.59 14.35 31.34

3.717 5.164 3.156 3.661 3.812 3.821 3.090

40.55 6.13 10.45 4.29 9.82 14.30 5.72

51.56 7.2071 9.64 7.412 17.61 22.213 9.85 34.00 10.7714 16.11 8.724 40.13 124.566 10.63 12.53 15.3033 22.48 17.343 13.57 28.383 12.53 28.46 26.9002 0.00 17.380 25.03 28.560 32.16 67.27 11.1184 16.37 7.805 25.72 19.883 44.69 33.22 12.5133 7.30 7.075 16.82 42.700 19.61 86.67 7.8950 6.69 7.138 43.97 19.966 14.43

~

Cu

(Oh)

0.0000 4.7252 0,0000 5.8043 0.5501 112.25 6.4100 0.6748 118.81 7.8210 0.2281 173.21 7.0790 0.0860 93.02 4.8263 0.0000 4.7183

0.2913 11.75 1.2042 3.46 0.3137 25.44 1.6577 44.56 0.2933 9.56 1.4100 17.02 0.2744 8.41 1.2808 15.38 0.3000 7.73 1.4751 5.17 0.2642 7.85 1.3132 33.57 0.2614 17.23 1.0155 49.03

3.61 15.09 17.89 20.17 21.54 25.04 0.6411 16.81

0.6960 0.5091 0.4293 0.4942 0.4193 0.5990

0.4834 0.4172 0.5934 0.5146 0.4464 0.5199 0.?799

Location Direction Metal content (pg. g-') and coefficient of variation and distance Al C V Fe CV Cd CV Cr CV

2.223 2.594 3.386 2.566 2.648 2.467 2.335

6.271 8.393 29.920 21.433 10.039 7.557 8.520

1.683 2.216 6.840 4.887 2.853 2.938 2.213

V

0.36 16.38 4.94 10.49 6.19 14.47 17.12

9.99 6.73 4.16 9.66 26.44 22.25 11.13

18.75 12.71 6.08 27.88 28.43 21.32 26.35

CV

9.323 10.233 11.100 11.571 10.097 9.225 8.083

52.760 69.273 56.296 71.980 52.676 80.055 57.730

42.455 42.637 44.683 53.033 43.507 48.880 39.353

Zn

19.07 19.00 22.19 22.76 16.44 5.53 17.42

7.88 32.42 9.83 24.17 25.05 32.88 17.51

28.29 8.18 20.74 14.99 6.88 29.45 7.50

CV

Transplanted thalli and moss cushions were exposed for 10 weeks (from May 15. 1991 to August 1. 1991) in the surroundings of a power station and an oil refinery; directions and distances of locations from the power station are indicated (Hungary, Szizhalombatta. 27 km south of Budapest)

Tab. 8. Contents of Trace Elements and Coefficients of Variation (CV; n = 3) in the Thalli of Cirrdonirr c o n ~ o h r o .in the Moss Cushions of Tortuku rurulis and in the Soil Substrates Exposed a t Locations Affected by Different Degrees of Environmental Pollution

5-

D

3

2

?.

Sumpling

of' Ti?rricolous Lichan and Moss Spccii?.y

429

trace element contents in the samples taken from a few (7) typical exposition sites; correlation coefficient calculated by using the data from all (14) exposition sites.). These two elements were emitted into the air in considerable amounts from local sources in the study area. Correlation coefficients calculated between the element contents in the test plants and in their soil substrates were 0.86 for Ni in the case of the moss and 0.89 and 0.96 for V in the case of C. convolutu and T. rurulis, respectively. Considering the other elements this relation was characterized by correlation coefficients between 0.18 and - 0.45. In other words, there was no significant difference (except for Ni and V) in the trace element contents of the soil substrates of the lichen thalli and the moss cushions between the most and the least polluted (control) localities, while the degree of pollution was indicated by the plants. Similar results were obtained for 7 lichen species by Seaward et al. (1978). It is suggested, that most of the trace elements deposited from the air were adsorbed and accumulated by the lichen and moss transplants. Ni and V are the two elements occurring in the highest amounts in this area. Partly due to this, and due to the fast leaching rate of Ni from both species (see below) these two elements were present in the soil in higher amounts. At the same time, the probability of leaching is largely reduced by the dry summer weather. Another evidence is that V which is characterized by a slower leaching rate than Ni, also accumulated in the soil to a certain degree. This result was not affected by the method of soil substrate samling, there were no significant differences between the trace element contents with the soil substrate sampled before or after homogenization (see Sect. 22.2). Measuring the trace element contents of the soil substrate of the lichen thalli and the moss cushions seems to be a suitable method for qualifying the accumulation capability of these bioindicators when used for detecting and measuring air pollution.

22.3.6 The Influence of Sample Washing on Trace Element Contents In bioindication studies, air pollution can be tracked by two ways depending on the goals of the study: (1) one can be interested in the total trace element content (sum of the amounts adsorbed on the surface and accumulated within the plant), while (2) the accumulated amount within the plants is addressed only. In (2) the samples should be free of pollutants (including critical elements, too) adhered to the outer surface. This question was investigated by washing the samples with deionized water for different periods. In cryptogam species, washing the samples prior to the element assays has two counteractive effects. O n one hand the element content (on dry weight basis) will increase with washing due to the removal the of the pollutants adhering to the surface and, on the other hand, it will decrease depending on features the of the test plants, the element in concern and the duration of washing due to leaching of the accumulated elements.

0 5 30 60

Mn 21 1.050 224.850 230.975 243.875

A1 3875.5 4140.7 4355.5 4802.5

0.00 3.38 4.55 25.89

6.2 11.2 3.8

5.1

cv Ni 10.670 15.487 12.333 10.569

5077.0 5400.2 5631.5 6044.5

Fe

5.6 8.5 14.8 5.7

cv

21.0 52.5 4.6 8.6

cv

4.6 9.4 15.5 2.4

cv

22.9 22.1 6.3 15.9

cv

Pb 20.597 23.405 2 1.827 23.3 15

Cd 0.6920 0.7441 0.7976 0.8178

Pb 1 1.601 14.538 14.948 15.090

6.2 12.2 4.6 5.2

cv

6.2 6.1 9.3 2.1

cv

19.4 3.5 6.2 3.3

cv

7.9 6.0 12.8 31.2

cv

4.7 8.2 14.9 1.8 Ti 41.630 44.765 47.960 59.725

cv co

19.6 7.0 1.6 2.5

cv

21.7 28.1 22.0 20.9

cv

2.8389 2.9660 3.0056 3.2498

Ti 12.851 16.655 18.740 17.637

0 5 30 60

Ni 1.512 2.090 1.835 1 .SO3

18.5 4.7 8.1 2.6

cv

27.1 15.3 12.4 14.1

Mn 20.463 24.713 27.465 24.665

0.3034 0.3605 0.3538 0.3608

0 5 30 60

20.8 6.7 10.0 2.4

0.3441 0.4915 0.3605 0.3316

588.7 765.4 832.0 749.0

506.8 632.2 721.8 666.4

Cladoniu convoluta 0 0.00 5 0.19 30 2.30 60 4.65

18.9 4.1 6.1 2.6

co

Duration of d. m. (%) Metal content (pg . g-') and coefficient of variation (YO) washing (s) Al cv Fe cv Cd cv

V 14.850 17.033 16.090 16.897

Cr 6.2192 6.5033 6.8972 7.5541

V 2.437 2.721 2.754 2.658

1.0855 1.4675 1.5927 1.2860

Cr

6.3 4.8 5.6 5.3

cv

4.6 8.9 14.8 5.1

cv

13.4 4. I 5.4 2.3

cv

8.1

15.1

24.9 8.6

cv

Zn 54.165 57.525 65.585 65.152

cu 8.329 8.790 10.117 12.290

Zn 30.350 36.915 37.638 36.025

3.882 3.868 10.21 1 5.987

cu

8.8 0.3 13.9 9.0

cv

3.4 3.6 23.5 11.3

cv

18.1 11.3 8.2 5.7

cv

44.9 6.4 7.6 14.8

cv

Tab. 9. Changes in Dry Mass (d.m.) and in Contents of Trace Elements. and Coefficients of Variation (CV; n = 3) in the Thalli of Cladoniu conoolztra and in the Moss Cushions of Torrulu ruralis after Cleaning with Deionized Water for 0, 5, 30 and 60 s Transplanted thalli of C. cot~cohtuwere exposed for 10 weeks (from November 4, 1991 to January 14, 1992) in the surroundings of a power station: direction: NE: distance: 1.9 km (Hungary. Ajka. 121 km east of Budapest, 17'31'E. 47 10". 239 m a.s.1.) Transplanted cushions of T. ruralis were exposed for 10 weeks (from May 15, 1991 to August I , 1991) near (5 m) a motor road with high traffic loads (motor road No. 6. Hungary, Szazhalombatta, 27 km south of Budapest)

?3

.-v t

Y5.

iU

Sampling of Terricolous Lichen und Moss Species

43 1

Washing the samples for 0, 5, 30, and 60 seconds has resulted in a continuous, linear decrease in weight of the thalli of C. convoluta, while the kinetics of this decrease was different for the cushions of T. ruralis (Tab. 9). In C . concoluta contents of Cr, Cu, Fe, Ti, AI, Zn and Mn increased in treatments not longer than 30 seconds, while treatments of 60 seconds resulted in decreases of various rates for the different elements (Tab. 9). Leaching of Co and Ni began during the 30seconds treatment, no changes were found for Cd and Pb after 5 seconds, i.e., the element contents neither increased nor decreased in the treatments successive to 5 seconds treatments. Trace element contents increased gradually as a function of the duration of washing, except for Ni, Zn and Cd in T. ruralis (Tab. 9). The Ni-content increased until 5 seconds of washing and decreased at successively higher rates thereafter. The rate of Cd increase was considerably repressed between 30 and 60 seconds and that of Zn between 15 and 30 seconds, suggesting that a 60 seconds treatment leaching was caused for part of the element content - smaller for Cd and higher for Zn. These changes were presumably caused in part by changes in the dry weight of the samples due to washing (see above). Literature data on the cleaning of lichen and moss samples by washing before determining the contents of trace elements are at least not consistent and sometimes even contradictory (Nieboer et al., 1972; Garty et al., 1977; Pilegaard et al., 1979; Nimis et al., 1993).We conclude, that this problem should be investigated separately for every object, even if there are conclusions from this kind of studies to be generalized. Conclusions in this respect from the present study are summarized below: The Ni-content (on dry weight basis) decreased markedly in both species, even after a few seconds of washing. The situation is similar for Cd, although the decrease is not as sharp as in the case of Ni. Leaching was found in the lichen for Co and Pb, and for Zn in the moss. For the other elements assayed, parallel decreases in their contents were found until 30 seconds of washing in the case of the moss and until 60 seconds in the case of the lichen. Therefore, for these plants it is not recommended to wash the samples with water for periods longer than 5 seconds, prior to the element assays when the analysis is carried out for measuring the contents of the elements above.

22.4 Summary Sampling to determine the element contents in lichens and mosses requires consideration of their peculiar features differing in many respects from those of higher plants and affecting the distribution of the elements in addition to the general sampling principles. So, the horizontal distribution patterns of the trace element contents within the thalli and cushions should also be considered as well as the information on vertical distributions.

432

2 Tuba et al.

In the present study we investigated these peculiar vertical and horizontal distributions of trace element contents in the thalli of two terricolous lichens, Cludoniu convoluta and Cladonia furcutu, and in the cushions of the terricolous moss, Tortulu ruralis. Further, the element content modifying effect of surface cleaning by washing in relation to the special element accumulation behavior of these species and the changes in trace element contents of the soil beneath the test plants were investigated for 2 meso-elements (Al, Fe) and 10 trace elements (Cd, Co, Cr, Cu, Mn, Ni, Pb, Ti, V, Zn). Considering the vertical distributions of the trace element contents and also the total trace element contents, the lower, elder parts showed higher values than the younger, upper parts in all of the species, with larger differences in C. furcata than C . ronuoluta and in T. ruralis. Horizontal variability of the contents of trace elements assayed was more extreme in the thalli of C.,furcutu, while it was nearly the same for most of the elements in the cushions of T. ruralis. Comparing the trace element contents and their distribution in the parts of the lichen thalli group in relation to the whole thalli the differences were not significant for the element contents while their distributions were more heterogeneous for the whole thalli than for the parts of them in the case of C. convolutu. In T. ruralis significant differences were found in both the trace element contents and their distribution between the sampled cushion parts and the whole cushions. Variability of trace element content distributions were similar within and between the thalli and cushions of nearly the same age. In the course of a 10 weeks transplantation bioindication case study conducted for the determination of the heavy metal loading in the surroundings of a Hungarian industrial town, it was found that changes in trace element contents of the so11 substrate covered fully by the test plants were consistent only for two elements (Ni, V) from local sources with those in the exposed lichens and mosses. In the case of Ni, this relationship was unambiguous only for the moss. When washing the samples with deionized water trace element contents (on dry weight basis) first increased, due to the removal of the surface-adhering pollutants, and decreased thereafter, because the accumulated elements were leached out. This leaching occurred at shorter washing periods from C. convoluta than from T. ruralis. Ni contents decreased considerably and Cd contents to a lesser extent after a few seconds washing period in both species. Further elements which were found to leach out were Co and Pb in the case of the lichen and Zn in the casc of the moss. Therefore, for these plants it is not recommended to wash the samples with water for periods longer than 5 seconds, prior to the element assays for measuring the contents of the elements above. Results confirmed, that consideration of the variability of trace element contents within and between the thalli and cushions is essential either during transplantation or when sampling for trace element content determination. Closing note. Because of the extreme difficulties encountered by the topic, we intend to use the present work as a pilot study for future research.

Sumpling of Terricolous Lichen and Moss Species

433

Acknowledgements. The work was funded by the Hungarian National Scientific Foundation (OTKA F/1. 5359). Present work was also supported by the Duna Petroleum Refining Compary (MOL, Szazhalombatta), the Duna Heat Power Plant Compary (Szazhalombatta) and the Heat Power Plant Station (Ajka).

22.5 References Andersen, A,, Hovmand, M. F., Johnsen, I . (1978) Environnientul Pollution 17, 133- 151. Bargagli, R.. Iosco, F. P., D'amato, M. L. ( I 987) Crypptogumie,Bryologie, LichPnologie 8,331 - 337. Boonpragob, K., Nash, T. H. (1990) Environ. and Eqwrim. Bot. 30, 4, 415-428. Brown, D. H. ( I 987) Internationul Symposiunz on Progress und Problems in Lichenology in tht, Eighties, Minster, FRG, March 1986, Proceedings: Peveling, E. (ed.). Bibliotheca Lichenologia. Berlin and Stuttgart: Cramer, Vol. 25, pp. 361 -375. Brown, D. H., Beckett, R. P. (1985) Ann. Bot. 55, 179- 188. Briining, F., Kreeb. K. H. (1993), in: Plunts us Biomonitors. Indicutors,for Heavy Metals in the Terrestriul Environment: Markert, B. (ed.). Weinhcim, New York, Basel, Cambridge: VCH Publisher, pp. 395-401. Burton, M. A. S., Le Sueur, P., Puckett, K . J . (1981) Cunadian Journul q f B u t a n j ~59. 91 -100. Campbell, R. C. ( 1 989) Sfutistics fiir biologists, Cambridge, New York, Port Chester, Melbourne, Sydney: Cambridge University Press, pp. 327. Farkas, E., Lokos, L., Verseghy, K . (1985) Acta Botunica Hunguricu 31, 45-68. Garty, .I.(1 993) in : Plunts as Biomonitors. Indicators,fiw Heavy Metals in the Terrestrial Environment: Markert, B. (ed.). Weinheim, New York, Basel, Cambridge: VCH Publisher, pp. 193 -263. Garty, J . , Ammann, K. (1987) Environmental and E.uperimental Botany 27, 127- 138. Garty, J., Galun, M., Fuchs, C., Zisapel, N . (1977) Wuter, Air and Soil Pollut. 8, 171 - 184. Garty, .I., Galun, M., Kessel, M. (1979) New Phjrologisf 82. 159-168. Goyal. R., Seaward, M. R. D. (1981) Nris Phytologist 89, 631 -645. Goyal, R., Seaward, M . R. D. (1982) N o r Phytologist 90, 85-98, Mikinen, A. (1987) in: Proceedings of the I A B C'onfitmce of Bryoecologj. Symposia Biologica Hungarica: Pocs, T., Simon, T., Tuba, Z., Podani. J. (eds.). Budapest: Akdemiai Kiado, Vol. 35, pp. 777 - 794. Markert, B. (1993) in: Plunts us Biornonitors. Indicators ,for H ~ u v jMetals in the Terresrriul Environment: Markert, B. (ed.). Weinheim, New York. Basel, Cambridge: VCH Publisher, pp. 65- 103. Meenks, J . L. D., Tuba, Z. (1992) in: Biological Indicutoi:r. in Environmental Protection, Kovrics, M. (ed.). Budapcst : Akademia Kiado, pp. 65 - 75. Meenks, J. L. D., Tuba, Z. Csintalan, Zs. (1991) Journ. Hattori Bot. Lab. 69, 21 -35. Nash, T. H. (1989) in: Mrtul Tolerance in Plunts: Evolutionary Aspects: Shaw, .I.(ed.), Boca Raton, Florida: CRC Press. pp. 1 19- 13 I . Nieboer, E.,Ahmmed, H. M., Puckett, K . J., Richardson. D. H. S. (1972) LicI1enologist5,292-304. Nimis, P. L., Castello. M., Perotti, M. (1993) in: Plunts as Biomonitors. Zndiccitorsfor Heavy Metals in the Terresfriirl Emironmen?: Markert, B. (ed.). Weinheim, New York, Basel, Cambridge: VCH Publisher. pp. 265-284. Orban, S. (1991) in: Bakterium-, alga-, gomba-, zuzmo- es inohahatrirozo (Taxonomical Handbook of Hunguriun Flora. Bucteriu, Algae, Fungi, Lic,hens cmd Brj.ophyfes):Simon, T. (ed.). Budapest : Tankonyvkiado, pp. 675 - 777. Pakarincn, P., Tolonen, K. (1976) Ambio 5, 38-40. Pakarinen, P., Makincn, A,, Rinne, R. J. K . (1978) Annales Botanici Fennici 15, 281 -286. Pilegaard, K. (1979) Wuter, Air und Soil Pollut. 11, 77-91. Pilegaard, K., Rasmussen, L., Gydesen, H. (1979) J . Appl. Ecol. 16, 843-853. Puckett, K . J . (1976) Cunudiun Journul of Botany 54, 2696-2703.

434

Z . Tuba

ei

nl.

Schutte, J. A. (1977) The Bryologist 80, 279-283. Seaward, M. R . D., Goyal, R., Bylinska, E. (1978) Nutiirtilist 103, 135- 141. Tuba, Z. ( 1 985) Ahstructa Botanicm 9, 23 I - 239. Tuba, Z., Csintalan, Zs. (1992) in: Pollution Knows N o Frontiers: Schleichcr, K. (ed.). New York: Paragon House, pp. 199-21 I . Tuba, Z., Csintalan, 2s. (1993a) in: Plants as Biornonitors. Irru’icutors f o r Heaiij M ~ t u l sin the Tcwestriul Enuironmmt: Markert, B. (ed.). Weinheim. New York, Basel, Cambridge: VCH Publisher, pp. 329 341. Tuba, Z., Csintalan, Zs. (1993b) in: Plants us Biotnonitors. Indicators Jbr FIeuvy Metals in tliv Terrestriul Environinent: Markert, B. (ed.). Wcinhcim, New York, Basel, Cambridge: VCH Publisher, pp. 403-41 1. Tuba, Z., Lichtenthaler, H. K.. Csintalan, Zs. ( 1 9 9 3 ~ Journul ) of’ Plant Pligsiology, (in press). Verseghy, K . (1991) in: Bakterium-, alga-, gomba-, zuzm6- es mohahatirozo ( Tasonomiccd Handhook of’ Hungarian Flora. Buetrriu, Algae, Fungi, Lichens und Bryopligtes) : Simon, T. (ed.). Budapest: Tankonyvkiado, pp, 575-674. ~

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

23 Comparative Investigation of the Distribution of Chemical Elements in an Aceri tatarico-Quercetum Plant Community and in Stands of Cultivated Plants Margit Kovacs, Karoly Penksza, Gabor Turcsunyi, Laszlh Kaszab, Sandor Thth, and Pal Szdke

23.1 Introduction Multielement investigations, belonging to a new branch of ecology (Lieth and Markert, 1985, 1988; Markert, 1986), are determining the chemical composition the so-called element concentration cadaster - of the components (soil, plants, etc.) of characteristic ecosystems in different zones of the Earth. With the help of basic data (which are mainly dependent on the geochemical environment) some impacts of loads caused by human influences (such as changes in quantitative relationships among chemical elements or appearance of new elements) can be detected. One of the greatest changes in the plant cover of the Earth has been caused by the eradication of the original vegetation (e.g.. forest) and the establishment of arable land on its habitat. The potential zonal plant community at the foot of the Hungarian Northern Central Mountains is wooded steppe (Aceri tatarico-Quercetum; Zolyomi, 1957, 1967). The characteristic soil type of the area is Chernozem Brown Forest soil on loess base rock. One of the still existing stands of wooded steppe is under protection at the boundary of the village of Kerecsend. About 150 - 200 years ago the forest had a larger extension, now crops are cultivated in one part of its original habitat.

23.2 Material and Methods Sampling of soils and plant materials was carried out after the prescriptions of the multielement cadaster investigations (Lieth and Markert, 1985, 1988). Samples of the characteristic species of the tree, shrub, field and moss layers of the Aceri tatarico-Quercrturn community as well as from the crop and weed species of the arable land were collected in June and August, 1991. Data about the element cadaster of the forest community have been already published elsewhere (Kovics et al., 1993). In this contribution only data are presented, which are important for a comparison of the chemical composition of the components of the forest and the arable land.

436

M . Kovac.s ct al.

At the time of investigation wheat and barley were the plants cultivated on the arable land. Samples were taken from the cultivated crops as well as from the following weed species : - in the wheat-field from Convoluulus arvensis, Matricaria muritima ssp. inodoru, Cirsium arvcnse and Bromus tectorum, - in the barley-field from Lathyrus tuherosus, Convolvulus arvensis, Centaurecr cyanus, Cirsiurn a r v ~ ~ n sand e Matricaria maritima ssp. inodoru. The plant samples were divided into different organs (root, stem, leaf, inflorescence) prior to analysis. Methods of digestion were published by Kovacs et al. (1992). In the solutions made of the digested samples the quantity of the following elements was determined by an ICAP-61 Thernio Jarrell Ash type ICP-AES: Al, As, B, Ba, Ca, Cd, Co, Cr, Cu, Fe, Ga, K, Li, Mg, Mn, Na, Ni, P, Pb, Se, Si, Sr, Ti, V, Zn.

23.3 Results and Discussion Besides the geochemical environment the different biogeochemical cycles of elements in the forest and on the arable land had also an influence on the element contents of the plants. The total quantity of elements in the foliage of the five forest tree species decreased in the following order of succession : Qucrcus cerris, Acer tutaricum, Quercus puheseens, Quercus rohur and Quercus petraea. In this rank of order the Ca and the Mg contents played the greatest role (Tab. 1). The lowest Ca, Mg and total element contents were detected in the leaves of Quercu.~prtraea. There was much more Ca in the leaves of trees than in the leaves of cereals, although the Ca content of the soil (upper l00cm layer) of the arable land significantly exceeded the comparable parameter of the forest (see Tabs. 1 -4). ‘lab. 1.

Elcmcnt Contents (pg/g dry weight) of Leaves of Trces in an Aceri tatarico-Quercetum Community near Kcrecsend, Hungary (means of the leaves collected from 10 layers of the crowns of 10 individuals) Qiiercus puhcwons

Al AS

R Ba

c;i Cd Co Cr CU

Fe Cia

11.9 0 48.6 104.7 I 1.284 0.02 0.05 0 6.5 182 0.06

Queriws Lerris

62.6 0 58.6 156 14,349 0.04 0 0 5.9 121 0

QutJrcus robur 36.9 0 52.9 16.1 10,596 0.04 0 0 1.4 148 0

Quercus petrueu 11.7 0 115 99.0 10.582 0.02 0 0 6.3 128.8 0

Acer tataricum 75.1 0 54.2 32.3 14,558 0.4 0 0 4.9 134 0

Element Distributions in Ecosystems ? f a Woodotl S i q i p A v w

437

Tab. 1. (continued) Quercus pubescens

K Li Mg Mn Mo Na Ni P Pb Se Si Sr Ti V Zn

6,656 0.7 2,027 1,465 0.5 122 10.1 1,629 2.5 0.07 115 18.5 1.3 0.9 28.8

Quercus cerris

5,386 0.7 2,100 1,462 0.02 123 6.4 2,010 1.8 0.2 I20 20.4 0.7 0.7 28.9

Quercus rohur

7,413 0.7 2,072 817 0.01 141 7.5 1,875 1.4 0 125 15.6 0.9 0.6 36.7

Quercus petvaea

6,445 0.6 1,616 780 0.1 119 7.5 2,070 1.2 0.2 89.9 23.8 0.7 0.9 31.7

Acer taturicum

6,056 0.3 2,103 79 1 0.02 118 4.3 1,362 2.2 0 99.9 26.5 1.1 0.7 57.7

Tab. 2. Element Contents (itg/g dry weight) of Wheat (Triticum aestivum) Plants on an Arable Land near Kerecsend, Hungary Root Al As

B Ba Ca Cd co Cr

cu

Fe Ga K Li Mg Mn Mo Na Ni P Pb Se Si Sr Ti V

Zn

370 0 0 53.5 2,835 0.8 3.7 6.0 6.5 6,979 3.2 5,127 8.9 1,386 274 0 194 8.6 1,009 6.4 0 168 11.7 54.5 11.5 384

Stem 58.3 0 2.1 19.9 968 0 0 0 2.8 61.5 0 8,65 1 3.0 576 15.3 0 109 0 1,420 0 0 25 1 4.2 0.7 0 17.3

Leaf 290 0 4.9 45.8 6,352 0 0 0 3.1 341 0 9,863 3.2 1.787 56.7 0 104 1.4 1,488 0 0 566 15.8 5.2 0.5 11.8

Spike 62.80 0 3.9 6.2 779 0 0 0 2.3 54.1 0 $21 1 0.7 407 23.2 0 117 0.9 960 0 0 184 2.5 0 0 20.5

Caryopsis 24.6 0 2.4 3.2 675 0 0 0 3.4 32.3 0 3,556 0.6 922 26.3 0 96.5 1.7 2,803 0 0 48.5 1.3 0 0

26.2

438

M . Kovdcs ef (11.

Tab. 3. Elcment Contents (pg/g dry weight) of Barley (Hordoum u u l g m ~ )Plants on an Arable Land near Kerccsend, Hungary Root Al AS B Ba Ca Cd

co

Cr

cu t;c G I

K Li M& Mn Mo Na Ni P Pb Se Si Sr Ti V Zn

242 0 0 43.92 2.01 1 0 0.61 4.02 6.63 4,005 4.5 2.532 5.9 585 185.5 0 259 6.47 404 0 0 98.16 10.43 36.9 3.58 29.46

Stem 65.72 0 2.11 17.47 1.993 0 0 0 2.4 I 79.84 0 14,125 3.53 423 11.83 0 545 0.81 865 0 0 196 9.42 1.62 0 16.55

Leaf 462 0 8.98 52.57 9,699 0.2 0.34 0 3.91 594 2.94 6.576 4.06 1,206 78.04 0 254 1.54 1,154 3. I4 0 259 32.38 8.65 I .27 2 1.62

Spike 82.26 0 3.36 16.94 2,026 0 0 0 3.8 I 223 2.74 6,545 1.1 1 840 26.62 0 208 71.78 1,574 0 0 324 7.68 I .06 0 25.6

Caryopsis 38.4 0 1.68 4.95 707 0 0 0 3.48 32.8 0 4,530 0 950 16.22 0.53 138 6.05 3.424 0 0 49.52 2.15 0.75 0 33.9 1

Average Element Contents (p(s/g dry weight) in the upper 40 ern Layer of Chernozeiii Brown Forest Soils in an Aceri tatarico-Qucrcetum Forest and on an Arable Land near Kerecscnd, Hungary

Tab. 4.

Al As B Ba Ca

Cd CO

Cr CU Fe Ga K Li

Forest soil

Arablc soil

7,366 0 0 Ill 3,398 1.05 7.95 10.7 7.7 7,877 6.7 824 7.7

8,296 0 0 98 5,628 1.7 10.1 13.7 8.9 10,288 7.6 931 7.5

Forest soil Mg

Mn Mo Na Ni P Pb Se Si Sr Ti V Zn

1.883 775 0 29.9 16.4 208 16.3 0 1,159 11.0 13.8 14.2 29.6

Arable soil 190.8 756 0 35.9 20.8 362 15.3 0 1,186 18.3 14.7 13.1 32.1

Element Distributions zn EcosysternJ o j a Wooded Steppe Areu

439

Ca, Ba, Mg and Mn occurred in the leaves of Quercuspubescens and Quercus cerris in very similar amounts. In the same species the highest quantities of heavy metals were present in the foliage (e.g., about 1700 pg/g Mn), whereas only 780- 817 pg/g of this element were detected in the leaves of Quercus robur and Quercus petraeu. The percentage of heavy metals in the leaves of Quercus cerris and Quercus pubescens was 6.2 - 7.1, and this value amounted only to 4.3 in the case of Quercus robur and Quercus petraeu. On the arable land plant species dominate that show a chemical composition which is different from that of the plants growing in the forest (Tabs. 2 and 3). For example, the foliage of the trees contains more Ca, B and Mn, whereas the stem and the leaves of the cereals accumulate more Si. The total element content of the tree leaves exceeds that of wheat and barley. There is a difference between herbaceous dicots and monocots as well; dicots accumulate a higher amount of elements than monocots (Fig. I). The intensive root system of gramineous plants seems to accumulate more Ca and heavy metals especially Fe - than the root system of the herbaceous dicots. The heavy metal contents in the roots of the dicots in the Aceri tutarico-Quercetum forest varied between wide limits (354 - 3416 pg/g). The comparable parameter of the plants belonging to the family of Gramineae was 6107 - 7876 pg/g. The heavy metal contents in the roots of the investigated dicotyledonous plants were 428 - 1251 pg/g, whereas in the roots of Bromus tectorum 3283 pg/g, of wheat 7330 pg/g, and of barley 4241 pg/g heavy metals were detected. The higher Ca and heavy metal contents in the upper soil layer of the arable land might be the result of element concentration by the roots of the cultivated cereals, although fertilizers could also have contributed to the concentration process. Additionally, the element contents of the soils could have been influenced by the following factors: In the forest, the biomass of the trees withdraws a high amount of elements from the biogeochemical circulation for a long time (50- 1.50 years). Nevertheless, from the litter supplemented yearly a significant amount gets back into the soil. From the arable land, a high amount of elements is withdrawn by the caryopses and the straw, whereas seed grains, fertilizers and pesticides supplement the losses. In addition, erosion or - on the relatively flat surface - deflation, leaching and fallout could have some influence on the distribution ofelements in both ecosystems as well. Weeds had only a subordinate importance because of their low abundance. Although their total element contents were higher than those of the crops, their little phytomass accumulated only a relatively small proportion of elements of the ecosystem (see Fig. 1). High element contents are characteristic for the plants of the Compositae family (e.g., Cirsium aruense, Centaurea cyanus, Matricaria maritima ssp. inodora). These species accumulate in higher quantities B and Cd. The high Ca, Mg, Sr and Li contents of Cirsium arvense is also conspicuous. In the leaves of Matricaria maritima ssp. inodora the following elements are present in higher quantities: Cu, Fe, Mn, Pb and Zn. Some microelements as, for example, Cr and Ga were detected only in the roots of wheat and barley. In these organs Al, Co, Fe, Mn, Ni, Pb, Ti and V were concentrated as well.

M .Korhcs et al.

440

k c r i trtorico-qucrcetum

IJg4

irable land

field layer dicots

70000 60000

50000

shrub layer

40000

ri

30000 20000 10000

0

I

Ih

L

> 18 20 22

23 25 21 2 8 2 9

345

678910

Fig. 1. Rank of order for plants in an Aceri taturie(~-Quercetun?community and on arable land based on the total element contents o f leaves (wig). Aceri tatarico-Quercetum: ( I ) Quercus ccrris, (2) Acer tafuricum, (3) Quercus puhrscenz, (4) Quor(w.7 rohur, ( 5 ) Quyrcus petraea. (6) Cornu.~mus, ( 7 ) LiLyustrunii ~ u l p - e ,(8) Crataegus monogynu, (Y) Rosrr crmincr, ( 10) Acrr cumnpcstrc, ( 1 I ) Crutuegus osyucantha, ( 12) Silewe cucuhulus, ( I 3) L~vchriis coronurirr, (1 4) Pulmonario mollissirna. ( I 5 ) Ncpetu punnonica, ( I 6) Asfrugalus glycj~pliyllos.( 17) Lithospermum purpureo-coerul~unz,(1 8) Inulri conyza, (19) Dictamnus albus, (20) Phloniis tuhcrosu, (2 1) Cynuncliurn vincetoxicum, (22) Hypericum montanum, (23) Festuca valesiacu, (24) Bruc~hypodilititn pinnuturn. (25) Ductylis glomeratu, (26) P(JU n~~rnorulis,(27) Festuca sulcata, (28) FIjpnuni c~uprc~ss~orm (29) e , PIrt,yg.yriiun repens. Arable land : (1 ) Cirsiurn aroense, (2) Mutricaria maritiniu ssp. itiodoru, (3) C o n v o h ~ ~ lurocmis, u.~ (4) Triticuni Lit>.viiiwii, (5) 3romu.r f ~ c t ~ r u n(6 i) , Ckviutn Lrrvsnsc. (7) Cetircturru cyanus, ( 8 ) Mart?curia niuririmrr ssp. inodoru, (9) Conrmlvulus urvrwsi.r, (1 0 ) Luthyrus ruhcrosi,is, ( I I ) H o r d e ~ o n lILiIgar(’.

The organic material content in the upper soil layer of the arable land was less (6.3%) than that of the comparable layer o f the forest soil ( 1 1.8%). In the uppermost 10 em layer of the arable land the amount of the investigated elements was higher (29,756 pg/g) than in the same layer of the forest (26,936 psis).This difference might be the consequence of the intensive fertilization on the arable land. In the soil of the arable land the quantities of Ca, P and heavy metals - especially of Fe - are also higher than in the forest. In a lower extent the Cd, Co, Cr and Ni contents are also increased (Tab. 4). The element concentration cadaster of the trees shows wider ranges (0.01- 10,000 pg/g) than that of the other plants (Tab. 5). Two microelements, M o and Se, were detected only in the leaves of the oak species. It seems that crops appearing in the habitat of a former forest decrease not only the diversity of species but also the number of detectable elements (“element-diversity”). The element concentration cadaster becomes narrower, too (Tab. 6).

Element Distributions in Ecosystems of u Wooded Steppe Areu

44 1

Tab. 5. Element Concentration Cadasters (pg/g) of the Leaves of Trees of the Aceri tataricoQuercetum Community near Kerecsend, Hungary

Acer trrturicurn

0.01

0.1

1

10

100

Mo

Cd Li V

cu Ni Pb

Al

Fe Mn Na

Ca

B Ba Si Ti Zn

1000

10,000

Quercus puhescens

Cd co Cia Se

Li Mo V

cu Pb Ti

Al B Ni Sr Zn

Ba Fe Na Si

Ca

Quercus cerris

Cd Mo

Li Se Ti V

cu Ni Pb

Al

Ba Fe Na Si

Ca

Li Ti V

cu Ni Pb

Al

Fe Mn Na Si

Ca

B Fe Mn Na

Ca

Quercus rohur

Cd Mo

B Sr Zn

B Ba Sr Zn

Quercus petrueu

Cd Li Mo Se

cu Ni

A1 Ba Si Sr Zn

Tab. 6. Element Concentration Cadasters (pg/g) of the Leaves of some Crops and Weeds on an Arable Land near Kerecsend, Hungary

Triticum uestiouni

0.1

1 .o

10

100

I000

V

B cu Li Ni

Ba Mn Sr Zn

A1 Fe Na Si

Ca K Mg P

Ba Mn Sr Zn

Al Fe Nil Si

Ca K Mg P

Ti Hordeum vulgure

Cd

co

B cu Cia Li Ni Pb Ti V

442

M . Kooucs

CI

u/.

Tab. 6. (continued) 0.1

1 .0

10

I00

I000

V CU Li Ni

Al B Ba Mn Si Sr

Fe Na

<:a Mg P

K

B Ba

Al Fe M II

Mg P

ca

Ti

zI1 Cd Li Ni Ti V

cu Sr Zn

K

Na

Si

23.4 Summary The potential zonal forest community on the Chernozem Brown Forest soils of the Heves-Borsod Plain at the foot of the Hungarian Northern Central Mountains is Accri tuturic,o-Qucrc,ctum. A part of the forest at the boundary of the village of Kerecsend was put into agricultural use about I50 - 200 years ago. The multielement cadasters of the rest of the forest and of the arable land were determined (in lhe year ofinvestigation barley and wheat were the plants cultivated on the arable land). Molybdenum and selenium could have been detected only in the foliage of trees. Based on the total amount of elements the element concentration cadaster of the trees showed higher values than the foliage of the cultivated plants. Because of the intensive root system of the cereals the quantity of calcium and iron was increased in the upper 40 cin layer of the soil. Acknowledgement. The work was financed by the National Scientific Research Funds (OTKA) of the Hungarian Academy of Sciences.

23.5 References Kovks, M., Turcsinyi, G., Penksza, K., Kaszab, L., Sz6ke. P. (1992) in: P/un/.s us hi om on it or.^ j h r / i e o c ~niclal po//utiori 01' tcrrc.strio/ riii!il.Oiiiii('ii/.s, Markert, R . (ed.), Wcinhcim - New York : VCIl Publisher Inc., pp. 495 50.5. Kovics, M., Pcnksza, K., Turcsinyi. G., Kasmb, L.. Sr6ke. P. (1993) Multielemenl-An~ilyseder Arlen eines Waldsteppen- Waldes. Ph?./oc.oeriolofiitr)/~~~ir~ (Berlin - Stuttgart) 23. pp. 257 - 267. Licth, H., Markert, B. (1985) Concentration cadasters of chemical elements in contrasting ~ ~ /322 i U ji ~324. ~t~, ecosystcms. N ~ t u r l ~ ~ i . s . ~ c ~ n .12, Licth, H., Markert, B. ( 1988) Ar~fs/lc//rrngund Au.sii~o/zrngiiko.s!,s/crncrri,r I : ' l ~ ~ / ~ i ~ ~ ~ / - K ~ ~ / i i r ~ t t / ~ ~ i / K u / L I . s / cBerlin ~. - lieidelberg- New York -London - Paris - Tokyo: Springer-Vcrlag. Markert, B. ( I 986) Multiclcnicnt-Analyse: Mogliche Darstcllungswciscn von Melldaten. /;rc~.seniu.s J . A n d . Ch011. 327, 329- 334. Zblyomi, B. ( 1957) Dcr Tatarenahorn-Eiclicii-LiiDwilld der zonalen Waldstcppc. Actrr Bot. A c ~ i d . S r i . //ifng. 3. 401 - 424. i r n&cri.s Zhlyomi, R. ( I 967) in : f;iihrcir t k ~ rk;.ukursioneu tkes Irr/ernrr/ionrr/i,rr Ccohotuni,~c/ioi~ ~ ~ r ? ? ~ ) o . s i in Vtkihttj/, Ur?gurn: Zblyomi, B. (cd.), Budapest: Hungarian Academy of Sciences, pp. S1 -54. -

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

24 Sampling of Tropical Terrestrial Plants with Particular Reference to the Determination of Trace Elements Ranjith Juyasekera

24.1 Introduction According to Polunin (1960), there are five main tropical vegetation types, namely the tropical rain forest, forests with seasonal rhythm or monsoon types of forest, the savanna woodland, the thorn woodland and the hot desert. Tropical forests consist broadly of two levels: the high forests of the hot and humid regions which lack a pronounced dry season, and the forests developed in seasonally dry climates. Little is known about the numerous biomes which are found in the tropical forests. With respect to the dynamics of organic and inorganic matter, tropical plant communities differ greatly from plant communities elsewhere due to their special characteristics such as the structure and composition of life forms, the relatively long lifetime of leaves, the complex and multilayered structure of the forests with their wide floristic diversity of species. Therefore, the sampling of plants in tropical forests poses its own problems. In contrast to the temperate regions, no marked seasonal variations can be observed in the tropical rain forest, but the vegetation of mountains is considerably affected twice a year by the influence of heavy monsoon rains. Uptake rates and interactions of trace elements in tropical plants are significantly affected by high levels of humidity, light intensity, temperature as well as by differing life cycles compared to plants growing in the Northern Hemisphere (Epstein, 1972; Barber, 1984). In established forest ecosystems, atmospheric input, being the only input pathway, is very significant (Adriano, 1986). In this respect, established forests such as the montane forest ecosystem would be ideal for long-term monitoring of atmospheric input of various chemical elements or constituents. Plants are sampled for many purposes (Lieth and Markert, 1990; Markert 1991) such as to assess the baseline concentration levels, availability of elements in soil, soil-plant chemical relationships, plant nutrient status, inorganic and organic pollutants or to investigate pathogenic problems. With respect to the pollution trend monitoring, investigation of biological and environmental materials from very remote and anthropogenically unaffected ecosystems such as tropical montane rain forest would provide interesting information on pollution trends as virtually no uncontaminated material from preindustrialized times is available in developed countries (Jayasekera and Rossbach, 1993). Irrespective of the purpose of sampling, sound sampling without any contamination is of vital importance to derive reliable and accurate information about the trace element chemistry of the plant population being studied. The objective of this

444

R. .Iqn.rr~keto

paper is to give a general account on some important aspects of sampling of tropical terrestrial plants for trace element analysis with special emphasis on the montane rain forest.

24.2 Statistical Aspects It is frequently assumed that measurements of many natural phenomena are normally distributed or nearly so. Thcrefore, the normal distribution is the most common symmetrical distribution and also most important in biological work. The normal distribution is one of the simplest distributions and is the basis of most parametric statistics, for, e.g., analysis of variance, correlation, regression and construction of confidence intervals (Zar, 1984). Therefore, before performing any parametric statistical test on data obtained from chemical analysis, it is advisable to ensure that the set of data being studied is normally distributed. Observations can be tested for normality performing an n-score and the Kolmogorov-Smirnov test (Zar, 1984) using standard statistical software packages like SAS (Helwig and Council, 1979), MINITAB (Shaefer and Anderson, 1989) or SYSTAT (Wilkinson, 1988). All elements may not show the normal distribution. In natural ecosystems, the elements that are essential for plant growth tend to be normally distributed in the plant tissues, while nonessential elements seem to have distributions with positive skewness (Pinder and Smith, 1975). Skewed distributions can not be fully described by the mean and standard deviation alone and significance tests must be modified to satisfy assumptions of equality of variances and normality of data. However, precision of measurements decreases if practical considerations limit the size or the number of samples that can be taken. For example, in protected areas the sampling method used depends upon the degree of destructive sampling permissible. Randomization and replication are two important ways of avoiding systematic error and poor precision (Zar, 1984). In some cases, however, replication may be impracticable due to site heterogeneity.

24.3 Plant Sampling Randomization is one of the most important methods of avoiding bias in experimental data. It involves selecting some proportion of a population so that there is an equal chance of every part selected being representative of the whole with regard to variability. When sampling at the individual level, for example, in an established primary forest, it is necessary to ensure that the trees being sampled are representative of the different tree species occurring in the forest. Subsequently, plant parts are taken at random, analyzed and the means computed and compared statistically. For most physiological studies, leaves are frequently sampled as they normally represent the most active component of plants (Vitousek and Sanford,

Sampling of Tropical Plants

445

1986; Jayasekera, 1992). In terms of quantitative significance, leaves and stems are the most important, and they represent short- and long-nutrient retention, respectively. It is always important to understand as much as possible of the ecology of the system being studied, especially the seasonal changes in activity, differences in the chemical composition of different species (Markert and Jayasekera, 1987; Jayasekera, 1992) and also growth stages. In sampling plant material for trace element analysis, two major aspects have to be kept in mind. Firstly, it is of central importance that the sample of a particular plant part taken at random is representative of the tree canopy under consideration. When sampling leaves, a pool of at least 25 subsamples of fully developed leaves representing sun, shade and intermediate leaves should be taken from different positions of the tree canopy. The total number of leaves in a pooled sample might be about 500. The pooled sample should be placed in a precleaned, large polyethylene bag and be transported to the laboratory as soon as possible for sample preparation. Senescent or damaged leaves, either by insects or mechanically, should not be sampled. Secondly, it is necessary to ensure that the sample taken is free from soil or dust contamination which is a serious problem in trace element analysis. Except for studies dealing with the extent of atmospheric deposition which is a potentially productive line of research, leaf surface contaminants can be wiped clean with a paper tissue soaked in double distilled water (Jayasekera, 1992). Intensive washing of leaves should be better avoided as it involves some leaching of ions from the tissues (Bowen, 1979). A high content of titanium together with aluminium, scandium and zirconium is a useful indication for soil contamination (Bowen, 1979). Cutting instruments like stainless steel inevitably cause contamination when used for sampling. It is also important to emphasize here that when working at the ecosystem level, all the vegetation on a statistically selected plot is cut, separated into vegetation components such as canopy leaves, canopy stems, understory leaves, understory stems, roots, fruits, flowers and sometimes litter. Subsequently, chemical analyses are performed on samples of each of the vegetation component. Although this method is widely used by ecologists for vegetation analysis, analytical results obtained in such studies are often doubtful with respect to precision and accuracy. It is necessary to ensure that the content of a particular trace element detected is of plant uptake and not due to any contamination. Tab. 1 shows baseline concentration levels of selected trace elements in leaves of six floristically dominant trees sampled at two different times, i.e., during the monsoon rainy period (September 1990) and in the dry period (March 1991) on five sites in a montane rain forest at an altitude of 1850 m in Sri Lanka. As the research sites were located within the premises of Strict Natural Reserve, the degree of destructive sampling permissible was limited. Sampling was performed according to the procedure described above. The details of the study site, sampling, material processing and chemical analysis procedures can be found in Jayasekera (1 992). Tab. 1 clearly shows that the leaves collected during the heavy, monsoon rainy season possess higher contents of sodium which could probably be attributed to the atmospheric input (Adriano, 1986) of sea-borne sodium transferred by monsoon rains (Veneklaas, 1990). In contrast, aluminium contents tend to be higher in the dry season, except in Eugenia mabaeoides leaves, probably caused by the dry

446

R. Jntwsrkrrci

deposition of dust consisting mainly of silica and aluminium (Deshmukh, 1986) during the windy, dry season. As the E. mubueoides leaves belong to the leaf size class “nanophylls” (average leaf area 2.9 cm2) with the smallest leaf area per leaf (Jayasekera, 1992), the contribution of dry deposition seems to be minimal. With regard to the other elements shown in Tab. 1, it seems that seasonality does not tend to affect considerably their contents in leaves. However, biological variation of trace element concentrations in leaves over the five different sites across the montane rain forest shows that the chemical variability between sites is considerably low, except for few cases (Tab. 1). Tab. 1. Comparison of Foliar Concentrations (mg/kg dry wt.) of Selected Trace Elements of Six Floristically Dominant Tree Species on Five Sites in a Montane Rain Forest in Sri Lanka ( I ) in the monsoon rainy season, (2) in the dry season; figures in parentheses indicate biological variation (Yo);analytical values for NlST citrus leaves (SRM 1572) are also given Al

I

0.58 (58)

2

0.68 (35) 0.13 (49)

1

2

I

-7

0.23 (52) 0.06 (37) 0.12 (58)

1

0.1 (32)

2

0.15 (33) 1.86 (30)

I

1

1.4 (48) 0.05

2

0.1 I

Found

(60) -

Cerlified

-

2

NlST Citrus leaves

Co

-

Cr

Cu

Fe

Mn

Na

Ni

Zn

Sampling

cf

Tropical Plants

447

24.4 Concluding Remarks Trace elements in plants are of general concern because of their importance in both animal nutrition and plant growth. However, the natural, baseline levels of trace elements in plants are affected by several factors such as plant species, plant parts and age, soil and climatic conditions. Two major problems in arriving at precise and accurate values of trace element concentrations are errors due to sampling and contamination, provided chemical analysis is performed under optimum conditions. Therefore, the spatial design and the method of collecting samples are central for accurate determination of trace element concentrations. Complete standardization of sampling methods, especially in tropical terrestrial ecosystems is undesirable and impracticable mostly due to site heterogeneity and complexity, but wide adoption of certain general principles will help to make the results of different investigations comparable. Depending upon the objective, samples taken should be prepared for trace element analysis. Acknowledgements. This work is a contribution to the biogeochemistry of a tropical montane rain forest project funded by the International Foundation for Science (IFS) in Sweden. Dr. M. Bredemeier performed chemical analysis on selected samples at the University of Gottingen.

24.5 References Adriano, D. C. (1986) Truce Elements in the Terrestrial Environment, New York, Heidclberg, Tokyo: Springer Verlag. Barber, S. A. (1984) Soil Nutritwt Bioavuilahility, Toronto, Singapore: John Wiley & Sons. Bowen, H. J. M. (1979) Environmental Chemistry uf the Elements, London: Academic Press. Deshmukh, I. ( I 986) Ecology and Tropical Biology, 1st Edition, Oxford, London : Blackwell Scicntific Publications. Epstein, E. (1972) Minerul Nutrition uf Plunts: Princijhs & Perspectives. New York: John Wiley & Sons. Helwig, J . T., Council, K. A. (1979) S A S User’s Guide,, Gary, North Carolina: SAS Institute Inc. Jayasekera, R. (I 992) Vegetatio 98, 73 - 8 I. Jayasekera, R., Rossbach, M. (1993) The Science qf the Total En:nt~ironmcnt, in press. Licth, H., Markert, B. (eds.) (1990) Hiwent Concentration Cudasters in Ecosystems: Methods and Assessment, Wcinheim: VCH Publisher. Markert, B. (1991) in: Modern Ecology; Basic and Applied Aspects, Esser, G., Overdieck, D. (eds.), Amsterdam, London, New York, Tokyo: Elsevicr. pp. 275-293. Markert, B., Jayasekera, R. (1987) Journal qf’Pluwt Nutrition 19 (7), 783-794. Pinder, J. E., Smith, M. H. (1975) in: Minerul Cycling in Southern Eco.ystem.s, Howell, F. G., Gentry, J. B., Smith, M . H. (eds.), Georgia: Technical Information Center, pp. 107- 125. Polunin, N. (1 960) Introduction to Plant Geography, London: Longman. Shaefer, R. L., Anderson, R. B. (1989) The Sfudeni Edition of’Minitah - User’s Manual, California, New York: Addison Wesley Pub. Company, Inc. & Bcnjamin/Cummings Pub. Company, Inc. Vcncklaas, E. J. (1990) Journal Uf’EcoIogy 78, 974-992. Vitousek, P. M., Sanford, R. L. (1986) Annual Review qf Ecology and Systematics 17, 137- 167. Wilkinson, L. (1988) S Y S T A T : The Systemfiir Statistics, Evanston, Illinois: SYSTAT Inc. Zar, J. H. (1984) Biostatistical Analysis, 2nd Edition. Ncw Jersey: Prentice-Hall, Inc., Englewood Cliffs.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

25 Sampling in the Stemflow and Throughfall Areas of Forests Gabor Turcsanyi, Karoly Penkszu, IrCn Siller, Ern6 Fuhrer, Sundor Thth, Margit Kovucs, and Saraltu Biittner

25.1 Introduction A part of the precipitation falling upon forests is retained by the foliage of the trees and is called interception. Unintercepted water reaches the floor or the understory either assembling and running down on the surface (bark) of the stems, or falling through the foliage while touching (and partly also penetrating) the leaves or the twigs of the trees. These parts of the precipitation are called stemflow and throughfall, respectively. Only in the gaps some precipitation directly reaches the shrubs, the field layer or the soil surface. When sampling in a forest or underneath trees one has to be very cautious, since the site of sampling can greatly influence the results of investigation. Differences in the chemical composition and other characteristics of samples cannot only be affected by the nutrient uptake of roots, the uneven deposition of litterfall and the shadowing effect of foliage, but also by stemflow and throughfall, in which the precipitation water, following different routes, can get either collected or dispersed, or can even undergo chemical changes. The distribution of the precipitation between stemflow and throughfall depends mostly on the physical properties of the tree crown and the bark surface, whereas the chemical composition of both fractions is strongly influenced by the chemical properties of the touched or penetrated substances. Thus, differences in the quantity as well as the composition of stemflow and throughfall can result in the formation of different habitats on the forest floor, and, as a consequence, the rise of different cominunities with species differing in their ecological, morphological, chemical, biochemical, physiological parameters etc. Attention to the effects of stemflow and throughfall has been focused only in the last decades, since the environmental dangers of acid precipitations were discovered.

25.2 Literature Data 25.2.1 The Amount of Stemflow and Throughfall in Stands of Different Tree Species Several data in the literature, originating from various climatic belts of the world suggest that different areas of forest floors receive uneven amounts of precipitation. In Pinus fuedu stands 20% (Switzer et al., 1988), in a Pinus sylvestris stand 12%

450

G. Tirrcsdn~,iet r i l

(Santa-Regina et al., 1989), in Piccu sirchensis plantations 13% (Anderson et al., 1990) and 18% (Anderson and Pyatt, 1986), in a Quorcusprinus forest 8.5% (Potter, 1992) of the unintercepted precipitation reached the ground a s stemflow. Pontailler ct al. (1988) found that stenitlow was 4 % for beech and negligible for oak. In a Qucrcus c w r i s coppice stand 7% (Mugnozza et al., 1988), in some Agatlzis lorcitzth~fdiaplantations 5.26 - 5.35% (Pudjiharta and Basuki, 1988), i n a Pinits rudiutu stand 10.8 & 1.9% (Huber and Oyarzun, 1990) of the precipitation were rccorded as steinflows. Due to stemflow, the soil near the base of trees in some Quercus uirginiana mottes received about 22.2% of annual precipitation (Thurow and E. internzediu was about et al., 1987).Annual stemflow down Euc,u(}~fu.spiliiluris 4% of annual rainfall but steniflow from individual rain events was up to l6'YO (Prebble, 1987). In a tropical rain forest in Venezuela stemflow was 8 % of precipitation (Jordan and Heuveldop, 198 I). In an Australian tropical rainforest stemflow volumes were found to be as much as two orders of magnitude greater than the voluinc of incident rainfall expected in a rain gauge occupying an area equal to thc trunk basal area (Herwitz, 1986b). In a New Zealand forest the 7 in tall palm Rhopu1ostyli.s supidu directed more water to stemflow than did much larger trees of other species in the study area (Enright, 1987). Raich (1983) also found that stemflow in a tropical forest averaged 9%) of rainfall, 92% of which was collected by palm trees, and more than 74% was collected by trees less than 10cm in diameter. Within an aspen-birch forest overstory throughfall generally contributed a much larger solute load than stemflow, whereas undcrstory stemflow contributed about six times more solutes than understory throughfall (Price and Wattcrs, 1989). Nevertheless, lots of publications suggest that steinflow does not play a significant role in the distribution of precipitation in all kinds of forests. In several different forest types stemflow amounted to not more than 3% of the total rainfall (e.g., Nizinski and Saugier, 1989; Almeida et al., 1990; Ettala, 1988; Johnson, 1990; Lloyd and Marques-Filho, 1988). The amount of throughfall and stemflow can be influenced by species (Navar and Bryan, 1990), individuals (Navar and Bryan, 1990), years and seasons (Cape et at., 1991), location and size (Switzer et al., 1988), crown area (Oyarzun et at., 1985; Weihe, 1985), branch gradient (Weihe, 1985), age of trees (Johnson, 1990), and possibly also by the proportion of the crown overhung by other trees (Wcihc, 1985). Stemflow in an eucalypt forest was influenced by event type, climate variation, tree characteristics, species composition, sylvicultural interventions like thinning and rain angle as well, this last one having a major effect (Crockford and Richardson, 1990a, b, c). The stemflow area of the beech tree (Fugus sylvuficu) attracts more precipitation than the rest of the forest area, on account of its infundibular treetop (Wittig, 1986a). Spatial variability of throughfall can be also dependent on the distance from the tree stem and on the density of the canopy cover (Johnson, 1990). Theoretical areas and distances over which stemflow is spread around Diospyros iexana, Ac,u('I'u,furnPSiunuas well as Prosopis luevignta trees averaged 0.320 mz and 0.115 m2, and 0.30 and 0.15 m, respectively (Navar and Bryan, 1990). Trees in agricultural cropland also change the amount of precipitation which reaches the ground in their immediate vicinity (Darnhofer, 1989).

Sampling in Stemflow and Tliroughfiill Areas

45 1

25.2.2 Chemical Composition of Stemflow and Throughfall Differences in the chemistry of the stemflow and throughfall waters were measured by several authors. As a consequence of the different chemical composition the pH-values of stemflow and throughfall waters also differ. Stemflow is generally more acidic than throughfall or precipitation (Miller, 1984; Wittig, 1986a; Edmonds et al., 1991; Oh et al., 1987; Rampazzo and Blum, 1990). The pH of throughfall can also show substantial spatial variation (Tajchman et al., 1991). Rainwater reaching the ground as throughfall usually shows an increase in pH, except in areas where sulphur dioxide deposition can be presumed to be high (Miller, 1984). Mahendrappa ( 1 989) established that stemflow components contributed to acidity, while throughfall reduced acidity of rainwater. Elemental concentrations in precipitation increase as it moves through the canopy generally in the order stemflow > throughfall > incident (Baker and Attiwill, 1987; Switzer et al., 1988). In some forest soils in the Netherlands stemflow caused extremely high localized inputs around the boles of trees, up to 1 1 times the input by throughfall (Ivens et al., 1989). Marschner et al. (1991) found that nutrient concentrations in stemflow water were 8 to 12 times higher than those in precipitation water. Sulphate, nitrate and phosphate are present in stemflow in higher quantities than in throughfall (Potter, 1992; Asche and Beese, 1987). In some oak forests the contribution of stemflow to forest floor input of S was 12- 18% (Kelly, 1984a). Organic acids are also important contributors to acidity in throughfall and stemflow (Edmonds et al., 1991). The chemistry of dead-tree stemflow is qualitatively different from that of live trees, with dead-tree stemflow contributing very large proportions of the total amounts of nitrate and phosphate available within the system (Watters and Price, 1988). In two tropical forest sites in Australia the stemflow inputs of Mg and K from single rainy days were higher than the mean annual rainfall input and almost of the same order of magnitude as the mean annual throughfall input (Herwitz, 1986a). The concentration of K is generally higher in stemflow than in throughfall (Potter, 1992). Leonardi and Fliickiger (1987) also state that K' shows a very high rate of leaching. Fliickiger et al. (1986) establish that a close correlation exists between rainfall pH and K content of stemflow. Highest concentrations of NH, were found in stemflow (Edmonds et al., 1991). In the vicinity of a pig fattening facility the seasonal average in winter yielded the highest amount of ammonium also in stemflow (Hartung, 1985). The greatest part of Cu available in the forest soil is also returned by stemflow (Santa-Regina et al., 1989; Santa-Regina and Gallardo, 1989). Dissolved organic carbon concentrations increase in both throughfall and stemflow (Koprivnjak and Moore, 1992). Beryllium concentrations in stemflow and throughfall are relatively low, albeit two to three times higher than those in rainfall (Neal et al., 1992). In industrial areas the distance from point sources of pollutants exhibits its greatest influence on stemflow inputs (Kelly, 1984b).

452

G. Turr cLinyr ef (11

The pH value and the element concentration of stemflow and throughfall are influenced by species (Yadav and Mishra, 1985c; Santos et al., 1981; Crockford and Richardson, 1987; Mahendrappa, 1987; Edmonds et al., 1991; Tajchman et al., i991), foliar senescence (Switzer et al., 1988), characteristics of stem surface (Crockford and Richardson, 1987; Satoh et al., 1989), branch inclination (Herwitz, 1987), crown area and stem diameter (Klemmedson et al., 1983; Yadav and Mishra, 1985c), successional stage (Potter, 1992), particulate and aerosol deposition due to seasonal differences in fossil fuel use and industrial emissions (Switzer et al., 1988; Werner et al., 1987), chemical composition of the precipitation (Edmonds et a]., 1991), duration of storm events (Potter, 1992), and amount of rain (Santos et al., 1981; Garcia-Bellido et al., 1989).

25.2.3 Physical Changes in Soils Due to Stemflow The soil samples may be more repellent to water of throughfall origin, and even more repellent to stemflow than to distilled water (Crockford et al., 1991). As a consequence of intensive weathering contaminated stemflow areas show higher clay and silt contents and smaller aggregates, as well as clay illuviation, compared to the non-contaminated reference areas between trees (Rampazzo and Blum, 1992). Changes in the pore distribution and the surface of the soil aggregates could have been also detected (Rampazzo, 1991).

25.2.4 Chemical Changes in Soils Due to Stemflow Chemical and mineralogical changes also occur in forest soils due to acid atmospheric depositions. The lowering of pH in the soil around the trunk of a 50-year-old mountain maple was reported by Grenzius (1988). In beech stands a zone of extreme soil acidification (1- 5 m2 in size) is formed around the base of the beech stems, the size and shape of this zone depending on various factors but always extending farthest downslope (Glatzel et al., 1983; Kazda and Glatzel, 1984; Glavac et al., 1985; Glatzel and Kazda, 1985; Schulte and Spiteller, 1987). A soil pH decrease of 1.2 units was observed for a distance of up to 1.2m from boles of beech trees (Spelsburg and Crossmann, 1984). Results of Falkengren-Grerup (1989a) suggest that stemflow reduces pH and base saturation in topsoil of mull soils especially within 150 cm of the stem. The acidification in a beech stand extended to a depth of approximately 60 cm (Offenberg, 1986). Pallant and Riha (1990) state that soil fungi endemic or more active in the near-tree environment of red pine and Norway spruce make a significant contribution to acidification of soil surface horizons. Falkengren-Grerup and Bjork (1991) establish that soil around stumps may not recover fully from acidification, or does so only slowly after the initial 15 years of

Sampling in Stemflow and Throughfall A r i w

453

recovery. According to Crabtree and Trudgill (1985) stemflow produces large concentrated point-source inputs to the soil. In the stemflow area higher contents of C, N, S as well as Fe,O, and of the heavy metals Pb, Zn and Cu were observed, accompanied by very low base saturation (especially of Ca and Mg) and high A1 saturation (Glatzel et al., 1983; Kazda and Glatzel, 1984; Glavac et al., 1985; Glatzel and Kazda, 1985; Rampazzo and Blum, 1990,1992; Rampazzo, 1991; Hajduk, 1987; Werner, 1988; Grenzius, 1988). Werner et al. (1987) found that Ca, Mg, Cd and Zn contents were lower, whereas Al, Pb and Cu contents were higher in the trunk base area than in areas not influenced by stemflow water. At the bases of the trunks of isolated specimens of Quercus rotundifolia Escudero (1985) measured a higher mineral content, whereas in the area beneath the crown but away from the trunk, increases were found in nutrient contents, pH and C/N ratio. At the base of spruces mean "'Cs activities were about twice, and under beeches 5 to 15 times as high as under more distant parts of the canopy (Forster et al., 1991). The cation exchange capacity decreases with increasing exchangeable Al content in the stemflow area, and the amount of Fe is higher than that of exchangeable Ca and Mg (Rampazzo and Blum, 1990). The amount of exchangeable Ca, Mg and Mn is generally low and the quantity of exchangeable K is higher close to the stem (Kazda and Glatzel, 1984; Glatzel and Kazda, 1985; Falkengren-Grerup, 1989a). In areas south of the Alps in Europe, stemflow causes changes in the soil at the stem base of beech trees only in polluted areas (Glavac et al., 1985). Samples taken in beech stands in the zone of infiltration show the highest variation in soil pH, organic material contents, Cd, Pb, Cu and Zn contents, while the zone between the crowns is least variable (Hertz and Angehrn-Bettinazzi, 1989). Soil changes are greatly influenced by its buffering capacity (Werner et al., 1987).

25.2.5 Living Organisms Influenced by Stemflow 25.2.5.1 Microorganisms Nitrification, dehydrogenase and alkaline phosphatase activities are inhibited in the zone influenced by stemflow, while ammonification increases (ZechmeisterBoltenstern, 1989; Kinzel et al., 1989; Werner, 1988).

25.2.5.2 Fungi Zhao (1989) found that soil acidification is favorable for vigorous beech mycorrhizas in the humus, provided that the soil pH is 2 4.0. The proportion of vital mycorrhizas decreases in the carbonate buffering zone, while the proportion of damaged mycorrhizas increases.

454

G . Tirrcsdnyi et d.

25.2.5.3 Mosses and Lichens Following the Chernobyl nuclear accident increased contamination was recorded for moss and lichen samples from tree stems with greater stemflow, particularly beech (Guillitte et al., 1990). The contents of lead in mosses are significantly higher in the trunk base area than they are in the center ofthe stand (Clement and Wittig, 1987). Acid precipitation affects epiphytes by reducing the buffering capacity of bark and increasing bark acidity (Farmer et al., 1991). The extent to which this occurs depends on tree species, soil chemistry and the nature of atmospheric inputs.

25.2.5.4 Other Plants In spring 1979, Carex pilosu died out in large patches downhill of the boles of old beech trees in the Vienna Woods (Glatzel et al., 1983). Differences were observed between the floristic cover of the trunk base area and that of soil outside these areas (Neite and Wittig, 1985). In the trunk base area an increase of acidity indicator plants and a decrease of alkalinity indicator plants was observed (Wittig, 1986a, b). Crozier and Boerner (1984) established that several herb species occurred at significantly closer mean distances to the base of Quercus alha or Fagus grandifiilia. The effects of higher soluble iron and lead contents on the distribution of herbaceous pants are discussed by Neite (1989). Falkengren-Grerup (1989a, b) found that the cover of field-layer plant species was closely correlated with pH, base saturation, exchangeable Ca and organic matter in the soil. Disturbed conditions of beech seedlings from natural regeneration in zones directly affected by stemflow were observed by Huttermann and Gehrmann (1 982). A highly interesting phenomenon is, that stemflow of Eucalyptus spp. yields understory suppressive leachates of allelochemicals (May and Ash, 1990).

25.2.6 Impact of Stemflow on the Roots of Trees Plant growth was reduced on the acidic substrate of the infiltration zone of beech stemflow and root damage, especially dieback of root tips could habe been observed (Glatzel and Kazda, 1985). Turcsrinyi and Fangmeier (1990) detected higher lead contents in beech roots of the stemflow area than in those of the throughfall area.

25.2.7 Impact of Stemflow on Animals The mesofauna of soils in the infiltration zone of highly acid- and heavy metalenriched stemflow of beech woods indicates changes in soil conditions (Kopeszki, 1988). The abundance of soil animals decreases, the dominance and species

Sampling in Stemflow and Throughfall Areas

455

composition alter and many species disappear in the infiltration zone (Kopeszki, 1992). However, in the investigation of Stockli (1991) the dominant small species did not prefer any particular micro-site, and species diversity in the micro-sites was quite similar.

25.2.8 Some Contradictions Some results conflicting with the acid stemflow theory were also published in the literature. In the investigation of Yadav and Mishra (1985a) stemflow samples showed higher pH values and higher concentrations of K, Ca and Na than throughfall and rainfall. Anderson (1991) found that in some oak stands nutrient addition via stemflow seemed to be the main reason for considerably higher pH values and calcium contents near the stems, and for an altered herb composition, mainly consisting of Mercuriulis perennis and Hepatica nobilis, in the stem zone of many trees. Similar conclusions were drawn by Moore and Dubreuil (1987) after their investigations carried out in two stands dominated by Fugus grandfolia or Acer saccharurn. No differences between throughfall and stemflow areas in a mull rendzina indicating optimal biological soil conditions and a large buffering capacity under a Fagus syluatica stand could have been detected by Scholten (1990). Wilmanns (1985) states that it is not stemflow water, as is often claimed, which makes moss rings around stems, but the wind, whirling around stems and keeping their immidiate surroundings free of litter. On a steep vegetated sand dune in Queensland a depression on the downslope side of tree bases created by enhanced erosion and accretion on the upslope side was attributed to the tree acting as an obstacle in the path of downslope movement of sand grains (Prebble, 1987). Papritz (1987) discusses in his review paper the “litter” hypothesis, the “stemflow” hypothesis, and the importance of pollution in changes of the chemical, morphological and physical properties of the soil and the herbaceous vegetation at the base of forest trees. His opinion is that most of the changes in the soil are attributed to the influence of stemflow, and it is concluded that in regions affected by pollution the stem base zone of beech trees is suitable for evaluating the pollution load and its effects in forest ecosystems.

25.3 Material and Methods In Hungary, the pH and the chemical composition of the equilibrium soil solution as well as the element accumulation by beech roots was investigated in two stands. One stand was growing on a slightly podsolized brown forest soil (Brennbergbanya) and the other on a Ranker soil (Parad). Samples were taken from the stemflow and the opposite sides of the stems and from the throughfall area on both sides of

456

G . Turcsdnyi ('t d.

the trees mentioned. Both stands were situated on slopes, thus exposing one side of the tree trunks to extreme stemflow during heavy rainfalls, whereas the opposite side mostly remains dry. Soil samples were taken from the A,, horizon. Root samples were collected at the same sites. The soil samples were dried and homogenized. Then, equilibrium soil solutions were made following the procedure suggested by Schierl et al. (1986). These equilibrium soil solutions were analyzed. The plant material was carefully washed by tap water. The cleaned plant samples were dried and homogenized. 200 mg of the samples, respectively, were weighed for digestion. Digestion was carried out in tetra-fluorine-ethylene (teflon) vessels under pressure with nitric acid at 170 "C during 10 hours. After cooling the solutions were filtered and analyzed. All extracts were analyzed by a Thermo Jarrel Ash ICP-AES for 26 elements (Al, As, Ba, B, Ca, Cd, Co, Cr, Cu, Fe, Ga, K, Li, Mg, Mn, Mo, Na, Ni, P, Pb, Se, Si, Sr, Ti, V, Zn).

25.4 Results and Discussion The pH of the podsolized brown forest soil was more acidic than that of the Ranker soil (Tabs. 1 and 2). Tab. 1. pH Values and Element Contents (pg . ml- ') of Equilibrium Soil Solutions in the Stemflow and Throughfall Areas of a Beech Stand in Brennbergbinya, Hungary Values are the means of 4 parallels Element

PH Ca Sr Ba

Mg Na K P Fe Al B

Mn Ni Si Ti Zn

cu V

Throughfall area, uphill (a)

Stem flow area, uphill

Throughfall area, downhill

(b)

Stemflow area, downhill (c)

4.20 42.70 0.18 0.19 8.24 4.2 1 32.23 2.64 1.71 6.78 0.18 14.31 0.05* (c) 18.20 0.13 0.37 0.03 0.22

4.18 49.52 0.24 0.28 8.15 4.65 31.82 2.53 1.60 7.23 0.17 20.92' (c), * (d) 0.05* (c) 21.62* (c) 0.10 0.47 ' (c) 0.02 0.14

3.60' (a, b, d) 17.78* (b, d) 0.07 0.12* (b) 7.80 6.04 72.04* (a, b, d) 3.44 3.16 6.15 0.15 6.15 0.02 15.85 0.07 0.31 0.03 0.18

4.33 49.68 0.20 0.19 9.19 3.97 32.36 2.69 1.48 5.75 0.24 11.62 0.05* (c) 19.43 0.11 0.38 0.05 0.21

' Significant diffcrcncc ( p < 0.01) from the denoted samples (Studcnt's t-tcst).

*

Significant diffcrcncc ( p < 0.05) from the denoted samples (Studcnt's t-tcst).

(4

Sampling in Stemflow and Throughfall Areas

457

Tab. 2. pH Values and Element Contents (pg . ml-') of Equilibrium Soil Solutions in the Stemflow and Throughfall Areas of a Beech Stand in Parad, Hungary Values are the means of 4 parallels Element

PH Ca Sr Ba Mg K P Fe Al B Mn Si Zn CU V

Throughfall area, uphill (a)

Stemflow area, uphill (b)

Stemflow area, downhill

Throughfall area, downhill

(4

(4

6.62 43.05 0.17 0.20 6.34 19.83 1.75 0.26 0.51 0.04* (b) 0.47 26.65 < 0.06 <0.01 0.006

6.53 73.86 0.28 0.43 9.74 30.22 2.65 0.30 1.06 0.08 1.67 30.26 0.10 0.01 0.008

4.64' (a, b, d) 15.72' (b), * (d) 0.05 0.13' (b), * (d) 4.46 40.74 1.87 1.25' (a, b, d) 3.01 ' (a, b, d) 0.07 3.12 26.65 0.23' (a), * (b) 0.02' (a), * (b, d) 0.010

6.64 55.27 0.21 0.32 7.53 25.26 2.49 0.28 0.87 0.06 1.42 22.69* (b) 0.16 0.01 0.006

' Significant difference ( p < 0.01) from the denoted samples (Student's t-test). * Significant difference ( p < 0.05) from the denoted samples (Student's t-test). < A mean value containing one or more values under the detection limit of the instrument.

At first sight it is conspicuous that the chemical characteristics of the soil solutions from the stemflow side of the trunks deviate systematically from those of the other sampling points. As shown in Tabs. 1 and 2, the pH of the soil solution of the downhill side of the stemflow area is - in accordance with data in the literature (e.g., Kazda and Glatzel, 1984; Glavac et al., 1985; Schulte and Spiteller, 1987; Falkengren-Grerup, 1989a) - in both stands significantly (p < 0.01) lower than that of the soil solutions collected of other sampling points. One of the trunks investigated in Parid showed a pH difference of 3.19 in the solutions sampled on its two sides (the distance between the two sampling points did not exceed 1 m). Similar to data in the literature (e.g., Kazda and Glatzel, 1984; Glatzel and Kazda, 1985; Glavac et al., 1986; Falkengren-Grerup, 1989a), pH differences in beech stands caused by stemflow influenced the composition of the soil solutions in our stands as well (Tabs. 1 and 2). In both stands the soil solutions at the downhill side of the stemflow area contained less Ca, Sr, Ba and Mg, and more K as well as Fe, than the solutions collected from other sampling sites. In some cases, the differences were highly significant ( p < 0.01). However, differences in the behavior of some elements (such as Mn and Zn) can also be detected between the two stands. In addition, at lower p H values in all soil solutions in Brennbergbanya the A1 and heavy metal contents were significantly increased compared to the soil solutions originating from

458

G. Turchj3r et u l

Parad. It seems that the acidification proccss could not have caused such significant changes in the chemical composition of the soil solutions in Brennbergbanya as in the soil solutions of the stand in Parad. Changes in the composition of the soil solutions in the stemflow area arc rcflccted by some alterations in the element accumulation by the beech roots as well (Tabs. 3 and 4). Thus, the Fe, Pb and V contents in Brennbergbanya, and the Pb content in Parad were significantly higher in the beech roots of the stemflow area than in those of all other areas. Additionally, in Parad a significant decrease in the Ca, Sr and Mg contents of the stemflow roots could have bccn dctcctcd, compared to other sampling points. A difference in the behavior of the elements between the two stands is also conspicuous. Some not consequent and thus not easily interpretable significant differences amongst samples not influenced by acidic stemflow show that, besides stemflow, the forest soil and its inhabitants are exposed to several other environmental factors as well.

Tab. 3. Element Contents (pg . g - ' ) of Beech Roots in the Stemflow and Throughfall Arcas of a Stand in Brennbergbinya, Hungary Valucs arc thc mcans of 4 parallels Element

PH Ca Sr Ba Li Na Mg K P Fe Al

B Mn Si Cd CO Zn CU Ni Ti Ph V +

*

Throughfall area, uphill

Stemflow area, downhill

Throughfall arca, downhill

(4

Stcmflow area, uphill (b)

(c)

(4

4.20 2638.25* (d) 18.14* (d) 77.70 0.67 92.31* (c) 476.40 1624.00 471.23* (d) 741.13 1497.50 3.14 237.75 12.82 0.60 1.16 42.22 5.40 3.70 27.67 5.54 I.57

4.18 2474.50* (d) 17.05 (d) 61.73 0.45 75.71* (c) 41 5.70* (c, d) 1494.25 416.85* (c, d ) 522.80 1 125.50 3.26 209.03 14.30 0.58 1.02 41.17 5.55 3.76 24.62 6.53 1.29

3.60' (a, b, d) 3 150.75 17.12' (d) 43.31 * (d) 0.98* (b) 160.83 633.43 2050.00 600.35 1510.25' (a, b,d) 1959.75' (b) 4.46 157.26 17.74 0.78 1.60 45.91 9.36 5.19 36.96* (b) 44.36' (a, b, d) 3.20' (a, b), * (d)

4.33 3662.75 25.73 100.8I 0.75 90.1 5* (c) 622.43 2076.00 623.35 763.23 1609.00 4.93 302.80* (c) 18.31 0.66 I .22 50.79 6.76 5.84 36.48* (b) 9.78 1.75

+

Significant difference ( p < 0.01) from the denoted samples (Student's (-test). Significant diffcrcncc ( p < 0.05) from thc dcnotcd samples (Student's I-test).

Sampling in Stelflow und Tliroughfall Areus

459

Tab. 4. Element Contents ( k g . g - ' ) of Beech Roots in the Stemflow and Throughfall Areas of a Stand in Parad, Hungary Values are the means of 4 parallels Element

PH Ca Sr Ba Li Na Mg K P Fe At B

Mn Si Cd co Zn

cu

Ni Ti Pb V +

*

Throughfall area, uphill

Stemflow area, downhill

Throughfall area, downhill

(4

Stemflow area, uphill (b)

ic)

(4

6.62 8229.0 42.13 306.50 1.24 71.30 997.35 3082.75 1122.60 1388.00 2451.88 5.27 250.50 8.97* (b) 0.79 1.53 54.84 6.48 13.57 80.03 3.12 4.18

6.53 7580.75 38.93 369.54 1.36 78.47 925.13 2692.25 89 I .98 1755.25 29 33.50 3.91 218.75 11.25 0.71 I .73 65.93 8.30 9.20 100.66 5.12 4.70

4.64' (a. b, d) 3618.50' (a, b, d) 16.54' (a, b, d) 234.49 1.14 7 1.43 672.40' (a, b, d) 2470.00 777.18* (a) 1833.25 2349.25 2.14' (a, d) 268.63 9.63 0.65 I .96 56.17 14.99* (a, b) 7.3 I 77.40* id) 55.37' (a, b, d) 4.59

6.64 7580.75 39.58 331.71 1.40 78.46 952.73 3074.50 1011.63 1812.75 3044.50 5.29 282.50 10.31 0.92 1.77 75.28 10.24 6.32 114.99 3.95 5.08

Significant difference ( p < 0.01) from the denoted samples (Student's t-test). Significant difference ( p < 0.05) from the denoted samples (Studcnt's t-test).

25.5 Conclusions Based on our and other investigations it is necessary to emphasize the divergence of the chemical and physical composition of the trunk base area from the remainder of forest stands in several woods of the Earth. Additionally, significant differences can be observed in the composition of samples originating from the throughfall areas of forests as well. The troubles of sampling in a forest are shown by the conclusions of Lloyd and Marques-Filho (1988), who established that the random relocation of throughfall gauges was better than fixed positions, and random relocation on a transect line better than over an area. The results of the study of Wolfe et al. (1987) indicate that increased sampling intensity combined with tracer studies will be needed to clearly determine the effect of stemflow, tree species, and tree size on subtly manifested soil properties such as pH and S 0 4 - S as well as other elements of interest. Duijsings et al. (1986)

460

G. Tiirc.sbnJ,ict cil

summarized the data on ion depositions in throughfall in a scheme which for a given dissolved constituent and a desired accuracy interval, indicates the number of sampling points needed. We think all data presented here support the opinion that before establishing element concentration cadasters in forest ecosystems preliminary investigations are needed to estimate the extent and the impacts of stemflow and throughfall as well as the number and size of samples necessary for getting reliable results. Acknowledgement. The work was financed by the National Scientific Research Funds (OTKA) of the Hungarian Academy of Sciences.

25.6 References Almeida, A. P. de, Rickerk, H., De-Almeida, A. P. (1990) Water balance of Eiicu1ypptu.s g1ohulu.s and Qiiercus suhrr forest stands in south Portugal. Forest Ecology und Muiiugement 38, 5 5 -64. Anderson, A. R., Pyatt, D. G. (1986) Interception of precipitation by pole-stage Sitka spruce and lodgepole piiic and mature Sitka spruce at Kieldcr Forest, Northumberland. Forestry 59.29 - 38. Andcrson, A. R., Pyatt. D. G., Stannard, J. P. (1990) The effects of clearfelling a Sitka spruce stand on the water balancc of a pcaty glcy soil at Kershope Forest, Cumbria.-Forr.~try O:y/i)rcl 63, 51-71. Andcrsson. T. (199 I)Influence of stemflow and throughfdl from common oak (Quercus robur) on soil chemistry and vegetation patterns. Cunadian .lournu/ of’ Forcst Rc>seurcli 21, 91 7 - 924. Asche, N., Becse, F. (1987) Schwcfeldioxid-Belastung und Sulfat-Deposition in einem emissionsnah gelcgcncn Waldiikosystem. AIlgcwic~ine2Forst-citsclirj~iNo. 34, 868 - 870. Baker, T. G., Attiwill, P. M. (1987) Fluxes of elements in rain passing through forest canopies in south-castcrn Australia. Biogroc,ke,mistry 4, 27 - 39. Cape, J. N., Brown, A. H. F.. Robertson, S. M. C., Howson, G., Paterson, I. S. (1991) Interspecies comparisons of through fall and stemflow ot three sites iii northern Britain. Fnrc~srEcology arid Mrmugcwicwt 46. I65 - 177. Clement, M., Wittig, R. (1987) Heavy metal content of the moss Miiium liornum growing in thc stem flow area of Fcigus sylcnticu. Actri Oec,ologicu, 0rw)Iogia Planturum 8, 257 - 264. Crabtrcc, R. W., Trudgill. S. T. (1985) Hillslope hydrochemistry and stream response on a wooded, permeable bedrock: the role of stemflow. Journal of’ Hyu‘rolngy 80, 161 - 178. Crockford, H., Richardson, D. P. (1987) Twknical Memorandirm No. 87/11, Division of Water Resources Research. Institute of Natural Resources and Environment, CSIRO, Australia. Crockford, R. H.. Richardson, D. P. (1990a) Partitioning of rainfall in a eucalypt forest and pinc plantation in southeastern Australia: I. Throughfall measurement in a eucalypt forest: effect of method and species composition. Hydrological Processes 4, I3 1 - 144. Crockford, R . H., Richardson, D. P. (1990b) Partitioning of rainfall in a eucalypt forest and pinc plantation in southeastern Australia: 11. Stemflow and factors affecting stemflow in a dry sclerophyll eucalypt forest atid B Pinus rcidicitu plantation. Hydrologicul Processes 4, 145- 155. Crockford, R . H.. Richardson, D. P. ( 1 9 9 0 ~ Partitioning ) of rainfall in a eucalypt forest and pine plantation in southeastern Austalia: IV. The rclationship of interception and canopy storage capacity. the interccption of these forests, and the effect on interception of thinning the pinc plantation. Hydrological Processes 4, 169 - 188. Crockford, H., Topalidis, S., Richardson, D. P. (199 I ) Water repcllcncy in a dry sclcrophyll eucalypt forest-measurements and processes. Hydrological Processes 5, 405 -420. Crozier, C. R., Boerncr, R. E. J. (1984) Correlations of understory herb distribution patterns with microhabitats under different tree species in a mixed rncsophytic forest. Oecologiu 62,337 - 343.

Sunipling in StemfloM’ and Throughfall Areus

46 1

Darnhofer, T., Gatama, D., Huxley, P., Akunda, E., Darnhofer, T. 0.. Huxley, P. A. (1989) in: Meteorology and agroforestry. Proceedings of’ an international workshop on the application of meteorology to agrqforestry systems planning and management, Nairobi 9- 13 February 1987, Reifsnyder, W. S., Darnhofer, T. 0. (eds.) Nairobi, Kenya: International Council for Research in Agroforestry (ICRAF), pp. 371 -382. Duijsings, J. J . H. M., Verstraten, J. M., Bouten, W. (1986) Spatial variability in nutrient deposition under an oak/beech canopy. Zeitschrift ,fur Pflanzenernahrung und Bodenkunde 149, 718 -727. Edmonds, R. L., Thomas, T. B., Rhodes, J. J. (1991) Canopy and soil modification of precipitation chemistry in a temperate rain forest. Soil Science Society qf America Journal 55, 1685 - 1693. Enright, N. J. (1987) Stemflow as a nutrient source for nikau palm (Rhopalost.vlis sapirla) in a New Zealand forest. Austruliun Journal of Ecology 12, 17 - 24. Escudero, A. (1985) Efectos de arboles aislados sobre las propiedades quimicas del suelo. Revue d’ Ecologie et de Biologie du Sol 22, 149 - 159. Ettala, M. (1988) Evapotranspiration from a Salix aquatica plantation at a sanitary landfill. Ayuu Fennicu 18, 3 - 14. Falkengren-Grerup, U. (1989a) Effect of stemflow on beech forest soils and vegetation in southern Sweden. Journal of Applied Ecology 26. 341 - 352. Falkengren-Grerup, U. ( I 989b) Soil acidification and its impact on ground vegetation. Amhio 18, 179-183. Falkengren-Grerup, U., Bjork, L. (1991) Reversibility of stemflow-induced soil acidification in Swedish beech forest. Environmental Pollution 74, 3 1 -37. Farmer, A. M., Bates, J. W., Bell, J. N. B. (1991) Seasonal variations in acidic pollutant inputs and their effects on the chemistry of stemflow, bark and epiphyte tissues in three oak woodlands in N.W. Britain. Neu, Phytologist 118, 441 -451. Fliickiger, W., Braun, S., Fliickiger-Keller, H., Leonardi, S., Asche, N., Buhler, U., Lier, M. (1986) Untersuchungen iiber Waldschiiden in festen Buchenbeobachtungsflichen der Kantone BaselLandschaft, Basel-Stadt, Aargau, Solothurn, Bern, Zurich und Zug. Schweizerische Zeitschrift ,Jir Fur.stwe.sen 137, 917- 1010. Forster, H., Schimmack, W., Kreutzer, K. E. (1991) Die horizontale Verteilung von Radiociisium im Waldboden unter Fichte und Buche. Zeitschrift ,fur Pflanzenernuhrung und Bodenkunde 154, 87-92. Garcia-Bellido, I., Garcia-Criado, A , , Escudero-Berian, A., Gomez-Gutierrez, J. M. (1989) Relacion entre volumen de agua y concentracion de bioelementos en la precipitacion y agua de lavado de Quercus rotundifolia Lam. y Quercus pyrenaica Willd. en ecosistemas de dehesa. Studiu Oecologica 6, 265 - 291. Glatzel, G., Kazda, M. (1985) Wachstum und Mineralstoffernihrung von Buche (Fagus sq’lvaticu) und Spitzahorn (Acer platanoides) auf versauertem und schwermetallbelastetem Bodenmaterial aus dem Einsickerungsbereich von StammabfluRwasser in Buchenwaldern. Zeitschrifi ,fzir Pflunzenernahrung und Bodenkunde 148, 429 -438. Glatzel, G., Sonderegger, E., Kazda, M., Puxbaum, H . (1983) Bodenverinderungen durch schadstoffangereichte Stammablaufniederschlage in Buchenbestanden des Wienerwaldes. ANgemeine Forstzeitschrift No. 26/27, 693 - 694. Glavac, V., Jochheim, M., Koenies, H., Rheinstadter, R., Schiifer, H. (1985) EinfluIj des Stammablaufwassers auf den Boden im StammfuIjbereich von Altbuchen in unterschiedlich immissionsbelasteten Gebieten. Allgenieine Forstzeitschrift No. 5 1 - 52, 1397 - 1398. Grenzius, R. (1988) Starke Bodenversiiuerung und Schwermetallanreicherung durch StammabfluIj in der Innenstadt von Berlin (West). Mitteilungen der Deutschen Bodenkundlichen Gesellschaft 56, 363 - 368. Guillittc, O., Brabant, B. de, Gasia, M . C., De Brabant, B. (1990) in: Bryology and lichenology in Belgium. A symposium celebruting the 10th anniversary of the Vlaamse Werkgroep Bryologie. Meise, 26 November 1988: Memoires de la Societk Royale de Botanique de Belgique, No. 12, pp. 89 - 99. Hajduk, J. (1987) Imisiami ovplyvneny obsah niektorych prvkov v podc v chranenej krajinnej oblasti Muranska planina. Ochranu Prirody 8, 123- 131.

462

G. Turcxinj,i c’t u l

Hartung, J. ( 1985) in : Proceedings, V. Internutionrrli~rKongrc$,fiir Tierliygim,, Hunttover, 10- 1 3 S(prwibrr 198.5, Brriid 11: Hilligcr, H. G. (ed.) GicRcn. German Federal Repuhlic; Deutsche Veterinirmedizinische Gesellschaft, pp. 659 - 664. Hertz, J., Angehrn-Bettinazzi, C. ( I 989) Ungcnauigkeit von Schwermetallanalysen - ein Homogenitiitsproblem. Bulletin - Bodenkundlicke Ge.sellscliuft der Scliweiz 13, 81 -92. Herwitz, S . R. (1986a) Episodic stemflow inputs of magnesium and potassium to a tropical forest floor during heavy rainfall cvcnts. Oecologiu 70, 423 -425. Herwitz, S. R. (1986b) Infiltration-excess caused by stemflow in a cyclone-prone tropical rainforest. Eurth Surfirce Processes und Lun&)rins 11. 401 -41 2. Herwitz, S. R. ( 1987) Raindrop impact and water flow on the vegetative surfaces of trees and the effects on stemflow and throughfall generation. Eurtli .Sur;/Ucc Proc~~sses utid Lmdfnrnis 12, 425 -432. Hubcr, A. W., Oyarzun, C. E. (1990) Variaciones annuales en precipitacion, escurrimiento c intercepcion cn un bosque adulto dc Pinus rudiuta. Turriulba 40, 503- 508. Huttermann, A , , Gchrmann, J . ( I 982) Auswirkungen von Luftverunreinigungen auf eine Buchcnnaturverjungung in immissionsexponierter Lage. Forst- und Holzwirt 37, 406-410. Ivcns, W. P. M. F., Draaijcrs, G. P. J., Bleuten, W., Bos, M. M. (1989) The impact of air-bornc ammonia from agricultural sources on fluxes ol‘ nitrogen and sulphur towards forest soils. Crrterrtr 16, 535-544. Johnson, R. C. (1990) The interception, throughfall and stemflow in a forest in Highland Scotland and the comparison with other upland forests in the U.K. Journal ofHydro1ug.y Amsterdun? 118. 281 -287. Jordan, (’. F., I leuveldop, J. (1981) The water budget of an Amazonian rain forest. Actu Atnazonic,u 11, 87-92. Kazda, M.. Glatzel, G . (1984) Schwermctallanreicherung und Schwermetallverfugbarkeit im Einsickerungsbereich von Stammablaufwasser in Buchenwaldern (Fugus syluuticu) des Wienerwaldes. Zritsclirift ,/%PflritizPvrerriiilrrung irnd Botkenkuntle 147, 743 - 752. Kelly, J . M. ( I984n) in: Forest soils rind treutmeni inipucts, Proceedings, Sixth N o r t h Atnwicun b’or~~.vt S0il.v C‘ot7fiwticr, Unioer.sit)~ nf’Tennessee, Knoxville, June 1983, Stone, E. L. (ed.) Knoxvillc, USA: Department of Forestry, Wildlifc and Fisheries, University of Tennessee, pp. 265 -289. Kelly, J . M . (1984b) Power plant influences on bulk precipitation, throughfall, and stemflow nutrient inputs. Journul of’ Enoironnientrrl Quulity 13, 405 -409. Kinzel, H., Baumgartcn. A,, Spadingcr, K., Zcchmcistcr-Boltenstern, S. (1989) in: Ecologicd itripuct of‘ucidification,Proccdings of the Joini Syinpo.sium Environnioniul T1irrwt.s to Forest and O t h ~ r Natirrol Ecosjwetris” held L I ~the Universitj. of Oulu, Finlurid, November, I - 4, 1988, Szabolcs, T. (ed.) Budapest, Hungary: Hungarian Academy of Sciences, pp. 69-77. Klemniedson, J. 0 . . Mcier, C. E., Campbell, R . E.. Marx. I). B. (1983) Effect of stand composition and season on chemistry of throughfall iind stemflow o f ponderosa pine forests. Forest Sciencc 29, 871 -887. Kopeszki, H. (1988) Populationsdynamik und Indikatorwert der Boden-Mesofauna iin Einflunbereich dcs saurcn Buchen-Staminablaufcs. Zoologischer Anzoiger 221, 368 - 378. Kopcszki. H. ( 1992) Vcrindcrungen dcr Mcsofauna eines Buchenwaldes bei Siurebelastung. P ~ d ~ ~ h36,i 295 ~ l-~305. ~ ~ i ~ Koprivnjak, J . F., Moore. T. R . (1992) Sources, sinks, and fluxes of dissolved organic carbon in subarctic fen catchments. Arctic arid Alpine R m w r c h 24. 204- 21 0. Leonardi, S.. Fluckiger, W . (1987) Short-term canopy interactions of beech trees: mineral ion leaching and absorption during rainfall. Tree Physiology 3, 137- 145. Lloyd. C. R., Marques-Filho, A. de 0. (1988) Spatial variability of throughfall and stemflow measurements in Amazonian rainforest. Agricultural and Forest Meteorology 42, 63 - 73. Mahendrappa, M. K . (1987) Tree species and urea treatment effects on sulfur and metals in throughfall and stemflow of some eastern Canadian forest stands. Cunndim Journul of’ Forest Re.S<~ar(,li 17, 1035 - 1042. Mahendrappa, M. K. (1989) Impacts of forests on water chemistry. Wuter, Air, rind Soil Pollution 46. 61-72, Marschner, B., Radei, K ., Renger, M. ( I 99 I) Riiumliche Variabilitit von Bodenlosungszusammensctzung und Sickerwasseraustrag unter einem Kiefern-Altbestand. Mitteilungrn cfer Deutschen Borlenkuridliclien Ge,vdl.scliuft 66, 359 362. ‘ I

-

Sampling in Stemflow and Throughfull Areas

463

May, F. E., Ash, J. E. (1990) An assessment of the allelopathic potential of Eucalyptus. Austruliun Journal of Botuny 38, 245 - 254. Miller, H. G. (1984) Deposition-plant-soil interactions. Philosophical Transactions of’ the Rojzl Society of London, B 305, 339 - 35 1. Moore, T. R., Dubreuil, M. A. (1987) The neutralization of acid precipitation by beech and maplc stands in southern Quebec. Nuturuliste Cunadien 114, 449 -457. Mugnozza, G. S., Valentini, R., Spinelli, R., Giordano, E., Scarascia-Mugnozza, G . (1988) Osservazioni sul ciclo dell’acqua in un bosco ceduo di Qziercus cerris L. Annuli Accudemia Italiana di Scienze Forestali 37, 3 - 2 I. Navar, J., Bryan, R. (1990) lntcrception loss and rainfall rcdistribution by three semi-arid growing shrubs in northeastern Mexico. Journal of’ H ~ ~ d r o l o gAmsterdam y 115, 51 -63. Neal, C., Jeffery, H. A,, Conway, T., Ryland, G. P., Smith, C. J., Neal, M., Norton, S. A. (1992) Beryllium concentrations in rainfall, stemflow, throughfall, mist and stream waters for an upland acidified area of mid-Wales. Journal qf Hydrology, Amsterdam 136, 33 -49. Neite, H. (1989) Zum EinfluB von pH und organischem Kohlenstoffgehalt auf die Loslichkeit von Eisen, Blei, Mangan und Zink in Waldboden. Zeitschrifi,fiir Pflanzenernahrung und Bodmikunde 152, 441 -445. Neite, H., Wittig, R. (1985) Korrelation chemischer Bodenfaktoren mit der floristischen Zusammensetzung der Krautschicht im StammfuBbereich von Buchen. Acta Oecologia, Oecologia Plantarurn 6, 375-385. Nizinski, J.. Saugier. B. (1989) Dynamique de I’eau dans une chenaie ( Q u u c u s petraeu (Matt.) Liebl.) en forst de Fontaineblcau. Annulrs des Sciences ForestiPres 46. 173- 186. Offenberg, K. (1986) Bodenversauerung im Wurzelbereich einer Altbuche. Forst- irnd Holzwirt 41, 295 - 297. Oh, J. H., Kim, Y. K., Chae. J. S., Yi, C. K . (1987) Studies on the changes of pH and chemical composition of throughfall and stemflow under coniferous, deciduous forest stands due to acid precipitation. Research Reports of the Forestry RiJsearch Institute Seoul No. 34, 159 - 165 (in Korean). Oyarzun, C . E., Huber, A. W., Vasquez, S. G . (1985) Balance hidrico en tres plantaciones de Pinus radiata. I. Redistribucion de las precipitaciones. Bosyue 6, 3 - 14. Pallant, E., Riha, S. J. (1990) Surface soil acidification under red pine and Norway spruce. Soil Science Society qf Americu Journal 54, 1 124 - 1 I 30. Papritz, A. (1987) Veranderungen der Bodeneigenschaften im StammfuBbereich von WaldbHumen. Schweizerische Zeitschrift ,fiir Forstwesen 138, 945 - 962. Pontailler, J. Y., Nizinski, J., Saugier, B. (1988) in: Studies on water transport in the soil-plantutnm~pheresystem, Calvet, R. (ed.) Paris: Institut National de la Recherche Agronomique (INRA), pp. 329-355. Potter, C. S. (1992) Stemflow nutrient inputs to soil in a successional hardwood forest. Plant and Soil 140, 249 -254. Prebble, R. E. (1987) Divisional Report No. 90, Division of Soils, CSIRO, Australia. Price, A. G., Watters, R. J. (1989) The influence of the overstory, understory and upper soil horizons on the fluxes of some ions i n a mixed deciduous forest. Journal qf’Hydrology,Netherlands 109, 185- 197. Pudjiharta, A,, Basuki, T. M. (1988) Intersepsi curah hujan pada tegakan Agathis loranthifi)lia. Buletin Penelitian Hutan, Bogor, Indonesia No. 5 12, 1 - 10. Raich, J. W. (1983) Throughfall and stem flow in mature and year-old wet tropical forest. Tropical E C O I O24, ~ J 234~ 243. Ranipazzo, N . (1991) Doctoral Thesis, University of Vienna. Rampazzo, N., Blum, W. E. H. (1990) Chemico-mineralogical changes in forest soils due to atmospheric deposition. Mitteilungen cler Deutschen Bodenkundlichen Gesellschaf’t 62, 133 - 136. Rampazzo, N., Blum, W. E. H. (1992) Changes in chemistry and mineralogy of forest soils by acid rain. Water, Air, and Soil Pollution 61, 209 -220. Santa-Regina, I., Gallardo, J. F. (1989) Biogeochemical cycles: return of bioelements in rainfall. Acta Oecologica, Oecologiu Pluntnrum 10, 433 -438. Santa-Regina, L., Gallardo, J . F., San-Miguel, C., Moyano, A. (1989) Intercepcion, pluviolavado y escorrentia cortical en una plantacion de Pinus syluestris de la Cuenca de Candelario (ccntro-oeste dc Espana). Bosyuc~10, 19-27.

464

G . Turcsanyi

CI a1

Santos, A. dos, Ribeiro, M. de N . G., Ribeiro, T. S. B., Bringel, S. R. B. (1981) Hidroquimica da Amazonia Central 111. Quimica da agua dc lavagem da florcsta no ecossistema Campina Amazonica (stemflow). Acts Aniuzonicu 11, 335- 346. Satoh. F.. Fujiwara, K., Komazaki, S. (1989) Chemical compositions of throughfall and stem flow under mixed forcsts: an cxamplc i n Misumai experimental site. Reserrrch Bulli~tinso f t h e College Erpcvimcnt Forosts, Hokknido Uniwrsiry 46, 829 -846 (in Japanese). Schierl, R., Gottlein, A,, Ilohmann, E., Trubenbach, D., Krcutzer, K. (1986) EinfluB von saurer Beregnung und Kalkung auf llumusstoffc sowic die Aluminium- und Schwermetalldynarnik in w,"[{.' . I igen Bodencxtrakten. t;r,~,vtii~i.r,serzsclz~~l~c/ie,s ~i,nirulb~utt 105, 309 - 3 13. Scholten, T. (1990) Vcrgleich der Humusformen in zwei Buchenbcsthndcn untcr dcm EinfluB hohcr atmosphirischer Sloffeintriige. Mittcilungcw cler Diwt.rihan Bodcwkiindlichcn Gesrll.schqft 62, 175- 182. Schulte. A , , Spiteller, M. (I987) Veriinderungen bodenchemischer Parameter im Stammablaufbereich von Buchen. Forst- und Holzii,irt 42. I52 - 154. Spelsburg, G., Crossmann, G. ( 1 984) Beobachtungen iiber die Bodenversauerung im StammfuBbereich zweier Altbuchen. Forst- iind Holzivirr 39, 286 - 288. Stockli, H. (1991) Influence of stemllow upon the decomposing sysiem in two beech stands. Revue, cl'Gologir et de BiologitJ dir Sol 28, 265 - 286. Switzer. G . L., Nelson, L. E., Shelton, M. G . (1988) Influence o f the canopy o f loblolly pine plantations on the disposition and chemistry of precipitation. Technical Bulletin - Mi.ssissippi Agricultural und Forestrv E.uperinient Station No. 154, 1 - 30. Tajchman, S. J . , Kcys, R. N., Kosuri, S. R. (1991) Comparison of pH, sulfate and nitrate in throughfall and stemflow in yellow poplar and oak stands in north-central West Virginia. Fr,rr.st Ecology crrid Muncrge~ue~it 40, I37 - 144. Thurow, T. L., Blackburn, W. H., Warren, S. D., Taylor, C. A., Jr. (1987) Rainfall interception by midgrass, shortgrass, and live oak mottes. Journal q f Range Management 40, 455-460. Turcsinyi, (3..Fangmcicr, A. ( 1 990) Blci- und Cadmiumgchalt von Buchenwurzeln (Fagu L.) in StammfuB- und Zwischcnstammbercichcn. Zeitschrift f i r Pflanzmerniihrung trnd Boderik u ~ i ~153, l t 197-200. Wattcrs, R . J., Price, A. G. (1988) The influence of stemflow from standing dead trees on the fluxes of some ions in a mixed deciduous forest. Cuntrdiun Journalqf'Forest Research 18, 1490 - 1493. Wcihc. J. (1985) Benetzung und Interzcption von Buchcn- und Fichtcnbcstinden. V. Die Verteilung dcs Rcgcns untcr Buchenkroncn. Allgenieine Forsr- wid Jugcl'zeirunR 156, 8 I - 89. Werner, W. (1988) Stickstoff- und Phosphor-Mineralisation im Versickerungsbcreich dcs Stammablaufwassers von Buchen (Fugus .syluuficcc L.). Hora, G D R 181, 339-352. Werner, W., Venanzoni, K., Wittig, R. (1987) Trunk base phenomena in Italian beech forests. A comparison with Central European conditions. A c t ~Uccologicxz, i Ucwhgiu Plunturztm 8, 359 - 374. Wilmanns, 0. (1985) Vegetation as an indicator of climate. General considerations and specific examples. .I. Riornc,tc,orology 10 (Suppl. 2). 86-95. Wittig, R . (l986a) Acidification phenomena in beech (I;ugu,i .q~lvuricu)forcsts of Europe. Water, Air, and Soil Pollution 31, 317-323. Wittig, R. (1 986 b) Belrstrmg und Scliiiden at~f'Okos~sreriichpnc. und ilirc Folgen, QuPr.seIinitt.s,sCtiiinrrr, 25. itnd 26. Noi~ernher1985, ~ r i c ~ i . s t a g ~ ~Rc~rIin: g ~ ~ ~Umweltbundesamt. ~ud~~ Wolfc, M. H., Kclly, J . M., Wolt, J . D. (1987) Soil pH and extractable sulfate-sulfur distribution as influenced by tree species and distance from the stem. Soil Scirnw Society qf Americu Journul 51, 1042- 1046. Yadav. A. K., Mishra, G. 1'. (1985a) Chemistry of stemflow and throughfall waters for some tropical dry deciduous forest trees I . pH, specific conductivity, nitrogen and phosphorus. Jourrztrl of' T r ~ / ~ iF~oa~lP . s ~I ,~51 J ' -60. Yadav, A. K., Mishra, G. P. (1985b) Chemistry of stemllow and throughfall waters for some tropical dry deciduous forcst trccs 11. Potassium, calcium and sodium. Journul of' Tropicul Forestry 1, 99- 1 1 I . Yadav, A. K., Mishra, G. P. ( 1 9 8 5 ~ )Distribution of precipitation under a tropical dry deciduous forest stand of central India. Journrrl of Tropicul Forestr~,1, 182- 197. Zcchmeister-Boltenstern, S. (1989) Doctorul Thesis, University of Vienna. Zhao, Z. (1989) Doctorul Tlicjsis, Univcrsity of Vienna.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

26 Sampling of Different Social Categories of Red Wood Ants (Formica s. str.) for Biomonitoring Vambolu Maavara, Ants-Johannrs Martin, Ahto Oja, and Pckku Nuorteva

26.1 The Role of Ants in Nature Red wood ants are one of the dominant elements in boreal forests. Their biomass is usually high and may under optimal conditions exceed that of all other invertebrates (Holldobler and Wilson, 1990). They have reached this dominant position in many terrestrial ecosystems, because during evolution the ants (especially the species belonging to the Formica s. str. group) have taken advantage of their genesis of a social way of life. It is worth remembering that the evolutionary development of social life in insects was a really great innovation, comparable to the development of multicellular organisms from unicellular ones through colony formation. In the development of the instinct societies of insects, multicellular organisms group together in order to achieve ecological advantages for single specimens. In environmental monitoring programs this kind of superorganisms requires special attention. This requirement has been considered, when the red wood ants have been listed for protection in the European red book of animals. In chemical biomonitoring, however, the social insects have received only minimal attention. Certainly, it is not wise to allow organisms with high biomass and great ecological potency to be neglected in biomonitoring. It is a bit astonishing to note that red wood ants, with their exceptionally high biomass, consume mainly the honeydew produced by aphids. More than 90% of their food consists of honeydew (Maavara and Martin, 1983; Rosengren and Sundstrom, 1987, 1991; Gosswald, 1983). In the Estonian environment one medium sized colony of Formica aquilonica collects more than 500 kg of aphid honeydew during the vegetation period, containing some 25-35 kg of sugar (Maavara and Martin, 1983). Only about 5 - 10% of the fresh weight of the food of red wood ants consists of prey animals, more if the amount is expressed in dry weight (Rosengren and Sundstrom, 1991). Despite the low percentage of animal food, red wood ants are important predators in the forest ecosystem because their number and biomass are so high. A colony of red wood ants studied by Rosengren and Sundstrom (1991) collected 5 million arthropod prey items with a fresh weight of 12 kg and a dry weight of 7 kg during one season. On the other hand, this same colony collected 300,000 seeds weighing 1.8 kg, and 240 kg of honeydew containing 36 kg of sugar. As predators ants are of great importance in the population regulation of other insects, including many forest pests, especially during mass outbreaks of the pests (Gosswald, 1952, 1979, 1989; Elton, 1958; Adlung, 1966; Otto, 1967; Finnegan,

466

v.MuaccliN C t al.

1975; Laine and Niemeli, 1980; Skinner and Whittaker, 1981; Warrington and Whittaker, 1985; Whittaker and Warrington, 1985; Rosengren and Sundstrom, 1987, 1991; Whittaker, 1991; Woodman and Price, 1992). Red wood ants are unusual predators in that their foraging activity and interaction with potcntial prey often seems to be more a consequence of their requirement for honeydew than a response to prey distribution and abundance (Whittaker, 1991). This is to say that much of the ants’ predation occurs in order to protect their honeydew trees against damage caused by herbivores. The case described by Laine and Niemeli (1980) is very illustrative in this respect. They showed that during a complete defoliation of subarctic birches by the geometrid moth Oporiniu autunznuta, there existed green islands of undamaged trees up to a radius of 2 0 m from the nests of F. aquilonia. By predation on grazing caterpillars the ants were able to protect the honeydew production in their foraging territory. The principle of density-dependent mutualism in an aphid-ant interaction is similar to that in the interaction in predation: Ant tending significantly improves the growth of small aphid populations, but the benefit from ant tending decreases or disappears at higher ant densities (Breton and Addicott, 1992). By high aphid density, when aphids are providing honeydew in excess, the ants start to prey on such aphid specimens, which have stopped feeding and are walking away from the group (Rosengren and Sundstrom, 1991). Out of the 5,000,000 arthropod prey items (weighing 12 kg), collected by foragers of a red wood ant colony in Finland, 2,000,000 were aphids (weight 0.8 kg) (Rosengren and Sundstrom, 1991). As predators upon herbivorous insects, ants are important food competitors for insectivorous birds (Catzeflis, 1979; Haeming, 1992). On the other hand, red wood ants may increase the population density of birds whose diet includes red wood ants (Otto, 1962). It is noteworthy that red wood ants are frequently predators also of evertebrates which are not herbivores. They also consume carcasses of dead animals. Through their nest-building, red wood ants have considerable influence on soil biology. They remove detritus and mix it into mineral soil (Gosswald, 1989; Holldobler and Wilson, 1990; Stanley et al., 1991). Sometimes they transfer detritus and mineral particles onto bare rock, thus giving the possibility for forest trees to take root there (Oinonen, 1956; Brandt, 1980a; Rosengren et al., 1985).

26.2 Pollutant Accumulation in Ants In respect to environmental conservation it is important to note that some functional groups of red wood ant workers are loaded with a content of metal among highest found in boreal forest animals (Fangmeier and Steubing, 1986; Stary and Kubiznakova, 1987; YIi-Mononen et al., 1989; Nuorteva, 1990; Nuorteva et al., 1992; Oja, 1992). In addition, lindane, dieldrin, DDT, DDE and PCB have also been noted in red wood ants (Debouge et al., 1987; Dcbouge and Thome, 1988, 1989). The primary source of metals for ants lies in detritus. From there a fraction of metals goes into the phloem of trees and further to the phloem-feeding aphids (Nuorteva, 1990). The aphids absorb mainly amino acids from their intestines,

Sumpling of Red Wood Ants

467

whereas sugars pass intact through their intestine to be excreted as honeydew. Metals from the phloem also go into honeydew (Crawford et al., 1983; Stary and Kubiznakova, 1987; Nuorteva, 1990) (Fig. I). Honeydew being the most important food for red wood ants, they get not only its sugars, but also high amounts of metals. Anthropogenic metal pollution as well as experimental dietary metal contamination can elevate metal levels in ants (Nuorteva et al., 1978b; Stary and KubiznQkova, 1987; Crawford et al., 1983; Nuorteva, 1988, 1990; Nuorteva et al., 1992; Oja, 1992).

CIRCULATION OF Cd IN FOREST BIOCOENOSIS

Fig. 1. Circulation of Cd in a forest biocoenosis (Nuorteva, 1990).

468

v.Muuuurtr e r NI.

Background metal levels in ants seem not to harm them. It has, however, been shown for Finnish Formica aquiloniu colonies that artificial cadmium feeding in the forest causes in both the surface workers and pupae detectable metabolic disorders. Such disorders include inhibition of metalloenzymes, induction of detoxifying enzymes, depletion of the adenine nucleotide pool and loss of homeostatic control by the reduction of the adenylate energy charge at least in cases where the cadmium content is elevated to a level occurring in polluted central European areas (Migula et al., 1993). The observation that metals from anthropogenic sources may accuinulate in ants to harmful levels is environmentally important and serves as a warning: One of the most dominant elements in forest ecosystems is in danger. It is consequently very necessary to monitor the fate and effects of metals in ant colonies, since the red wood ants is an Achilles’ heel for metals in the forest ecosystem (Nuorteva, 1990; Nuorteva et al., 1992). This kind of monitoring is, however, difficult because ants are superorganisms whose colonies consist of different kinds of functional groups of workers. Although morphologically similar, they perform different tasks in order to support the function of the colony. Quite like man in different occupations, ants too, are in their different works exposed to pollutants to different degrees. One can say that environmentalists have the job of studying occupational medicine in red wood ant colonies. To understand the fate and importance of the metals in the ant colony, it is necessary to take separate samples from each of the different castes, functional groups of workers and development stages. This kind of sampling requires knowledge of the function of the ant colony. First, the red wood ants are described (Sect. 26.3) and the structure and biological significance of their nests (Sect. 26.4) as well as the social structure and life cycle of the colonies (Sect. 26.5). From this theoretical basis we can furnish practical guidelines for collection of ant samples and of necessary background data (Sect. 26.6). We also describe the use of artificial feeding experiments (Sect. 26.7) and finally present pilot studies which elucidate the pilot work performed and the perspectives they offer (Sect. 26.8). The guidelines are based on a synthesis of the myrmecological research skill of Vambola Maavara and Ants-Johannes Martin, to the metal biomonitoring research skill of Pekka Nuorteva and Ahto Oja.

26.3 Definition of the Red Wood Ants In the genus Formica includes the subgroup of red wood ants. The subgroup comprises the following species: (1) Forniiru rufu L. (2) F. polyctenu Forst. (3) F. aquiloniri Yarr. (4) F. luguhris Zett. ( 5 ) F. pratensis Retz. ( 6 ) F. nisFricms Em.

Sumpling of Red Wood Ants

469

These all live in the forest and build mound nests. According to some authors the red wood ants include only the two (Otto, 1962) or four (Dlussky, 1967) first named species. Others add F. truncorum F. to the whole group as a seventh species (Gosswald, 1989). Red wood ants occur in the forested areas of the Palearctic region, ranging froin the Atlantic to the Pacific. The highest nest densities have been reported from mixed woods, where the seasonal differences in honeydew production from different species of trees gives the most favorable honeydew-access for these ants (Miiller, 1960; Rosengren et al., 1979; Laine and Niemela, 1989). Of the red wood ants, F. uquiionia and F. lziguhris are generally seen as the most dependent on conifers, but in the subarctic they extend their range to pure mountain birch forests (Laine and Niemeli, 1989). From the subarctic their range extends to a latitude of 56-57' in the south. F. rufa and F. polyctena occur mainly in the zone of mixed forests, whereas F. prcltensis and F. nigricans have their highest population densities in mixed forest and steppe zones. F. truncorum is scattered in its occurrence, living only in very xeric habitats (Stary, 1988).

26.4 Structure of the Nests and Territories In general, the red wood ants build their nests out of coniferous needles, twigs and resin, but F. prutensis sometimes prefers to build its nests of sand and fine gravel. The three first mentioned species often build rather high nest mounds, up to 2 meters. The nests are built instinctively to create a suitable microclimate for all developmental stages and for hibernation (Heimann, 1963a, b, 1964; Kneitz, 1970; Martin, 1975, 1980a, b, 1987; Rosengren et al., 1987; Gosswald, 1989; Horstmann, 1990). In addition, the nest gives protection against many kinds of enemies. The nest mound of red wood ants consists of three parts: (1) mound, (2) base and (3) underground part (Fig. 2). In addition, each nest colony has a foraging territory, often also sub- and superstructures.

26.4.1 The Nest Mound The mound consists of a surface layer built up mainly of needles fixed with melted resin and a more spongy nest interior constructed of twigs of different sizes and containing nest chambers. The mound surface is penetrated by holes and passages. Microbial activity in the nest material produces considerable amounts of heat through aerial metabolism, in addition to the mass-specific heat production of ants (Coenen-Stass et al., 1980). The nest warming effect of insolation and microbial activity is considered more important in small or weak colonies, whereas endogenous nest heating, based on the metabolism of the ants and their clustering behavior is more compatible in the case of vigorous colonies (Rosengren et al., 1987). The

470

V. Murivum et a/.

Fig. 2. The most important sampling sites of red wood ants during summer. ( I ) Ant road, (2) surface of nest top, (3A) brood chamber for pupae, (3B) brood chamber for larvae. (4) hibernating chambers for queens and reserve ants.

ventilation of the mound is effective: During cold and windy days the windward nest intcrior cools down, and the brood is carried to the opposite side of the mound where it is warmer (Heimann, 1963b; Martin, 1980a, b ; Horstmann, 1990). Because this kind of removals occurs quite often, it is important to determine the brood location prior to sampling.

26.4.2 The Base of the Nest The base is formed of sand and soil dug out when underground chambers are created and forms a circular wall around the mound. The base wall is penetrated by a system of passages for housing the reserve ants (store ants) and for ventilation.

26.4.3 The Underground Part of the Nest In the underground part of the nest chambers exist for diapausing reserve ants, for queens, and for the whole colony in winter. The underground part usually has a larger volume than the part above groud.

Sampling of’ Red Wood Ants

47 1

The mound shape is hemiellipsoid or paraboloid in general. A young nest is more flattened and has the shape of a hemisphere, with its radius about equal to its height. It is possible to estimate the nest volume by using special mathematical formulae (Maavara, 1991).

26.4.4 The Foraging Territory The colony has a guarded and marked foraging territory for collecting food. This territory is defended against members of competing nests and against other species of ants. The size of the colony determines the size of the territory. A nest territory may reach a quarter of a hectare in size. The number of ant roads (routes) leaving from the nest and the ant activity on them indicates territory size and its occupation.

26.4.5 Super- and Substructures When the ant colony grows to dimensions where the territory cannot support its life any more, colony fission occurs. In fission, a part of workers depart with a queen from the mother nest colony and occupy a new nest outside the territory of the old nest. At first, a polydomous colony develops, a supercolony, where two or several nests are connected to each other by ant roads (Martin, 1978; Rosengren et al., 1985; Gosswald, 1989; Zakharov, 1991; Fortelius et al., 1993). Along them a balancing exchange of queens, nurses, inside workers, pupae and larvae occurs (Fig. 3). When the daughter colony has developed a productive territory for itself it may attain independence. In some cases, usually in early spring or in late summer, the independent nests may, however, temporarily join each other and form a supercolony for a while (Zakharov, 1974, 1991 ; Maavara and Martin, 1983a). One can speak of a dynamic federation of ant colonies (Fig. 3). In different habitats the ant colonies are consistently polydomous and have a high degree of internest mobility, because the habitat necessitates thermoregulatory migrations (Martin, 1980a, b; Rosengren et al., 1985). Red wood ant colonies have sometimes, in addition to long feeding or connecting routes, one or several small mound nests without the above described nest interior and underground chambers. These annexes have the function of a refuge for the ants during bad weather. Sometimes, with the support of favourable living conditions for the main colony, such a substructure nest may develop to a persistent daughter nest with brood chambers and underground chambers for hibernation.

412

V . Muciucira

el

rrl.

w

Daughter colony

o

Abendunednest Feeding buds

Supercolony bounday

,--------

Fig. 3. Federation of supercolonies mappcd in the ant protection area Akste in Polva county, Estonia (Maavara and Martin, 19X3a).

26.5 Social Structure of Ant Colonies In monitoring the danger to ant colonies caused by the load of pollution, it is necessary to identify the different castes, developmental stages and functional groups of workers. There are three castes in the colony: females, males and workers. The workers are divided into functional groups based on their age-determined behavior and tasks in the colony (Schmidt, 1974).

26.5.1 Sexuals In the colonies there always exist reproductive queens able to produce eggs. In monogynic nests thcre is only one queen, in oligogynous ncsts these are from two to several queens, and in polygynous nests several hundreds and sometimes even more than a thousand queens coexist (Schmidt, 1974; Gosswald, 1989). The queens appear only seldom on the nest surface. The best possibility to catch them is in early spring during the basking time (in German “Sonnung”). Later they are hiding in the deepest chambers, are timid and thus difficult to catch. Under Estonian

Sampling qf Red Wood Ants

413

conditions young winged sexuals (males and females) of the red wood ants emerge generally in May or June. The sexuals of F. prutensis and F. nigricans emerge in June and August. The sexuals leave the nest shortly after their emergence in order to perform a matching flight. The lifetime of males is brief; they die shortly after mating and do not return to the nest. Fertilized females may return to their old nest, go to some other nest, or in some cases a female can intrude on a Formica,fuscu nest and form a new colony as a social parasite. Samples of young females and males are easy to collect during the mating flight.

26.5.2 Workers Red wood ant colonies include a non-reproductive laboring caste of workers, which form the most numerous part of the colony. They are sterile females, and their number in a nest depends on species and varies between some thousands and 3 million. If a nest contains less than a hundred thousand workers, the colony is not independent and is only part of a polydomous supercolony (Sect. 26.4.5). Workers are reared up during the whole summer. There exist three emergence peaks of young workers (sometimes called as generations) in June, July and August. It is possible to identify the times of emergence peaks by means of empty cocoons carried to the surface of the nest base or outside the nest. The longevity of workers is about two years (Maavara and Martin, 1983b). It is possible to note mass death of old outside workers especially at the times of the emergence peaks of young workers. Dead specimens are carried to the “graveyards” in abondened nest parts (especially at the northern side) or to the periphery of the base chambers and passages, but most often into moss and herb cover quite far from the nest, often at the boundary of the colony’s territory. For sampling it is easiest to find graveyards on forest patches or in wheel ruts of abandoned forest roads. With increasing age, the hormonal regulation of workers determines their different tasks of the colony in a quite fixed order (age polyethism); i.e., each worker proceeds during its life through a series of professions: (1) reserve worker, (2) nurse, (3) inside worker, (4) outside worker (e.g., forager, builder). The stages of age polyethism run in only one direction. A newly emerged ant cannot change immediately into an outside worker, nor can an old worker change to a nurse, because its behavior is under strong age-bound hormonal regulation.

26.5.2.1 Reserve Workers Ants emerging from pupae are reserve ants, which initially have no tasks. They are fed very intensively by inside workers, their consumed food being transformed into glycogen and lipids which accumulate as store substances in their fat bodies (Schmidt, 1974). When they have enough of store substances they migrate to the

414

v.MuuvurLl ct ul.

A. In early spring

B. In summer mass basking

&

reserve ants

Fig. 4. Temperature ( ‘C)and relative humidity (RH%) in red wood ant nests in early spring and summer. The arrows show: (A) migration of rcactivated reserve ants from the hibernation chambers to surface basking (“Sonnung”) when the cold barrier (T ‘C in brackets) has disappeared; (H) migration of reserve ants from the brood chambers to thc hibcrnatioii chambers.

hibernation chambers. There, glycogen and other sugars are consumed gradually during the inactive period. The use of lipids is, in contrast, blocked in diapausing reserve ants as long as the temperature is low (Schmidt, 1969; Martin, 1980b; Martin et al., 1985; Maavara and Martin, 1983b). Most reserve ants are reactivated in early spring, but for a minority activation is delayed until the middle of July. The diapause may last up to one year. Reactivation of the reserve ants coincides with the rise of their temperature preferendum (Fig. 4). Guided by positive thermotaxis they climb to the surface of the nest, where they will find the highest temperature (Kneitz, 1970; Schmidt, 1969; Martin, 1980a, b, 1991).

26.5.2.2 Nurses When the reserve ants are basking on the warm nest surface, the decomposition of their store lipids swiches on and results in food secretion from their labial and postpharyngeal glands (Schmidt, 1969, 1974). Simultaneously, hormonal changes alter their behavior, they movc into the inside of the nest and attain the state of nurses. They offer their food secretes to the larvae as wcll as also to the egg producing queen. Lack of food secretion results to development of small specimens. During the transformation of the lipids to sugars, energy is released as heat, which elevates the temperature in the brood chambers Lo 26-31 “C (Fig. 4). This is necessary for the development of the offspring (Heimann, 1963a, b, 1964; Schmidt, 1969, 1974; CoenenStass et al., 1980; Brandt, 1980; Martin, 1980a, b, 1991; Rosengren et al., 1987).

26.5.2.3 Inside Workers When the production of food secretion ceases, the bchavior of nurses changes into that of inside workers. Their tasks inside the nest are varied: They participate in

Sumpling of Red Wood Ants

475

the reception of food from outside workers, they cut up the prey, mix it with honeydew and other fluids in their crops and distribute it from mouth to mouth to other ants (trophallaxis). In addition, some groups of inside workers participate in rebuilding and cleaning of the brood chambers, in cleaning and removal of the brood inside the nest cone (Otto, 1962; Gosswald, 1989; Holldobler and Wilson, 1990; Zakharov, 1991). With the aging of inside workers their poison glands become active, their agressiveness increases and they become positively phototactic. They have thus developed into effective defenders of the nest. In this state their thready ovarioles increase to normal size (Schmidt, 1974).

26.5.2.4 Outside Workers Later the growth of the ovarioles stops, and soon they degenerate rapidly. In this period of life the ants change to outside workers (Schmidt, 1974). Now they participate in nest building, foraging of food, collection of nest materials and removal of all kinds of colony litter. Most of the outside workers are engaged in the collection of aphid honeydew (Maavara and Martin, 1983b; Rosengren and Sundstrom, 1987, 1991; Sect. 26. I). The foragers have a stable specialization either in honeydew collection or in predation. Small fraction of the outdoor workers have the task of recruitment of other workers for special tasks (Rosengren, 1971, 1977). One fraction of the outside workers have the task of territory defense. Some are engaged in carrying behavior and they participate in the exchange of brood and materials between different colonies. The eldest and most experienced outside workers (observers) maintain the memory about the colony and its territory (Rosengren, 1971, 1977; Zakharov, 1980). In spring, they stake out the nest routes and colony territory in the previous pattern by means of pheromone marks (Rosengren, 1971, 1977; Rosengren and Fortelius, 1987).

26.6 Sampling Practical application on monitoring of the theoretical knowledge on ant colonies given above is reported in the following.

26.6.1 What Kind of Nest Mounds is Suitable for Sampling? If one wishes to know the level of pollution of red wood ants in any given locality, it is not necessary to study very many nests, because the variation in pollutant levels between different nests is relatively small (see, e.g., Tab. 3 for control nests No. 8- 1 I , with a variation of about 30%).

416

V . M a m m a ct LII.

For continuous monitoring and more detailed investigations it is reasonable to select some strong colonies, which can easily resist monthly sampling without being weakened. Usually one can find such nests in well-lighted, fertile coniferous forests. It is possible to identify the vitality of the nests by the existence of several wellpopulated routes and by continuous, although periodical, increase in mound height. If one wishes to test experimentally the effects of pollutants, it is necessary to choose colonies of the same size. Nest size, however, is not allways correlated with colony size, and the colony size is, of course more important. The best method to insure size similarity is to count the number of ant roads and to calculate the activity on them. In addition, one must check that the nests have no contacts with any other nests (Fig. 3 ) . If the study nest belongs to a polydomous supercolony, dilution of the applied pollutant will occur and may result in erroneous conclusions as to poison tolerance. For F. polycfma, F. aquikonicc and F. lziguhris it may be difficult to find an independent colony, but nests of these species are nice objects for studies on pollutant distribution in supercolonies (Nuorteva et al., 1992).

26.6.2 Sample Taking Samples of red wood ants can be collected from ant roads, from the surface of the nest mound, from the brood chambers and from the nest base (Fig. 2). From the ant roads as well as from the surface of the nest it is possible to collect material by inserting test tubes into the soil or into the nest surface up to their edges. To increase the effect of such pitfall trap tubes one can use ethylene glycol (YlkMononen et al., 1989; Vogel et al., 1988) or pretreat them with pyrethrin spray. With high ant density they serve, however, quite satisfactorily also without any additives. It is also possible to collect ants by letting them climb on a small sheet board which is then quickly transferred into a polyethene bag (method used by Karel Mondspiegel in order to get ants without nest material). Bare hands are also a quite good instrument for the collection of ants from roads or nest surfaces. Hands are also the best instrument for taking ants from the nest interior. It is possible to move the hand to the warmest part of the nest, where the brood is present, or to the coldest part where the diapausing ants have their residence (Fig. 4). It is also possible to feel where the needles and twigs give way to mineral soil. It is possible to minimize the damage caused by the thermotactic hand orientation towards thc warm brood by identifying its site with long thin thermometers or through the sond of an electrical thermometer. In all cases it is necessary to fill the cavities caused by sampling because they may destroy the system of thermoregulation in the nests (Heimann, 1964). To take a sample from the desired area, it is mere1 necessary to transfer a handful of nest material and ants to a plastic bag and kill the ants by cold. Three replicates from different sides of the nest are necessary for statistical verification of the analyses. It is possible to perform most analyses from samples con-

Sampling of Red Wood Ants

471

taining only 20 - 50 worker ants, but the statistical stability of the results increases with 100 or more ants per sample. As for queens, winged females and males, which are difficult to catch but are heavier, about ten specimens of each are enough. When the samples have been separated into different fractions of ants and nest material, we dry them in a thermostat at 40 "C for about 72 hours. Dried samples are labeled and stored in glass tubes or plastic envelopes until analyzed chemically.

26.6.3 Sampling Objects For very sophisticated studies it is theoretically possible to collect nearly twenty different kinds of ants. In addition, it is possible to collect nest material from the sampling sites. The nest material may contain ant feces and is therefore potentially useful in illustrating the circulation of pollutants in ant nests. Obviously it would also be of interest to study the levels of pollutants in relation to ant diseases. Pollutants may decrease the resistance of ants to different diseases. It is also possible that some diseases with unknown etiology (e.g., the labial gland disease; Rosengren, 1979; Elton, 1989, 1991) are directly caused by some pollutants. Studies on the fate of pollutants in red wood ant societies are, however, only at their beginning. The first task is to study the main pollutant streams in the main ecological groups. During our pilot work (Nuorteva, 1988, 1990; Yli-Mononen et al., 1989; Nuorteva et al., 1992; Oja, 1992; Migula et al., 1993) we have gradually widened our sampling program. In our present studies we are collecting the following types of samples (Fig. 2).

26.6.3.1 Foragers Leaving the Nest We collect three separate samples of foragers leaving the nest from three ant roads within a radius of two meters. These ants most likely belong to the nest and do not comprise incoming food and pollutants, but a pure sample of foragers.

26.6.3.2 Foragers Traveling to the Nest We collect three separate samples of foragers traveling to the nest from three different ant roads. When roads are coming from trees infested by aphids, we take specimens descending the tree. In the study of Stary and Kubiznakova (1987) this mode of specimen catchment produced evidence as to the role of aphid honeydew as a metal source for red wood ants.

26.6.3.3 Surface Workers from the Top of the Nest The surface workers on top of the nest comprise of builders, carriers, foragers and observers, and to some degree also older inside workers. Sometimes there may occur masses of basking reserve ants as well. It is easy to recognize basking behavior

478

V . Mtiavoro el al.

by the dense masses of ants. Basking behavior is a quite common phenomenon especially in weaker colonies. In our pilot studies we collected surface workers from the top of the nest in pitfall test tubes (7 cm long, 2 cm in diameter) buried into the top of the nest. In our present studies we take the samples by scraping ants and nest material (to a depth of about 1 cni) into a plastic bag by hand from the nest surface on the top.

26.6.3.4 Workers and Brood from the Brood Chamber Larvae, pupae and indoor workers from the warmed brood chamber can be reached when the site of the chamber has been detected through thermometers or by hand. Larvae live together with nurses and other inside workers in the lower part of the brood chamber. The pupae live in the warmest upper part together with numerous inside workers and newly emerged ants with faint pigmentation. It is possible to separate from the samples larvae, pupae, inside workers, newly emerged ants and nest material for chemical analyses. It is important to note that outside workers are attracted to the brood chambers during rain and chilly weather. Under such conditions it is consequently not possible to catch pure samples of inside workers from the brood chambers. Therefore, it is necessary to restrict the sample taking to warm and sunny days.

26.6.3.5 Reserve Ants Reserve anls (store ants) can be reached through penetration of the nest base wall into the mineral underground part of the nest (Fig. 2). Recently dug sand or soil at the base of the nest mound indicates the easiest place to dig the tunnel to the site of the reserve ants. They are mainly passive, negatively photoactic and show preference for low temperatures - i.e., they are shy and try to hide in the cool sand. This behavior differentiates them from older active and agressive workers, which inhabit the hibernation chambers in order to dig them larger and to clean them. Because it is possible to identify the reserve ants just on the basis of their behavior (Maavara and Martin, 1983; Martin, 1991), it is not wise to kill any ants in reserve ant samples before they have been separated from the older ants. As to absence of reserve ants, that can be considered as a bad omen for the future of a red wood ant colony (Sect. 26.8.3).

26.6.3.6 Nest Material Nest material is always automatically collected when samples are taken from the nests by hand. As controls for these samples it may be useful to collect mineral soil and forest detritus (needles and twigs) within a distance of 2 - 5 m from the nest. At present, we have no idea about the utility of studying nest material. The

Sampling of Red Wood A n t s

479

levels of lindane, dieldrine, D D T and PCB found by Debouge et al. (1987) in the nest material of several species of ants were not higher than those in the surrounding environment.

26.6.4 Collection of Background Data and the Order of Sampling Red wood ants are very sensitive to any disturbance. When disturbed they excrete an alarm pheromone that induces nest defense and leads to mixing of the different working groups (Holldobler and Wilson, 1990). This is harmful not only for the quality of the samples, but also for the estimation of the vitality of a given colony. I n our present work we have tried to minimize the disturbance by taking the samples and making observations in the following order.

26.6.4.1 Traffic Density of Foragers on Ant Roads Upon arrival at the nest we first number the ant roads clockwise. Then we count, at a distance of 2 m from the nest, the traffic density of foragers per minute in both directions. The total number of ants counted is a significant ecological indicator that reflects very well the vitality of the nest. At the same time we register the weather conditions (air temperature, sunshine, clouds).

26.6.4.2 Ant Activity on the Nest Surface After the traffic density calculation we next estimate visually the ant density on the nest surface by using a ten point scale (0 = no ants, 10 = very dense population).

26.6.4.3 Temperature of the Nest One of the most significant indicators of colony vitality is nest temperature. High temperature within the whole nest interior reflects a vigorous colony. We have measured the temperature with long mercury thermometers or with a sond of electric thermometer from the nest surface to the depths of 5 , 10, 30, 40 and 50cm. The brood is found in the temperature range of 23 - 3 1 "C, the larvae in the lower part of the range, and pupae in the higher part of the range and also higher up in the nest (Fig. 4).

26.6.4.4 Sample Taking from Ant Roads Three samples from different ant roads are taken as described above in Sect. 26.6.3.1 and Sect. 26.6.3.2.

480

V . Monvarri et ril.

26.6.4.5 Measuring the Size of the Nest Mound It is most practical to measure the dimensions of the nest by using long measuring sticks. To describe the nest shape and to calculate the nest volume one measures the maximal diameter of the nest base and of the mound cone in the directions N-S and E-W, the total height of the nest base and the height of the mound cone (Maavara, 1991).

26.6.4.6 Sampling from the Nest In all cases we try to obtain samples from the nest in three replicates from different directions, taking the samples by the modes described above and in the following order: First from the top (Sect. 26.6.3.3), then from the upper part of the brood chamber (including pupae, Sect. 26.6.3.4), from the lower part of the brood chamber (including larvae, Set. 26.6.3.4), and from the diapause chambers of reserve ants (Sect. 26.6.3.5). We then take mineral soil and detritus from the vicinity of the nest (Sect. 26.6.3.6).

26.7 Feeding Experiments I t is possible to study the fate and effects of pollutants in red wood ant colonies by artificial feeding in the forest. Feeding has until now been performed at only onc point very near the nest, not necessarily by an ant road, because directional recruitment to food is known to occur in red wood ants (Rosengren and Fortelius, 1987). In future experiments higher efficiency may be achieved if the feeding is performed at several points near every ant road leading to the nest. Some data exists indicating that foragers on one particular road belong consistently to only one sector of the nest (Martin, 1978; Zakharov, 1974, 1991). By supplying the roads separately with different kinds of pollutants or radiomarkers it will be possible to test the validity of this sector hypothesis. If the artificial foods contain pollutants which may endanger the health of children or wild animals, it is necessary to offer the food under protection of a wire net cage. If beekeepers have their nests in the vicinity of the ant experimental area, it is necessary to keep the poisoned honey under a plastic cover with holes of size allowing ants but not bees to go through them.

26.7.1 Honey as a Feeding Substrate In honey feeding it is only necessary to mix the water solution of the pollutant with honey, and place it in the vicinity of an ant nest in early spring, when the honeydew of aphids is available only in limited amounts. Later in summer, the ants no longer

Sampling of Red Wood Ants

481

will eat polluted honey. During dry periods, when ants are thirsty, they will eat honey also in summer, if it thinned with water. The reason for ants’ summer avoidance of pure honey and preference for honeydew may merely depend on the water content. Honeydew contains some 90% water (Rosengren and Sundstrom, 1992) and the ants need much water for niantaining suitable humidity (85-92% RH) in brood chambers (Martin et al., 1985).

26.7.2 Fish as a Feeding Substrate In the middle of summer it is possible to offer pollutants mixed in grinded fish meat to the ants. An alternative is fish bearing a high load of contaminants which have been caught in polluted watercourses. Ants can take small pieces of fish as such and carry them to their nests (Rosengren and Fortelius, 1987), but mainly they take fly larvae which have developed in decaying fish (Nuorteva, 1977, p. 1088). Fly larvae have a much higher mercury content than do the fish consumed by them (Nuorteva et al., 1978a; Nuortevaet al., 1980; Nuorteva and Nuorteva, 1982). The transfer of mercury from fish to F. aquiloniu has also been shown (Nuorteva et al., 1978b). If the fish bait dries out during hot, rainless periods, the decay of the bait, as well as the action of the sarcosaprophags cease. Then the ants are unable to use this source of food. In the rains of autumn the dry carcass is softened and the sarcosaprophags return.

26.8 Description of some Pilot Studies already Performed Some four years ago the authors of this contribution realized that the ants represent in boreal forests an Achilles’ heel for metal pollution (Nuorteva, 1990, 1992; Nuorteva et al., 1992). We also realized that it is not sufficient to monitor the fate and effects of metals in red wood ant colonies only by taking samples of the surface workers. They need special monitoring where consideration is paid to all social groups separately. In the following we describe some pilot type results.

26.8.1 Natural Cd Levels in Different Castes and Worker Groups The first of our pilot study trials was performed on F. uquilonia material collected during the years 1983 and 1990 from the unpolluted forest of the Estonian ant protection area Akste and from the Valgesoo forest not far away. Data on Cd levels are given in Tab. 1, data on Mn, Zn and Cd levels in Fig. 5.

250 2000

1500

1000

500

O

l

R

Mn

Q

O

l

R

Q

Cd

Zn

Fig. 5. Levels of Mn, Zn and Cd in somc functional groups of I;. uquiloniu in the Akste ant protection arca in Estonia. White histograms show the maximum levels, the black blocks thc minimum lcvcls (modified from Oja, 1992).

Negative bioaccumulation occurred inside the ant colony, with the highest Cd levels in the foragers and surface ants, lower levels in inside workers and very low levels in the pupae. The Cd levels of the reserve ants were lower than of the other worker, but higher than those of the pupae. This results from the fact that inside workers had fed the reserve ants with Cd contaminated food in the period between their emergence and their migration to the hibernation chambers. The Cd levels of the sexuals remains, in contrast, near the low level of the pupae. This is of course very important for protection of the eggs.

Mean Levels of Cd of b’ortnicrr crquilonin in the Unpolluted Estonian Red Wood Ant Protection Area Akste and in the nearby Valgesoo Forest

Tab. 1.

Workers Foragers Surface workers Inside workers Pupae Kescrvc worken Malcs Queens

Numbcr of samples

ppm Cd/dwt

2 7 7 4 5 I 4

5.7 5.9 43 0.09 2.6 0.01 0.25

Sampling of Red Wood Ants

483

26.8.2 Cd Transfer to Ant Colonies under Pollution Stress Cd contents exceeding natural levels occur in ants under industrial pollution. Stary and Kubiznakova (1987) have produced some important pieces of evidence through their study on metals in Czechoslovakian red wood ant colonies living in semi-polluted and polluted areas. Though their material was not very extensive, they showed nicely that the F. polyctena foragers leaving the nest clearly have lower Cd levels than those having visited the aphid colonies and returning to the nest (Tab. 2). This difference did not, however, appear in their F. rgfu material. The occurrence of very low levels of Cd in pupae and sexuals of F.polyctenu was similar to that we noted for F. uquiloniu in Estonia. It remained unclear why they had two aberrant samples of inside workers of F. polyctenu, in which the Cd levels were higher than in the foragers and surface workers. This feature did not occur in their F. rgfa material. Tab. 2. Mean Cd Levels (in ppm/dwt) for Different Castes and Groups of Workers of Formiru polyctena and F. rgfu in Czechoslovakian Areas with only Background Pollution (Vcelna, Klet, Spalenec) and in a Locally Polluted Industrial and Mining Area (Kvetna) (compiled from data of Stary and Kubiznakova, 1987)

F. rufu

F. polyctenu

Samples ~~

ppm Cd

Background pollution workers : Foragers from the nest Foragers to the nest Surface workers Inside workers

8.4 22.6 24.3 47.6

Local pollution workers: Foragers from the nest Foragers to the nest Surface workers Inside workers Pupae

8.8 73.4 40.2 97.0 0.06

Queens

Samples

ppm Cd

~

35.3 34.6 27.6 29.8

2.0

Males

2

0.0

26.8.3 Artificial Cadmium Feeding Experiments In order to test behavior of Cd in concentrations exceeding Finnish natural levels, we fed ant colonies with honey containing different levels of Cd. While performing our feeding experiments in Tenala, Finland, we noted that Cd given to one nest also elevated the Cd levels of neighboring nests (Tab. 3). This

484

V. Mauvcrrrr et (11.

Tab, 3. Effect of Feeding CdCI, in Honey to a Series of Formica uyuiloniu Nests in Tenala, Finland Nests 1 - 1Oc situated in a row, 160 m long. in sunny ecotonc bctwccn an old fcrtilc coniferous forest and an agricultural field, nests 11 - 13 located separately inside a shady forest; ecotone nests ; of Cd contaminated honey: 0.5 kg, given in in three groups: 1-2. 3-6-7 and 8- 1 0 ~ amount early spring; samples of surface workcr ants taken on July I and 2, 1990 and 1991; NA = no ants; maximum temperatures ("C) inside the nests measured on July 17, 1990 and July I , 1991; the sond of the electronic thermometer (Gordon model 5300) was pushed into the nest vertically from the top until a depth of 50 cm; temperatures were noted at 10 cm intervals; * = missing Nest No.

Cd Treatment (PPm/Year)

Ecotone nests in a row (160 m) 1 2 Cd20- I990 3 4 5 6 7

-

Cd200- 1990 ~

Cd (ppm in ants)

Temperature in nests

1990

1991

1990

1991

13 13

11 11

25.8 24.9

26.0 18.0

18 22 23 28 28

12 13 Na 9

25.4 25.8 14.9 25.8 26.3

22.1 19.4 14.9 16.7 21.2

Na 9 Na 9 7

26.6 26.6 25.0 21.0 26.8

17.7 20.9 20.9 23.1 22.3

8

26.1 27.8

24.4 21.2 27.4

8 9 10a

10b IOC

Nests inside the shady forest, far from each other 8 II 12 Cd400-1990 55 13 Cd600- 1991 *

15

*

72

*

was an illustration of the existence of a supercolony. Such artificial feeding experiments obviously make it possible to measure the intensity of nest connections in supercolonies. During that experiment we noted also that temperature in the brood chambers, which is essential for the survival of red wood ant colonies, was not much affected by the artificial Cd feeding (Tab. 3). In an the artificial feeding experiment at Viikki on the outskirts of Helsinki, we followed the Cd distribution to the different castes and worker groups of F. ayuiloniu colonies using the methodology described above in Sect. 26.7.1. The results are given in Tab. 4. Here, the artificially added Cd was well distributed to the ant colony and at a larger scale showed the features less clearly seen in the material from unpolluted Estonian forests (Sect. 26.8.1, Tab. I). Proportionally, elevated dietary Cd access induced higher resistance to Cd transfer to reserve ants. From the results of the Viikki experiments it is clear that the offspring-sparing filter effect for Cd was already functioning when Cd was transferred from outdoor workers to inside workers. The most effective filter seemed, however, to exist between

Sampling

of Red Wood Ants

485

Tab. 4. Mean Cd Contents in Different Castes and Worker Groups of Formica aquilonia in an Artificial Feeding Experiment Performed in Viikki, Helsinki, Finland, April 25 - September I , 1991 The control nest colony was fed three times with 0.5 kg honey, Cd treated nests simultaneously three times with 0.5 kg honey containing 500 ppm CdC1, (summarized from Table 10 by Oja, 1992) Worker group

Number of samples

Control nests Surface workers Internal workers Pupae Reserve workers Queens Cd treated nests Surface workers Internal workers Larvae Pupae Reserve workers Queens

11

11 1 2 7 3

ppm Cd (dwt) in ants

Percentage in relation to surface workers

7.1 6.1 0.1 3.2 0.5

100 86

143.1 99.6 6.1 6.5 36.0 6.8

100 70 4 5 25 5

2 45 I

the nurses and the larvae (Tab. 4). As mentioned above in Sect. 26.5.2.2, the nurses feed the larvae first with their gland secretes, and later also with the liquid food circulating within the colony. It is, however, unclear whether the nurses themselves are really able to inhibit the transfer of metals to the larvae. The other possibility is that the older indoor workers are capable of binding the metals in their own tissues. In order to clear up this question it would be necessary to have separate samples of nurses and other indoor workers, which is important but a difficult task. One can say that the ability of ant workers to protect their offspring through filtering of poisons is negative bioaccumulation in the socio-biological food chain. This means that the ants have two poison tolerance systems: (1) the normal biochemical tolerance system and (2) the socio-biological tolerance system. This is one of the features giving exceptional possibilities for survival to the superorganisms of social insects. The exceptionally high poison tolerance of the fire ants (Solenopsis inuicta Buren and S. richteri Forel) is well documented (Banks et al., 1978, 1985). In future studies it would be desirable to relate more exactly the size of the red wood ant colony to the amount of pollutants offered to the colony by artificial feeding. This would make it possible to calculate exactly the amount of pollutants needed to elevate the pollutant level in red wood ants to dangerous dimensions. It is, however, not possible to think that one would be able to define some tolerance limits. Like most environmental pollution damage, ant decline is most possibly also a multistress syndrome. Thus the metals are only one factor among a multitude of other stresses, including the natural ones, and the decline of an ant colony occurs when the stress total exceeds the tolerance limit.

486

V . Muuourri ef ul.

26.8.4 Effects of Metal Pollution on the Enzyme Balance Artificial feeding of ant colonies in Tenala offered a possibility to study what will happen to ants, if the Cd and Hg pollution of the forests continues. In collaboration with enzyme specialists from the Silesian University it was possible to note that enzyme disturbance occurs only if the Cd content of Finnish ants rises to the levels now occurring in the polluted areas of Central Europe. These studies showed that Cd and Hg inhibit the action of enzymes participating in the AMP, ADP and ATP syntheses (Migula et al., 1993) on which directly depends the formation of lipids and the development of newly emerged ants into reserve workers (Maavara and Martin, 1983b). In addition, it has been noted that Cd decreases in insects’ food digestibility and consumption (Migula et al., 1989). In this respect it was shocking to note during the year 1993 that reserve workers were absent in some nests treated continuously with Cd and Hg. It was also reminder of the fact that pollution damage occurs not only where pollutant levels are highest. Organisms may have a low pollutant content merely because they are especially sensitive to a pollutant and have developed systems to keep the content at a low level during their evolution.

26.5 References Adlung, K . Ci. (1966) A critical evaluation of the European research on use of red wood ants (Formicu ridu group) for the protcction of forests against harmful insects. Zeitsclir. ungew. fknfOf7lOl.57, I67 - 189. Banks, W. A,, Ltifgrcn. C. S., Wojcik. D. P. (1978) A bibliography of imported fire ants and the chemicals and methods used for their control. United States Agric. Res. S e r i k e Puhl. 180, 1-35. Banks, W . A , , Lofgren, C. S.. Williams, D. F. (1985) Development of toxic baits for control of imported fire ants. In: T. M. Kancko and L. D. Spicer (eds.) Pesticidr Formulutions urid Appplicution system.^: Fourth Sympasium. Amer. Soc. Testing Materials, Philadelphia, pp. I33 to 143. Baroni Urbani, C., Collingwood, C. A. (1977) The Zoogeography of ants (Hymenoptera, Formicidae) in Northern Europe. Actu Zaol. Fennicu 152, 1-34. Brandt, D. Ch. (1980) The thermal diffusivity of the organic material of a mound of Formica polyetma Foerst. in relation to the thermorcgulation of the brood (Hymenoptera, Formicidae). Nethrrlands Journ. Zool. 30, 326 - 344. Breton, L. M., Addicot, J. F. (1992) Density-dependent mutualism in an aphid-ant interaction. E c o l o ~ y73, 2175-2180. Catzcflis, F. (1979) Etude qualitative et quantitativc de l’avifaunc dc la pressiere jurassienne du Chalet a Roch. Vaud. Nos Oiseuitx 35, 75 - 84. Cocncn-Stass, D., Saarschmidt, B., Lambrecht. I . (1980) Temperature distribution and calorimetric detcrmination of heat production in the nest of thc wood ant, Formica po/-vctena (Hymenoptera, Formicidac). Ecology 61, 238 - 244. Crawford, L. A,, Hodkinson, I. D., Lepp, N. W. (1983)Thc cffcctsofcopper and cadmium ingestion on aspects of the biology of herbivorous insects. Proc. Internat. Conf: Henvy Metals in the Enuirownent, Heidelberg, pp. 810-813. Debouge, M. H., Thome, J. P. (1988) Dynamique d’accumulation des biphenyles polychlores (PCB) chcz Formicapolyctena (Hymenopteres, Formicidae). Ann. Soc. r. Zoo/. Belgique 118, 131 - 139.

Sampling of’ Red Wood Ants

487

Debouge, M. H., Thome, J. P. (1989) Dispersion du lindane dans 5 especes de fourmis en Wallonie et cinetique d’accumulation chez Formica polyctena. Ann. Sci. Naturelles Zool., Paris 13 (lo), 25 - 30. Debouge, M. H., Thome, J. P., Jeuniaux, Ch. (1987) Bioaccumulation de trois insecticides orgdnochlores (lindane, dieldrine et DDT) et des PCB chez plusieurs especes de fourmis (Hymenoptera-Formicidae) en Belgique. Entomophaga 32, 55 1 - 561. Dlussky, G. M . (1967) The Ant Genus Formica (in Russian), Nauka, Moscow, 236 pp. Elton, E. T. G. (1958) The artificial establishment of wood ant colonies for biological control in the Netherlands. Proc. Xth Internat. Congr. Entomol. 4, 573 - 578. Elton, E. T. G . (1989) On transmission of the labial gland disease in Formica rufa and Formica polyctena (Hymenoptera, Formicidae). Proc. Koningklijke Nederland.7e Akad. Vetenschappen Ser. C, 92, 415-455. Elton, E. T. G . (1991) Labial gland discase in the genus Formica (Formicidae, Hymenoptera). Insectes Sociaux 38, 9 1 - 93. Fangmeier, A,, Steubing, L. (1986) Cadmium and lead in the food web of a forest ecosystem. In: H.-W. Georgii (ed.) Atmospheric Pollutants in Forest Areas, Dodrecht, pp. 223 - 234. Finnegan, R. J . (1975) Introduction of predacious red wood ant, Formica lugubris from Italy to eastern Canada. Canada. Entomol. 107, 1271 - 1274. Fortelius, W., Rosengren, R., Cherix, D., Chautems, D. (1993) Queen recruitment in a highly polygynous supercolony of Formica lugubris (Hymenoptera, Formicidae). Oikos 67, 193 - 200. Gosswald, K. (1952) Die rote Waldameise im Dienste der Waldhygiene, Metta Kinau Verlag, Liineburg, 160 pp. Gosswald, K. (1979) Auswirkungen der Waldameisen im Eichenwald. Wuldhygiene 13, 3 - 10. Gosswald, K. (1989) Die Waldameise, AULA-Verlag, Wiesbaden, 660 pp. Haemig, P. D. (1992) Competition between ants and birds in a Swedish forest. Oikos65,479-4483. Heimann, M. (1 963a) Zum Warmehaushalt der kleinen Roten Waldameise (Formica polyctena Foerst.) I. Waldhygiene 5, 1-21. Heimann, M. (1963b) Zum Warmehaushalt der kleinen Roten Waldameise (Formica polyctena Foerst.) 11. Waldhygiene 5, 58-67. Heimann, M. (1964) Zum Warmehaushalt der kleinen Roten Waldameise (Formica polyctena Foerst.) 111. Waldhygiene 5, 136- 146. Holldobler, B., Wilson, 0. E. (1990) The Ants, Springer Verlag, Berlin, Heidelberg, New York, 860 pp. Horstmann, K. (1 990) Zur Entstehung des Wgrmezentrums in Waldameisennestern (Formica polyctena Forster: Hymenoptera, Formicidae). Zool. Beitr. N.F. 33, 105 - 124. Kneitz, G. (1 970) Saisonale Veranderungen des Nestwarmehaushaltes bei Waldameisen in Abhangigkeit von Konstitution und dem Verhalten der Arbeiterinnen als Beispiel vorteilhafter Anpassung eines Insektenstaates an das Jahreszeitenklinia. Verhandlungsbericht Deutsche Zool. Gesellschaft 64, 3 18 - 322. Laine, K. J., Niemela, P. (1980) The influence of ants on the survival of mountain birch during an Oporinia autumnata (Lep., Geometridae) outbreak. Oecologia 47, 39 -42. Laine, K. J., Niemelii, P. (1989) Nests and nest sites of red wood ants (Hymenoptera, Formicidae) in Subarctic Finland. Ann. Entomol. Fennici 55, 81 -87. Maavara, V. (1991) Volume determination of the nest mounds of red wood ants (in Russian). Proc. I X Symposion Myrrnecologists of USSR in Kolotshava, pp. 11 - 14. Maavara, V., Martin, A. (1983a) Protection and utilization of red wood ants in Estonia (in Russian). Acta et Commentationis Univ. Tartuensis 641, 94- 101. Maavara, V., Martin, A. (1983b) The role of reserve ants in red wood ant colonies (in Russian). Proc. IIIrd All-Union Con5 Ethological Mechanisms in Moscow, pp. 102- 104. Martin, A.-J. (1975) The shape and orientation of the mound nest of Formica uquilonia Yarrow depending on the location conditions (in Estonian). Eesti NSV Teaduste Akad. Toim. B i d . 24, 109- 117. Martin, A.-J. (1978) Nest territories in red wood ant supercolonies (in Estonian). Tartu iilikooli Noorte Teadlaste Eteekannede Teesid, pp. 19 - 20. Martin, A.-J. (1980a) Vernal thermoregulation in the nest mounds of the red wood ant Formica aquilonia Yarrow. I. Passive warming of the nest (in Russian). Eesti N S V Teaduste Akad. Toim. B i d . 29. 103-107.

488

V . Mauvura et ul.

Martin, A.-J. (1980b) Vernal thermoregulation in the nest mounds of the red wood ant Formicu uquilonia Yarrow. 11. The active heating of brood chambers (in Russian). Eesti NSV Teadustr Akud. Toim. Biol. 29. 188 - 197. Martin, A,-J. (1 987) The variability of the shape and inside temperature of ants’ nest mounds in relation to their ecological adaptation (in Russian). Proc. X Symposion Myrmecologists of‘ U S S R in Novosibirsk, pp. 127 - 132. Martin. A.-J. (1991) The behaviour and role of reserve ants in red wood ant colonies (in Russian). Proc. I X Symposion Myrmecologists of U S S R in Kolotshavu, pp. 98 - 101. Martin. A,-J., Maavara, V.. Mand, M. (1985) The effcct of artificial heating or ant nest for the life of a colony of the red wood ant Furmicu uquilonia Yarr (in Russian). Proc. Meeting on Methods m d Results in Insect Physiology in Turtu, pp. 55 - 60. Migula, P., Kedziorski, A., Maconieczny, M., Kafel, A. (1989) Combined and separate effects of heavy metals on energy budget and metal balances in Achuetu domesticus. Uttur Pradeh Journ. ZOO^. 9, 140- 149. Migula. P., Nuorteva. P., Nuorteva, S.-L., Glowacka, E., Oja. A. (1993) Physiological disturbances in ants (Formica aquiloniu) to excess of cadmium and mercury in a Finnish forest. Sci. Total Environment (in press). Muller, H . (1966) Der Honigtau als Nahrung dcr hiigelbauenden Waldameisen. Entomophugu 5, 55 - 75. Nuorteva, P. (1977) Sarcosaprophagous insects as forensic indicators. In: C. G . Tedeshi, W. G. Ekkcrt and L. G. Tcdeshi (eds.) Forensic Med. ZI, W. B. Saunders Co, Philadelphia, London, Toronto, pp. 1072- 1095. Nuorteva, P. (1’188) European forest decline, metals and insects. Proc. X V I I I Internat. Congr. Entornol. Vuncouver B.C., Cnnada, p. 436. Nuorteva, P. (1990) Metal distribution patterns and forest decline. Seeking Achilles’ heels for metals in Finnish forest biocoenoses. Puhl. Dept. Environm. Conservation Univ. Helsinki 1 I, 1 - 77. Nuorteva, P. (1992) Achilles’ heels for metals in the boreal forest ecosystem. Proc. X I X Internut. Congr. Entntnol. Bejjing, p. 435. Nuorteva, P., Hlslnen, E., Nuorteva, S.-L. (1978a) Bioaccumulation of mercury in sarcosaprophagous insects. Norwegian Journ. Entomol. 25, 79 - 80. Nuortcva, P., Wuorenrinne, H., Kaistila, M. (1978b) Transfer of mercury from fish carcass t o Formicu uyuilonia (Hymenoptera, Formicidae). Ann. Entomol. Fennici 44, 85 - 86. Nuorteva, P., Nuorteva, S.-L., Suckcharoen, S. (1980) Bioaccumulation of mercury in blowflies collected near the mercury mine of Idrija, Yugoslavia. Bull. Environm. Conturn. Toxicology 24, 51 5 - 521. Nuortcva, P., Nuorteva, S.-L. (1982) The fate of mercury in sarcosaprophagous flies and in insects eating them. Arnbio 11, 34-37. Nuortcva, P., Nuorteva. S.-L., Oja, A,, Lehtinen, H., Salo. S. (1992) Two Achilles’ heels for metals in the Finnish forest ecosystem. Proc. VIth Internut. Conj: Bioindicatores Deteriorisalionis Regionis, C‘eske Budejoirice, pp. 72 - 17. Oinonen, E. (1956) On the ants of the rocks and their contribution to the afforestation of rocks in southern Finland. Actu Enrol. Fennici 12, 1-212. Oja, A . (1992) Monede metallide (Al, Fc, Mn, Zn, Cu ja Cd) sisaldus Eesti ja Soome eri saastutusastmcga piirikondadest kogutud metsakuklastes (Hymenoptera, Formicidae, Formicu spp.). Thesis in Estonian language by the Department of Environmental Conservation of Helsinki University in Finland, 74 pp. Otto, D. (1962) Die roten Wuldameisen. Die Neue Brehm-Bucherei 293, A. Ziemsen Verlag. Wittenberg-Lutherstadt. 15I pp. Otto, D. (1967) Die Bedeutung der Formica-Volker fur die Dezimierung der wichtigsten Schadinsektcn. - Ein Literaturbericht. Wuldhygiene 7, 65-90. Rosengren, R. (1971) Route fidelity. visual memory and recruitment behavior in foraging wood ants of the genus Formica (Hymenoptera: Formicidae). Acta Zool. Fennica 133, I - 102. Rosengrcn, R. (1977) Foraging strategy of wood ants (Forrnicu rufu group) I. Age polyethism and topographic traditions. A c t u Zool. Fennica 149, 1-30. Rosengren, R. (1979) Labialkortel-syndromet som “epidcmi” i en t l t stackpopulation av F o r m i u uquiloniu (Hymenoptera, Formicidae). Memor. Soc. Fuunu Flora Fennicu 55, 73 - 84.

Sampling of Red Wood Ants

489

Rosengren, R., Fortelius, W. (1987) Trail communication and directional recruitment to food in red wood ants. Ann. Zool. Fennici 24, 137- 146. Rosengren, R., Sundstrom, L. (1987) The foraging system of a red wood ant colony (Formica s. str.) - collecting and defending food through an extended phenotype. Experientia, supplementum: Behavior in Social Insects 54, 1 17 - 137. Rosengren, R., Sundstrom, L. (1991) The interaction between red wood ants, Cinaru aphids, and pines. A ghost ofmutualism past? In: C. R. Huxley and D. F. Cutler (eds.) Ant-Plunt Interactions, Oxford University Press, Oxford, New York, Tokyo, pp. 80-91. Rosengren, R., Vepsiliinen, K., Wuorenrinne, H. (1 979) Distribution, nest densities, and ecological significance of wood ants (the Formica rufa group) in Finland. Bulletin SROP (Org. Internal. Lutte Biologique contre des Animaux et les Plantes Nuisibles) 2, 181-213. Rosengren, R., Cherix, D., Pamilo, P. (1985) Insular ecology ofthe red wood ant Formica truncorum Fabr. I. Polydomous nesting, population size and foraging. Bull. SOL..Entomol. Suisse 58, 145- 175. Rosengren, R., Fortelius, W., Lindstrom, K., Luther, A. (1987) Phenology and causation of nest heating and thermo-regulation in red wood ants of the Formica rufu group studied in coniferous forest habitats in southern Finland. Ann. Zool. Fennici 24, 147- 155. Schmidt, G. H. (1969) Ein Beitrag zur Fruhjahrsaktivierung der hugelbauenden Waldameisen. Zool. Beitrage, N.F. 15, 171 - 183. Schmidt, G. H. (1974) Steuerung der Kastenhildung und Geschlechtsregulation im Ameisenstaat. In: E. G. Schmidt (ed.) Soziulpolymorphismus bei Insekten. Prohleme der Kastenbildung im Tierreich. Wissensch. Verlagsgesellschaft m.b.h., Stuttgart, pp. 404- 512. Skinner, G., Whittaker, J . B. (1981) An experimental investigation of the interrelationships between the wood ant (Formica rufa) and some tree canopy herbivores. Journ. AnimalErol. 50.31 3 - 326. Stary, 9. (1988) Atlas of inserts beneficiul to forest trees, Vol. 2, Elsevier, Amsterdam, Oxford, New York, Tokyo 1988, IOOpp. Stary, P., Kubiznakova, J. (1987) Content and transfer of heavy metal air pollutants in Formica ssp. wood ants (Hym., Formicidae). Zeitschr. ungew. Entomol. 104, 1 - 10. Stanley, R. J., King, W., King, T. J. (1991), pp. 521 -535, in: C. R. Huxley and D. F. Cutler (eds.) Ant-Plant Interactions, Oxford University Press, Oxford, New York, Tokyo, 601 pp. Vogel, W. R., Nopp, H., Fuhrer, E. (1988) Stoffwechsel, Entwicklung und Fortpflanzung von Insekten und Antagonisten unter SchwermetalleinfluB. FIW-Symposium 1988: Waldsterhen in &terreic/z. Theorien, Tendenzen, Therapien. Universitat fur Bodenkultur, Wien, pp. 228 - 238. Warrington, S., Whittaker, J. 9. (1985) An experimental field study of different levels of insect herbivory induced by Formica rufa predation on sycamore (Acerpseudoplatanus). I. Lepidoptera larvae. Journ. Appl. Ecol. 22, 775-785. Whittaker, J . 9 . (1991) Effects of ants on temperate woodland trees. In: C. R. Huxley and D. F. Cutler (eds.) Ant-Plant Interactions, Oxford University Press, Oxford, New York, Tokyo, pp. 80-91. Whittaker, J . B., Warrington, S. (1985) An experimental field study of different levels of insect herbivory induced by Formica rufa predation on sycamore (Acer pseudoplatanus). 111. Effects on tree growth. Journ. Appl. Ecol. 22, 797 - 8 1 1. Woodman, R. L., Price, P. W. (1992) Differential larval predation by ants can influence willow sawfly community structure. Ecology 73, 1028 - 1037. Yli-Mononen, L., Salminen, P., Wuorenrinne, H., Tulisalo, E., Nuorteva, P. (1989) Levels of Fe, Al, Zn and Cd in Formica aquilonia, F. polyctenu and Myrmica rufinodis (Hymenoptera, Formicidae) collected in the vicinity of spruces showing different degrees of needle-loss. Ann. Entomol. Fennici 55, 57 - 61. Zakharov, A. A. (1974) Structures and formation of the colonies of red wood ants (Formica s. str., Hymenoptera, Formicidae) (in Russian). Zool. Zhurn. 51, 58 -65. Zakharov, A. A. (1991) The Organization c f A n t Communities (in Russian), Nauka, Moscow, 278 pp.

Part IV Literature Survey

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH VerlagsgesellschaftmbH, 1994

27 Overview o References for Sampling and Related Topics Susanne Hannuppel

27.1 Introduction As sampling is part of every environmental investigation, literature about sampling can be found in many fields, the most important being air, water, soil, plants, animals, and waste. Each field has its own characteristics and problems concerning sampling plan and statistics, sampling equipment and techniques, sampling transport and preservation. Although not many books cover every step of the sampling process systematically, much -research has been done on its various aspects. This is reflected in the vast amount of journal articles and reports on sampling which can be found in large databases such as Chemical Abstracts, Analytical Abstracts, and Biosis. The following survey is intended to give a broad overview of sampling literature. In order to gain deeper insight into a certain field, the reader should investigate the references in the literature given.

27.2 Literature 27.2.1 General Aspects Adriano, D. C., 1992. Biogeochemistry qf trace metals, Boca Raton: Lewis Publishers. Ambe, Y., Okamoto, K., 1992, Specimcn hunk activities at NIES, in: Rossbach, M., Schladot J. D., Ostapczuk P. (eds.). Specimen Banking, Berlin, Heidelberg, New York: Springer Verlag. Anders, 0. U., Kim, K. J., 1911, Representative sampling and the proper use of reference materials, Journal of Radioanalytical Chemistry 39, 435 -445. Barcelona, M. J., 1988, Overview of the sampling process, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society. Baiulescu, G. E., Dumitrescu, P. Zugravescu, P. G., 1991, Sampling, New York: Ellis Horwood. Behne, D., 1981 Sources of error in sampling and sample preparation ,fir trace element analysis in medicine, J. Clin. Chem. Clin. Biochem. 19, 115- 120. Berg, E. L., 1982, Handbook.fbr sampling and sample preservation, Washington D.C. : U.S. Government Printing Office. Black, S. C., 1988, Dcifining control sites and hlunk sample needy, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society. Boehringer, U . R., 1992, Progress of six years experience with environmental specimen banking in the Federal Republic qfGermany, in: Rossbach, M., Schladot J. D., Ostapczuk P. (eds.), Specimen Banking. Berlin, Heidelberg, New York: Springer-Verlag. ~

494

S. Hanniippel

Boswcll, M. T., Patil, G. P., 1987, A perspective in composite sampling, Communications in Statistics Theory and Methods 16 (lo), 3069 - 3093. Bryden, G. W., Smith. L. R., 1989a, Sampling ,for environmental analysis. part I : planning and preparation, American Laboratory 21 (7), 30-39. Bryden, G . W.. Smith, L. R., 1989b, Sampling f b r environmcJntul analysis, part 2 : sampling methodology, American Laboratory 21 (9), 19 - 24. Bundcsministerium fur Forschung und Tcchnologie, (ed.), 1988, UmMdtprohenhank, Berlin, Heidelberg, New York: Springer Verlag. Cale, W. G., McKown, M. P., 1986, A cost analysis techniqucSr,r re.rearch management and design. Environ. Management 10, 89 -96. Cochran, W. G., 1977, Sampling fechniques, 3rd ed., New York: Wiley. Desu, M. M., Raghavarao, D., 1990, Sample size nwthodology, Boston: Academic Press. Evans, R. D., Dillon, P. J . , 1986, Puperpresentedat the EUCHEM conJ>rerrceon sampling strategicJs and techniques in environmental analysis, The Netherlands, Bilthoven: RIVM. Flatman, G. T., Englund, E. J., Yflantis, A. A., 1988, Geostatistical approaches fo the design of sampling regimes, in: Keith, L. H . (ed.), Principles of environmental sampling, ACS Professional Refcrence Book, Washington D.C.: American Chemical Society. Fleming, G. A,, Tunney, H., O’Riordan, E. G., 1986, The sampling ofsoils, herbage, animalmunures rind sew5age sludge f o r trace element and other cmalyses - Irish experiences, in: Gomez, A,, Leschber, R., L’Hcrmite, P.(eds.) Sampling problems for the chemical analysis of sludge, soils and plants, London, New York: Elsevier Applied Science Publishers. Gy, P. M., 1982, Sampling of particulare materials, Amsterdam: Elsvier Publishers. Holcombe, L. J., 1988, Relations of sampling design to unalytical precision estimates, in: Keith, L. H. (cd.), Principlcs of environmental sampling, ACS Professional Reference Book, Washington D.C.: American Chemical Socicty. Hushon, J. M. (ed.), 1990, Expert .systemsfor environmental irpplications, Weinheim: VCH Verlagsgcsellschaft mbH. Jolly, G. M., 1979, Sampling oflarge ohjects, in: Cormack, R., Patil, G. P., Robson, D. S., Sampling biological populations, Burtonsville, Maryland: International Co-operative Publishing Housc. Jolly, G. M., Watson, R. M., 1979, Aerirrl sample survey methods in the quantitative as.se.ssrnci~/ of ecological resources, in: Cormack, R., Patil, G. P., Robson, D. S., Sampling biological populations, Burtonsville, Maryland : International Co-opcrative Publishing House. Keith, L. H., Johnston, M. T., Lewis, D. L., 1988, Defining quality assurance and quality control sampling requirements: expert systems andaids, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society. Keith, L. H., 1991, t.Jnnironmenta1 sampling and analysis, A Practical Guide, Chelsea: Lewis Publishers. Kraft, G., 1980, Theoretische Grundlagen tier Probenahme, Probenahme - Theorie und f’raxis, Schriftenreihe der Gesellschaft Deutscher Metallhutten- und Bergleute, Weinheim : VCH Verlagsgescllschaft mbH. Kraft, G., 1993, Probenahme an,fisten StqNen, in: KienitL, H., Bock, R., Frescnius, W., Huber, W.. Tolg, G. (eds.), Analytiker Taschenbuch, Bd. I , Berlin, Heidelberg, New York: Springer Verlag. Kratochvil, €3.. Taylor, J. K., 1981, Samplingfor chemird analysis, Anal. Chem. 53, 924-938. Kratochvil, B., Wallace, D., Taylor, J. K., 1984, Sampling for chemical analysis, Anal. Chem. 56, 113R- 119R. Leschber, R., 1986, Samp1ing:fundamentd aspects, in: Gomez, A,, Leschber, R., L’Hermite, P.(eds.) Sampling problems for the chemical analysis of sludge, soils and plants, London, New York: Elsevier Applied Science Publishers. Lewis, D. L., 1988, Assessing and controlling sample contamination, in: Keith, L . H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C.: American Chemical Society. Markert, B., Klausmcyer, N., 1990, Computer - aided samplitig of’ the environment ( C A S E ) Ji)r multi-element analysis, in: Lieth, H., Markert, B., Element concentration cadasters in ecosystems, Weinheim: VCH Verlagsgesellschaft mbH. Martin, M., Couphtrcy, P., 1982, Biological monitoring of heavy metal pollution, London, Ncw York: Appl. Sci. Publ..

Overvienl qf R
495

Matthies, M., Altschuh, J., Briiggemann, R., Voigt, K., 1992, Duta needs, duta availuhility unddatu e.stimution,fbrenvironmentul e.uposure und huzuru’iisses.smPnt, GSF Report 1/92, Ncuherberg : GSF. Maskarinec, M. P., Moody, R. L., 1988, Storuge and preservation qf environmental sunzples, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society. M iiller, P., Wagner, G., 1 986, Prohenahme und gtvwtisclie Vergleichbarkeit (Prohendqfinition) von repriisentutiven Umweltprohcn im Riihmen des Um,~’eltprobenbank-Pilotprqjekte.s,BMFT-Forschungsbericht T 86-040. Muntau, H., Son, M., von Baudo, R., Schramel, P.. Marengo, G., Lattanzio, A,, Amantini, L., 1986, Puper presented at the EUCHEM conserenee on sampling strutegies and techniques in environmental analysis, The Netherlands, Bilthoven: RIVM. Myers, W. L., Shclton, R. L., 1980, Survey methoukfor eco&s.ystemmanugement, New York: Wiley. Pandurang Sukhatme, V. (et al.), 1984, Sumpling-theory surveys applications, Iowa: Iowa Statc University Press. Pao. Y. H., 1989, Adupfive pattern recognition and neurul networks, Reading: Addison-Wesley. Parker, L. V., 1992, Suggested guidelines f b r the use of PTFE, P V C and stainless steel in samplers und i o d casings, ASTM STP 1118, 217-29. Rose, Ch. D., 1992, A fructul model,fhr sampling C O C J ~ ores, , und other nutural populations, J . Coal Quai. 11 (1-2). 6-13. Smith, F., Kulkarni, S . , Myers, L. E., Messner, M. J., 1988, Evuluating and presenting yuulity ussurance sanzpling &a, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C.: American Chemical Society. Spengart, J., Wagner. G., 1993, Umweltprobenbank des Bundes - Richtlinie zur Prohenuhme und Prohenuufbereitung, Universitit dcs Saarlandcs. Stoeppler, M., Diirbeck, H. W., Nuernberg, H. W., 1982, Environmental specimen bunking, A challenge in trace analysis, Talanta 29. 963 - 972. Streit, B., Stumm, W., 1993, Chrmiccilproperties of’metuls and the process ojhioaccumulation in terrestriulplunts, in: Markert, B. (ed.), Plants as biomonitors, Weinheim: VCH Verlagsgesellschaft. Thompson, G., Bankston, D. C., 1970, Sumple contamination,from grinding unrl sieving determined hv emission .spectometry, Applied Spectroscopy 24 (2), 2 10- 2 19. Thompson, M.. Howarth, R. J., 1976, Duplicate anulysis in geochemicalpructice, Purr I , Theoreticul upproach and estimation q f analyticul reproducihility, Analyst 101, 690-698. Tomlison, R. C., 1959, Sumpling, in: Wilson, C. L. Wilson, D. W. (eds.), Comprehensive Analytical Chemistry, Vol. IA, Amsterdam. Vctter, L., 1989, Evnluierung und Entbvicklung stutistischer Verfahren zur Auswahl von reprusentutiven Untersue~iunRsobjekten.fur i;koto.uikologischr Problemstellungen, Kk!: Dissertation. Wagner, G., 1990, Konzept und Betrieh der Deutschen Umweltpr~ihenbunk,in: Okosystemforschung im Bereich der Bornhoveder Scenkette, Kiel: Heft I . Williams, H., 1978, A sampler on sumpling, New York: Wiley. Wise, S. A,, Zcisler, R., 1985, International revim1 of ent)ironmental specimen banking. NBS Special Publication, No. 706. Wittig, R., 1993, Generul aspects o hiomonitoring heuiiy metals by plunts, in: Markert, B. (ed.), Plants as biomonitors, Weinheim: VCH Verlagsgesellschaft. Woodbet, B. W., Copper, D., 1987, Sumples and standards, London: Wiley. Zeisler, R., Koster, B. J., Wise, S. A,, 1992, Specimen bunking at the Nutional Institute qf Standurds and Technology, in: Rossbach, M., Schladot J . D., Ostapczuk P. (eds.), Specimen Banking, Berlin, Heidelberg, New York : Springer Verlag.

27.2.2 Statistical Methodology Aitchison, J., Brown, J. A.C., 1966, 7he lognormul distribution, Cambridge: Cambridge University Press. Barlow, F. J., 1989, Statistics, New York, Wiley.

496

S. Hannappel

Belsley, D. A,, Kuh, E., Welsch, R. E., 1980, Regression diagnostics: identifying influential data and sources qf collinearity, New York: Wiley. Borgman, L. E., William, F. Q,, 1988, Sampling for tests of hypothesis when data are correlated in space and time, in: Kcith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C.: American Chemical Society. Box, G. E. P., Jenkins, G. M., 1976, Erne series analysis, forecasting and control, San Francisco: Holden-Day. Box, G. E. P. (et al.), 1978, Statisticsfor experimentulists, New York: Wiley. Brookes, C. J., Betteley, I. G., Loxston, S. M., 1966, Mathematics and statistics, London: Wiley. Campbell, S., Scott, H., 1985, Quality a,s.surance for environmental undysis, in: Taylor, J. K., Stanley, T. W. (eds.), Philadelphia: American Society for Testing and Materials. Carroll, R. J., Ruppert, D., 1988, 7'ranJformation and weighting in regression, London: Chapman and Hall. Chernoff, H., 1973, 7he use o f f i x e s fo represent points in k-dimensional space graphically, J. Am. Statist. Assoc. 68, 361 - 168. Cohen, A. C. Jr., 1959, Simpl@d estimators ,for the normal distribution when samples ure singly censored or truncated, Technometrics 1, 217- 237. Conover, W. J., 1980, Practical nonparametric statistics, New York: Wiley. Deming, S. N., 1984, Experimental design: response sufaces, in: Kowalski, B. R. (ed.),Chemometrics - Mathematics and Statistics in Chemistry, New York: D. Reidel. Devries, G. P., 1979, Line intersect sampling - statistical theory, applications and suggestions for extended use in ecological inventory, in: Cormack, R., Patil, G. P., Robson, D. S., Sampling biological populations, Burtonsville, Maryland: International Co-operative Publishing House. Doerffel, K., 1990, Statistik in der analytischen Chemie, Leipzig: Deutscher Verlag fur Grundstoffindustrie GmbH. Dutilleul, P., 1993, Spatial heterogeneity and the design of ecological field experiments, Ecology 74 (6), 1646- 1658. Du Toit, S. H. C., Steyn, A. G. W., Stumpf, R. H., 1986, Graphical explorntory data analysis, New York: Springer Verlag. Ebdon, D., 1977, Statistics in geography. A practical approach, Oxford: Blackwell. El-Shaarawi, A. H., D o h , D. M., 1989, Maximum likelihood estimates ofwater quality concentrations ,from censored data, Can. J. Fish. Aquat. Sci. 46, 1033- 1039. El-Shaarawi, Nadcri, A,, 1991, Statistical inference ,from multiply censored environmental data. Environ. Monit. Assess. 17, 339 - 347. El-Shaarawi, Esterby, S. R., Replucernent of censored observations by a constant: an evuluation, Water Res. 26, 835-844. Fahrmeir, L., Hamerle, A , , 1984. Multivariate stotistische Verfuhren, Berlin, New York : Walter dc Gruykr. und mathemutische Statistik, Berlin: VEB Dcutscher Fisz, M., 1989, Wahrsclieinlichkeit.sre~~ht~ung Verlag der Wissenschaften. Funk, W., Dammann, V., Donncvcrt, G., 1992, Qualititssicherung in der analytischen C'hrwzic,, Weinheim: VCH Verlagsgcsellschaft mbH. Gauch, H. G., Chase, G . B., 1974, Fitting the Guitssiun curve to ecologicul duta, Ecology 55, 1377-1381. Gauch, H. G.. 1982. Multivariate [znalysis in community ecology, Cambridge: Cambridge University Press. Gray, J . S., 1983, USPandrni.su.se ofthe log-normulplotting method,for detection ofefyects qfpollution - a reply to Shaw et ul., Mar. Ecol. Prog. Scr. 11, 203-204. Green, R., 1979, Sampling design arid statistical methods fijr environmental biologists, New York: Wiley & Sons. Hampel, F. R., Ronchetti, E., Roussceuw, P. J., Stahel, W. A,, 1980, Robust statistics - tho approach bused on influence functions, New York: Wiley. Hartung, J., Elpelt, B., Klosener, K. 1-1.. 1991, Stutistik, Lehr- und Handbuch der ungewundten Statistik, Miinchen: Oldenbourg Verlag GmbH.

Overview of References f o r Sampling and Related Topics

497

Haswell, S. H., 1992. Pructical guide t o chemornetrics, New York, Basel, Hong Kong: Marcel Dekker Inc.. Hoaglin, D. C.. Mosteller, F., Tukey J. W. (eds.), 1983, Understanding rohust and elrploratory data analysis, New York: Wiley. Joileffe, I. T., 1986, Principle compunenf analysis, New York: Springer Verlag. Jornel, A. G., 1988, Nonpararnetric geostafistics f o r risk nnd additional sampling assessment, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society. Kateman, G., Piipers, F. W., 1981, Quality control in analyticul chemistry, New York: Wiley. Kateman, G., 1987, Chemom~~trics-.sampling strategies, Chemometrics and Species Identification, Topics in Current Chemistry, 141, New York: Springer Verlag. Krzanowski, W. J.. 1988, Principles qf multivariate analysis - a user’s perspective, Oxford: Oxford University Press. Legendre, P., 1993, Spatial autocorrelution: trouble or new’ paradigm?, Ecology 74 (6), 1659- 1673. Lienert, G. A,, 1986, Verteilungsfreie Methoden in der Biostutistik, Bd. 1 und Bd. 2, Konigsstein/Ts.: Verlag Anton Hain. Liteanu, C., Rica, I., 1980, Statistical theory and methodology of trace analysis, New York: Halsted. Mancy, K. H., Allen, H. E., 1982, Design of measurement systems, in: Suess, M. J. (ed.), Examination of water for pollution control. A reference handbook, Oxford: Pergamon Press. Maritz, J. S . , 1981. Distrihution-free statistical methods, London: Chapman and Hall. Myers, R. H., 1990, Classicrrl and modern regression with applications, Boston: PWS-KENT Publishing Company. Patil, G. P., 199 I Encountered duta, statistical ecology, environmental statistics, and weighted di,strihution methods. Environmetrics 2 (4). 377 -41 5. Potvin, C.. Derek, A. R., 1993, Distrihution -,free and rohust statistical met1iod.s: viahle alternatives to purumetric statistics Y, Ecology, 74 (6), 1617 - 1628. Ramsey, M. H., Thompson, M., Hale, M., 1992, Uhjectiw evu[uation ufprecision requirements f o r geochemical anal~vsisus in^ robust analysis of’ variance, Journal of Geochemical Exploration 44, 23 - 36. Ramsey, M. H., 1993. Sampling and analytical quality control ( S A X ) j o r improved error estimation in the measurement of heavy metals in the environment, using rohust analysis of variance, Applied Geochemistry 2, 149- 153. Rousseeuw, P. J., Leroy, A. M., 1987, Rohust regression and outlier detection, New York: Wiley. Sachs, L., 1974, Angewandte Statistik, Berlin : Springer Verlag. Sharaf, M. A,, Illman, D. L., Kowalski, B. R., 1986, Chemometrics, New York: Wiley. Shaw, R. G., Mitchell-Olds, T., 1993, A N O V A .for unbalanced data: an overview, Ecology 74 (6), 1638- 1645. Skalski, J. R., Robson, D. S . , 1979, Tests of homogeneity and goodness-offit to a truncated geometric model ,for removal sampling, in: Cormack, R., Patil, G. P., Robson, D. S., Sampling biological populations, Burtonsville, Maryland : International Co-operative Publishing House. Slocomb, J., Stauffer, B., Dickson, K. L., 1977, On fitting the truncated lognormal distribution to species-ahundance data using maximum likelihood estinmtion, Ecology 58, 693 - 696. Snedecor, G . W., Cochran, W. G., 1989, Statistical method.s, Ames: Iowa State University. Sokal, R . R., Rohlf, F. J . , 1981, Biometry, San Francisco:W. H. Freeman. Sprent, P. 1989, Applied nonpurumetric statistical methods, London : Chapman and Hall. Taylor, J. K., 1988, Defining the accurucy, precision and confi:dence limits of sample data, in: Keith, L. H. (ed.), Principles ofenvironmental sampling, ACS Professional Reference Book, Washington D.C.: American Chemical Society. Taylor, J. K., 1987, Quality assurance of chemical measurements, Chelsea: Lewis Publishers. Trexler, J. C., Travis, J., 1993, Nontruditionul regression analyses, Ecology 74 (6), 1629- 1637. Weber, E., 1986, Grundrijl der biologischen Statistik? Stuttgart: Gustav Fischer Verlag. Weisberg, S., 1985, Applied linear wgression, New York: Wiley. Zar, J . H.. 1984, Biostatisricul analysis, New Jersey: Prentice-Hall Inc., Englewood Cliffs. ~

498

S. Hannappel

21.2.3 Air Allegrini, I., De Santis, F., Di Palo, V., Febo, A . , Perrino, D., Possanzini, M., 1987, Annzrlur rlenudw metliod,for sunzpling reactive gases and aerosols in the atmosphere, Sci. Total Environ. 67, 1 . Appel, B. R., 1993, Atmosphericsampk unalysis andsampling artifucts, in: Willeke, K . , Baron, P. A . (eds.), Aerosol measurement: Principlcs, tcchniques and applications, Ncw York : Van Nostrand Reinhold. Berger, K., 1990, Untersuchungen zur Probenuhme von iikotoxischen Metullen und Halhmetullerz aus Rauchgasen unter hesonderev Berucksichtigung des,filtergingigen Anteils, Dissertation, Hamburg: Universitit Hamburg. Berlin, A.. Brown, R. H., Saunders, K . J., 1987. DII@sive .rumpling: an alternatiue approach to ivarkplice air monitoring. London: Royal Society of Chemistry. Bcrner, A.. 1976a. Z u m Problem der Sc,kundurnied~rschliigein Ka.skadenimpuktciren. in : Gesellschaft fur Aerosolforschung (ed.), Acrosolc in Medizin. Forschung und Technik. n ~ e n einfachrr und Berner, A., 1 9 7 6 ~ . Zur Theorie der Messung lion A e r o ~ s o l g r ~ ~ ~ e n v e r t e i l uniittel.s uielfaclier Ku.Fkuclenimpaktoren - Teil II, Staub-Reinhalt. Luft 36, 417-419. Cheremisinoff. P. N., Morresi, A . C.. Air pollution .sampling & analysis deskbook, Ann Arbor: Ann Arbor Science Publishcrs Inc.. Clements, J . B., Lcwis, R. G., 1988, Sumpliizg,fi)r organic compounds, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society. Coffman, M. A., Singh, J., 1991, Sumpling and analysis qf gases and vapors, Report, Ordcr N o . PB92- 1146636, Clayton Environmental Consultants, Novi, MI, USA. Dasch, J. M., Cadle, S. E., Kennedy, K. G., Mulawa, P. A , , 1989, Comparison qf annular denuders und.filter pucks .for atnzospheric sampling, Atmos. Environ. 23, 2775. Duetsch, H. R., 1992. Measuretnent o f nitrogen dio.uide emissions with pussive collectors, Prax. Naturwiss., Chem. 41 (3), 21 -4. Eatough, N. L., McGregor, S., Lewis. E. A., Eatough, D. J., 1988, Comparison U f s i u dt.nir&r methods and u.filter puckjbr the collection qf' ambient HNO,,,,, HNO,(,,, and SO,,,, in the 1985 N S M C study. Atmos. Environ. 22, 1601. Evans, J . S., Ryan, P. B., 1983, Stutistical uiic~ertaintiesin cierosol muss conciwtralions mrusuri~tl hy iiirtual impactors, Aerosol Sci. Techno. 2, 531 -536. Fujita, E. M., Collins, J. F., 1989, Quality ussurunce ,for the Southern Culiforniu uir quulity study, 82nd Annual Meeting in Anaheim, Pittsburg: Air & Waste Management Association. Gaind, V. S., Wu, W. S. S, Chai, F, 1992, Chromutogruphic determination qf low I t w l s af' chlorine in uir, J. High Resolut. Chromatogr. 15 (12), 840-2. Grimmer, G., Glaser, A., Schncidcr, D., 1984, Longterrn storuge (fuirpollutunts - sampling systems for atmospheric purticulate matter and volatile compounds for a long-term storuge in un muironmrntulspec,inzenhank, Biochemical Institute for Environmental Carcinogens, Ahrensburg, F.R.G. Grosjcan, D., Williams, E. L.. 11, 1992, A passive .sampler ,fiw ,fiirmakkhyde, Atmos. Environ., Part A 26A (16), 2923-8. Hangal, S., Willeke, K., 1992, Aerosol sampling at small ,forwurd:facing angles: d$ferentiarion q f y a w f r u m pitch, Atmos. Environ., Part A 26A (16), 2913-21. Harrison, R. M., Kitto, A,-M. N., 1990, Field intercomparison ofjilter pack und denuder .sampling methods for reactive gaseous and purticulate pollutants. Atmos. Environ. 24A. 2633. Hicks, B. B., Meyers, T. P., Baldocchi, D. D., 1988, Aerometric measurement requireinents ,fbr quunt$king dry depusition, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society. Jacob, J., Grimmer, G., 1992, Air,filter samples - necessaryfor environmental specimen banking.?, in: Rossbach, M., Schladot J. D., Ostapczuk P. (eds.), Specimen Banking, Berlin, Heidelberg, New York: Springer Vcrlag. Katz, M. (ed.), Methods ( f a i r sampling and unalysis, 2nd rd., Washington D.C. : American Public Health Association. Klockow, D.. 1987, Z u m gegenwiirtigen Stand i k r Prohenahme von Spurenstqjyen in der f r r i t v Atmosphiire, Fresenius J. Anal. Chem. 326, 5.

Overview of References ,for Sampling and Related Topics

499

Koutrakis, P., Wolfson, J. M., Slater, J. L., Brauer, M., Spangler, J. D., Stevens, R. K., Stone, C . L., 1988, Ewluation of an annular denuder/filter pack system to collect acidic aerosols and gases, Environ. Sci., Technol. 22, 1463. Koutrakis, P., Wolfson, H. M., Bunyaviroch, A,, Froehlich, S. E., Hirano, K., Mulik, J. D., 1993, Measurement of umhient ozone using a i?itrite-coatedfilter,Anal. Chem. 65 (3), 209- 14 Lewin, E. E., Hansen, K. A,. 1984, Diirusion denuder assembly ,for collection and determination ofgase.s in air, Anal. Chem. 56, 842. Lodge, J . P., 1989, Methods of air sumpling and analysis, Chelsea Michigan: Lewis Publishers. Lund, W., Starkey, R., 1992, Sampling in the air, Chemtech 22 (Il), 648-54. Miller, C . E., Lewis, R. H., 1992, The reproducibility of',field meusurements qf urban air pollution: a pragmatic approach, Environ. Technol. 13 (1 I ) , 1053-60. Niessner, R., Malejczyk, M.. Schilling, M., Klockow, D., 1987, Die Diffusion als Prohenahmeprinzip zur GuslPartikel - Trennung. VDI - Berichte 608, 153. Perez Ballesta, P., Gonzalez Ferradas, E., Manana Aznar, A., 1992, Simultaneous passive sampling of volatile organic compounds. Chemosphere 25 (12), 1797-809. Peters, J. A , , 1988, Quality control infusion into stationary source sampling, in: Keith, L. H. (ed.), Principles of environmental sampling. ACS Profcssional Reference Book, Washington D.C.: American Chemical Society. Schwikowski, M., Naumann, K.. Dannecker, W., 1988, Determination of nitric acid andparticulate nitrate in the marine atmosphere employing a diijiusion denuder and a,filter pack. J . Aerosol Sci. 19, 131 1 . Shaw Jr, R. W., Stevens, R. K., Bowermaster, J., Tesch, J . W., Taw, E., (1982). Measurements of atmospheric nitrate and nitric acid: the denuder diflhrence experiment, Atmos. Environ. 16, 845. Sickles, J. E., Perrino, C., Allegrini, I., Febo, A,, Possanzini, M., Paur, R. J., 1988, Sampling and analysis of ambient air near Los Angeles using an annular denuder system, Atmos. Environ. 22, 1619. Tanner, R . L., 1988, Airborne sampling and in situ measurements of atmospheric chemical species, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C.: American Chemical Society. Vincent, J. H., 1989, Aerosol sampling: science and practice, London: Wiley. Watson, J. G., Chow, J. C.. 1993. Ambient air sampling, in: Willeke, K., Baron, P. (eds.), Aerosol measurement: Principles, techniques and applications, New York: Van Nostrand Reinhold. Wells, W. H., 1986, Evaluation of the coadsorption effect of water vapor and organic. vapors on charcoal sampling tubes, Ph.D. Thesis, Ann Arbor: UMI. Zhang, X., McMurry, P. H., 1992, Evaporative losses of fine particulate nitrates during sampling, Atmos. Environ., Part A 26A (18). 3305- 12.

27.2.4 Water Adams, W. J., Hoogheem. T. J., Michael, P. R., 1986, Aquatic monitoring: A rationalefor obtaining and interpreting aquatic ecosystem chemical exposure data, in: Isom, B. G. (ed.), Rationale for sampling and interpretation of ecological data in the assessment of freshwater ecosystems, ASTM STP 894, 3-21. American Public Health Association, 1985, Standard methods ,for the examination of water and wastewater, Washington D.C.: American Public Health Association. Baykut, G., Vetters, H. P., Schiller, A., Taddigs, R., 1992, Rechnergesteuerte Probennuhme , f i r fliichtige organische Bestandteile in Wusserproben, Wasser, Luft und Boden 36 (6), 30 - 3 1. Beck, M . B. (ed.), 1981, Uncertainty analysis of aquatic ecosystems, Berlin : Springer Verlag. Beck, M. B., Van Straten, G., 1983, Uncertainty and,forecusting of water quality, Berlin: Springer Verlag. Bothner, M. H., Robertson, D. E., 1975, Mercury contamination of seawater samples stored in polyethylene containers, Anal. Chem. 47, 592- 595.

500

S . Hannuppel

Brandl, H., Hanselmann, K. W., 1991 Evaluation and application of dialysis p o r e w t e r sump1er.s,for microbiological studies and sediment-water interfaces, Aquatic Sciences 53 (1 ), 55 - 73. Bredemeier, M., Lamersdorf. N., Wiedey, G. A., 1990, A new mobile und easy to handle suction Iq’simeterJor .soil water sampling, Fresenius J. Anal. Chem. 336 (I), 1-4. Briigmann, L., Geyer. E., Kay, R., 1987, A new teflon .sampler f o r trace metal studies in seawuter - W A T E S ” , Marine Chem. 21, 91 -99. Bruland, K. W., Franks. R. P., Knaucr, G. A,, Martin. J. H., 1979, Sumpling undanalytical~netliods for the determination qf copper, cadmium, zinc, and nickel at the nanogram per liter level in scu water, Anal. Chem. Acta 105, 233 - 245. Brusske, A., Willemsen, H. G., 1990, Rheinkurzzeitiiherwachung in N R W-Konzepr und erste Ergehnisse, Vom Wasser, 74. Band, Weinheim: VCH Verlagsgesellschaft mbH. Burn, D. H., 1991, Water quality sampling f o r nutrient loading estimation, in: Wrobcl, L. C . , Brebbia, C. A., Water Pollution. Modelling, Measuring and Prediction, Southampton: Computational Mechanics Publications. Cairns, J., Jr., Pratt, J. R., 1986, Developing a sampling strategy, in: Isom, B. G. (ed.), Rationale for sampling and interpretation of ecological data in the assessment of freshwater ecosystems, ASTM STP 894, 168 - 86. Canter, L W., 1985, River water quality monitoring, Chelsea: Lewis Publishers. Collins, H. J., Miinnich K., 1991, Reprusentanz von Wasserproben aus Grundwassc,rleitern, Wasser, Abwasser 132 (lo), 546-550. Cowgill, U . M., 1988, Sampling water: the impuct of sample vuriability on planning und confidence levels, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C.: American Chemical Society. Cox, A. G., McLeod, C . W., 1992, Fieldsumpling technique for speciation qf inorganic chromium in rivers, Microchim. Acta 109 (1 -4), 161 -4. De Boer, J. L. M.. Fortezza, F., 1992, The retention of trace elements on,funnel surfiices of sampling equipmentfbr wet deposition, Water, Air, Soil, Pollut. 64 (3 -4). 467-74. Desilets, L., 1988, Criteria f o r basin selection and sampling station macrolocution, Scientific Series, Inland Water Directorate, Ottawa: Ministry of Supply and Services Canada. Deutscher Verband fur Wasserwirtschaft und Kulturbau e. V., 1979, Empfehhmg zu Umfang, lnhalt und Genauigkcitsanf?forderungenhei chemischen Grundu,asseruntersuchungen, DVWK-Regeln zur Wasserwirschaft, 11 I , Hamburg: Verlag Parey. Deutscher Verband fur Wasserwirtschaft und Kulturbau e. V., 1982, Entnuhme von Prohen f i r hydrogeologische Grundwusseruntersuchungen, Merkblatter zur Wasserwirtschaft, 203, Hamburg: Verlag Parey. Deutscher Verband fur Wasserwirtschaft und Kulturbau e. V., 1990, EinJiisse von Mestellenaushau und Pumpenmaterialien auf die Beschaflenheit einer Wasserprohe, DV WK-Mitteilung, 20, Bonn. Fitzgerald, W. F., Lyons, W. B., 1975, Mercury concentrations in open ocean waters: sampling procedures, Limnol. Oceanogr. 20, 468 - 47 1. Friege, H., 1987, Monitoring qf the river Rhein - experience guthered,from accidental events in 1986, in : Angeletti, Bjorseth, Organic micropollutants in the aquatic environment, Commission of the European Community, Proceedings, Dordrecht : Kluwer Academic Publishers. Gassen, M., Woffen, B., 1978, Zur Huufigkeit der Probenahrne undder Beurteilung der Leistungsfahigkeit van Klaranlagen. gwf - Wasser/Abwasser 119, 455. Acta hydrochim. hydroGnauck, A,. 1990, ~ahrschein1ic.hkeitsverteilungvan Wasserinhaltsto~~en, biol. 18 (I), 119- 132. Gortz, W., Grubert, G., 1986, Prohleme bei der Entnahme von Mischproben f u r die Untersuchung von Wasser und Ahwasser, Gewasserschutz - Wasser - Abwasser 86, 21. Gudernatsch, H., 1983, Prohenahme und Proheaujbereitung von Wassern, in: Bock, R., Fresenius. W., Giinzler, H., Huber, W., Tolg, G. (eds.), Analytiker Taschenbuch, Bd. 3, Berlin, Heidclberg, New York: Springer Verlag. Hansen, P. D., 1992, On-Line Monitoring mit Biosensoren am Gewasser zur ereignisgesteuerten Prohenahme, Journal for Water and Wastewater Research 20 (2), 92-95. Hellawell, J. M., 1986, Biological indicators of,freshwater pollution and environmental management, London, New York : Elsevier Applied Science Publishers. “

Overview of References f o r Sampling and Related Topics

50 I

Helsel, D. R., Hirsch, R. M., 1992, Statistical methods in water resources, Amsterdam: Elsevier. Heyman, U., Ryding, S.-O., Forsberg, C., 1984, Frequency distributions of water quality variables. Relationships between mean and maximum values, Water Res. 18, 787 - 794. Humpesch, U. H., Elliott, J. M. (eds.), 1990, Methods of biologicalsampling in a large deep river the Danube in Austria, Wien: Wasser und Abwasser. Isom, B. G., Gooch, Ch., 1986, Rationale and sampling designsfor,freshwater mussels Unionidae in streams, large rivers, impoundments, and lakes, in: Isom, B. G. (ed.), Rationale for sampling and interpretation of ecological data in the assessment of freshwater ecosystems, ASTM STP 894, 46-59. Kearl, R. M., Korte, N. E., Cronk, T. A., 1992, Suggested modifications for ground water sampling procrdures based on observations,from the colloidal borescope, Ground Water Monitoring Review, Spring, 155 - 160. Kent, R . T., Payne, K. E., 1988, Sampling groundbvater nionitoring wells: special quality assurance and quality control considerations, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society. Kovalak, W. P., Dennis, S. D., Bates, J . M., 1986, Sampling ejrort required to ,find rare species qffreshwater mussels, in: Isom, B. G. (ed.), Rationale for sampling and interpretation of ecological data in the assessment of freshwater ecosystems, ASTM STP 894, 34-45. Krajca, J. M., 1989, Water sampling, Chichester: Horwood. Krhtz, T. K., Magnuson, J. J., Bowser, C. B., Frost, T. M., 1986, Rationalefor data collection and interpretation in the northern lakes long-term ecological research program, in : Isom, B. G. (ed.), Rationale for sampling and interpretation of ecological data in the assessment of freshwater ecosystems, ASTM STP 894, 22-33. Kretzer, H., 1992, Eine Methode zur tiefenorientierten Beprobung von Grund- und Bodenwasser, Wasser und Boden 44 (2), 77-78. Kritzner, W., 1992, Vecfahren und Techniken zur Entnahme reprasentativer tiefenorientierter Grundwasserproben, Wasserwirtschaft 82 (I), I 3 - 17. Ligget, W. S., 1988, Assessment c!fmeasuretnent uncertainty: designs,for two heteroscedastic error components, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C.: American Chemical Society. Mart, L., 1979, Prevention of contamination and other accuracy risks in voltametric trace metal analysis cj’nutural waters, Fresenius J. Anal. Chem. 313, 200- 202. Maul, A,, El-Shaarawi, A. H., Block, J. C., 1990, Bacterial distribution and sampling strategiesfor drinking water networks, in : McFeters, G. A. (ed.), Drinking water microbiology. Progress and recent developments, New York, Berlin, Heidelberg: Springer Verlag. Maul, A,, El-Shaarawi, A. H., Block, J. C.. 1990, Bacterial distribution and sampling strategies,fiir drinking ivater networks, in: McFeters, G. A.(ed.), Drinking water microbiology, New York: Springer Verlag. McGuire, P. E., Lowery, B., 1992, Evaluation of several vacuum solution samplers in sand and silt loam at several water potentials, Ground Water Monit. Rev. 12 (4), 1 5 1- 60. Metzger, I., 1986, Interpretation o f i n situ groundwater quality f r o m well samples, in: Isom, B. G. (ed.), Rationale for sampling and interpretation of ecological data in the assessment of freshwater ecosystems, ASTM STP 894, 76-85. Nash, R. G., Leslie, A. R. (eds.), 1992, Groundwater residue sampling design, ACS Symposium Series No. 465, Washington D.C. : American Chemical Society. Newburn, L. H., 1988, Modern sampling equipment: design and application, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society. Norris, J. E., 1988. Techniques f o r sampling surfirce and industrial water: special considerations and choices, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society. Office of Water Data Coordination, 1982, National handbook of recommendedmethodsfor wuter-data acquisition, Reston : U .S. Geological Survey. Parr, J., Bollinger M., Callaway, O., Carlberg, K., 1988, Preservation techniques ,for organic and inorganic compounds in water samples, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society.

502

S. Hannappel

Petzler, E. T., Brewer, P. G., 1993, Some practical aspects of measuring DOC - ~samplin~ urtjfuctx and analytical problems with marine sumple.s, Mar. Chem. 41 (1 - 3), 243 - 52. Pohlmann, K. F., Hess, J. W., 1988, Generalized ground water sampling device matrix, Ground Water Monitoring Review, Fall, 82 - 84. Powell, R. M., Puls, R. W., 1993, Passive sampling ofground-water monitoring wells withoutpurging: multilevel well chemistry and tracer disappearance, J. Contam. Hydrol. 12, 51 - 71. Puls, R. W., Clark, D. A,, Bledsoe B., Powell, R. M., Paul C. J., 1992, Metals in groundwater: sampling artifircts and reproducibility, Hazard. Waste Hazard. Mater. 9 (2). I49 - 62. Puls, R. W.. Powell, R. M., 1992, Representative ground water quality sarnples,for metals, Ground Water Monitoring Review, Summer, 167- 176. Pyatt, F. B., Beaumont, E. H., 1990, A simple technique t o anal,vse the vertical distribution ofplankton in ,freshwa!er systems, Environmental Education and Information 9 (I),48 - 50. Raynor, G. S., Hayes, J. V., 1983, Sampling and analysis of rain. in: Campbell, S . A. (ed.), Philadelphia: American Society for Testing and Materials. Remmler. F., 1992, Reprusenfativifat.T[~erlu.~te von Grundwasserprohen durch Mriteriul~~infliiss, Wasser und Boden (121). 15-16, Ronen, D., Magaritz, M., Levy, I., 1987, A n in situ multivlevel sampler j b r preventive monitoring and study of h,ydrochemical profiles in aquifirs, Ground Water Monitoring Review 7 (4), 69 - 74. Sanders, T., Ward, R., Loftis, J, Steele, T., 1973, Design of networks,for monitoring water quality, Littleton: Water Resources Publications. Selent, K., 1988, Die amtliche AbwcisserprobenahmL, Probenahmeschulung Wasser und Abizwser., Dusseldorf: Landesamt fur Wasser und Abfall Nordrhein-Westfalen. Smith, S. Jr., 1993, Pesticide retention by a programmable automutic M~uterlsuspendt~d .sediment sutpler, Bull. Environ. Contam. Toxicol. 50 ( l ) , 1-7. Smith, J. S., Steele, D. P., Malley, M. J., Bryant, M. A,, 1988, Groundwater sampling, in: Keith, L. H . (ed.), Principles of environmental sampling. ACS Professional Reference Book, Washington D.C.: American Chemical Society. Snelting, H., 1979, Mini-screens sumpling .systenis, Quarterly Report, 16, Leidschendam: Nal. Inst. Water Supply/Netherlands. Stednick, John D., 1991, Wildland water quulity sumpling und analysis, San Diego: Academic Press. Stevenson R. J., Lowe, R. L., 1986, Sampling arid interpretation nf algal pntterns,for water quality assessments, in: Isom, B. G . (ed.), Rationale for sampling and interpretation of ecological data in the assessment of freshwater ecosystems, ASTM STP 894, 118- 149. Stock, H. D., Grubert, G., Te Heesen, D., Selent, K.. 1990, Der selbstentleerende Ruckstellprobenehmer - neue Chancen der Abu,assrruberwachung, Gewiisserschutz - Wasser - Abwasser 118, 285. Suess, M. J., 1982, Examination of water for pollution control, Vol. 1, Oxford: Pergamon Press. Suffet, 1. H., Suffel. M., Malaiyandi, M., 1987, Organic pollutants in water: sampling, analysis and toxicity tesring, 188th Meeting of the American Chemical Society, Washington D.C. : American Chemical Society. Thornton, K. W., Kennedy, R. H., Magoun, A. D.. Saul, G . E., 1982, Reservoir water yuulity sampling design, Water Resour. Bull. 18, 471 -480. Toussaint. B., 1989, Anforderungen an den Bau von GrundH~assermepstellennus I~ydrogeologischr Sicht, Oberrhein. Geol. Abh. 35, 1 1 I - 128. Watts, C. D., Moore, K.. 1987, Fate and transport of organic compounds in rivers, in: Angeletti, Bjorseth. Organic micropollutants in the aquatic environment, Commission of the European Community, Proceedings, Dordrecht: Kluwer Academic Publishers. Wehrens, R., van Hoof, P., Buydens, L. Kateman, G., Vossen, M., Mulder, W., Bakker, T., 1992, Sampling of aquatic sediments. Design o j a decision-support system and a case study, Anal. Chim. Acta 271 ( I ) , 11 -24. Wilson, A. L., 1982, Design of sumpling programmes, in: Suess, M. H. (ed.), Examinalion of water for pollution control. A reference handbook, Oxford : Pergamon Press. Ziegler, H., I99 I , Sachgerechte Prohenahme zur esakten Kontrolle der Grund,vas.serclualitut. Wasscr, Luft und Boden 35 (6), 38.

Overoiew o f References f o r Sampling and Related Topics

503

27.2.5 Soils, Sediments, Sludges, Rocks and Mining Aichberger, K., Eibelhuber, A., Hofer, G., 1986, Soil sampling,for trace element analysis and its statistical evaluation, in: Gomez, A., Lescbber, R., L’Hermite, P. (eds.), Sampling problems for the chemical analysis of sludge, soils and phnts, London, New York: Elsevier Applied Science Publishers. Bodenschutzzentrum des Landes Nordrhein- Westfalen I 99 I , Mindestdatensatz Bodenuntersuchungen, AbschluRbericht des Arbeitskreises Mindesdatensatz Bodenuntersuchungen der Sonderarbeitsgruppe Informationsgrundlagen Bodenschutz. Burgess, T. M., Webster, R., 1980a, Optimal interpolation and isarithmic mapping qf soilproperties, I - The semi-variogram and punctual kriging, Journal of Soil Science 31, 3 15- 33 1. Burgess, T. M., Webstcr, R.. 1980b, Optimal interpolation and isarithmic mapping qfsoilproperties, II- Block kriging, Journal of Soil Science 32, 643 - 659. Burn, T., Lopez, C., 1986, Some applications q f regionalized variables theory to soil sumpling problems, in: Gomez. A., Leschber, R., L’Hermite, P. (eds.), Sampling problems for the chemical analysis of sludge, soils and plants, London, New York: Elsevier Applied Science Publishers. Cameron, D. R., Nyborg, M, Toogood, J. A , , Laverty, D. H., 1971, Accuracy of,field samplingfor soil tests. Can. J. Soil Sci. SL, 165-175. Chen, 2. S.. 1988, A survey cfheavy metal concentration qfsoils and rice grain in northern Taiwan, Taiwan: Project report of Council of Angriculture. Chen, Z. S., Lee, D. Y., Huang, D. N., Liau, S. H.. 1992, The relationships between heavy metals in soils, waters, and .teditnents: case study in Shiang-Shun tndustrial park, Taiwan : Project reports of EPA-ROC. Chen, Z. S., Lee, D. Y.,1992, The relationships between heavy metals in soils, waters, and sediments: ‘ case study in An-Nan Industrial park, Taiwan: Project reports of EPA-ROC. Colin, F., 1986, Application qf data analysis techniques to sludge and soil sampling operations, in: Gomez, A , , Leschber, R., L’Hermite, P. (eds.) Sampling problems for the chemical analysis of sludge, soils and plants, London, New York: Elsevier Applied Science Publishers. EPA, 1990, E P A specificutions andguidance,for obtaining sontaminantTfree sample containers, Seattle: EPA Region 10. Feichtinger, H.-K., 1980, Probenahme aus ,flussigen Metallen zum Zwecke der Gnsanalyse, Probenahme - Theorie und Praxis, Schriftenreihe der Gesellschaft Deutscher Metallhiitten- und Bergleute, Nr. 36, Weinheim, Deerfield Beach, Basel: Verlag Chemie. Frinzle, 0..Kuhnt. G., 1983, Regionalreprasentative Auswalzlder Boden f u r eine Umweltprobenhank, FE-Vorhaben 106 05 028 des Umweltbundesamtes, Berlin: Umweltbundesamt. Franzle, 0.. 1990, Representative sampling of soils in the Federal Republic of Germany and the EC Countries, in: Lieth, H., Markert. B., Element concentration cadasters in ecosystems, Weinheim: VCH Verlagsgesellschaft mbH. Gomez, A , , 1986, Sampling techniques ,fur sludge, soils and plants, in: Gomez, A,, Leschber, R., L’Hermite, P. (eds.), Sampling problems for the chemical analysis of sludge, soils and plants, London, New York: Elsevier Applied Science Publishers. Hbkanson, L., 1982, A modifiedpolar grab samplerfor course and consolidaled sediments, Uppsala: National Swedish Environmental Protection Board. Herkelmann, H., Koppe, P., 1986, Experiences with sludge sampling in the Ruhr River basin, in: Gomez, A., Leschber, R., L’Hermite, P. (eds.), Sampling problems for the chemical analysis of sludge, soils and plants, London, New York: Elsevier Applied Science Publishers. Holstein, F., 1980, Probenahme von NE-Konzentruten, (Beispiel: Kupferkanzentrate), Probnahme - Theorie und Praxis, Schriftenreihe der Gesellschaft Deutscher Metallhiitten- und Bergleute, Nr. 36, Weinheim, Deerfield Beach, Basel: Verlag Chemie. Horowitz, A. J., 1991. A primer on sediment - trace element chemistry, Chelsea: Lewis Publishers. Kraft, G., 1980, Theoretische Grundlagen der Probenahme, Probenahme - Theorie und Praxis, Schriftenreihe der Gesellschaft Deutscher Metallhiitten- und Bergleute, Nr. 36, Weinheim, Deerfield Beach, Basel: Verlag Chemie. Mather. P. M., 1972. Spatial analysis in geomorpholagy, London: Methuen.

504

S. Hannappel

Mawhinney, M. R., Bisutti, C., 1987, Common corers and grab samplers: operating manual, Burlington: National Water Institute, Environment Canada. Merks, J. W., 1985, Sampling und weighing qf bulk solids, Clausthal - Zellerfeld: Trans. Tech. Publications. Mori, H., 1981, International manual f o r the sampling ofsoft cohesive soils, Tokyo: Tokai University Press. Murdoch, A., Macknight, S. D., 1991, C R C Handbook of techniquesfor ayuatic sediment sampling, Boca Raton: C R C Press. Nothbaum, N., Scholz, R. W., 1991, Probenplanung und Datenanulyse bei kontaminierten Biiden, Projektbericht, T U Z 107 03 007, Berlin: T U Berlin. Oortner, H.. 1980, Uber die Problematik der Probenuhnle hei Sondermetallen. Probenahme Theorie und Praxis, Schriftenreihe der Gesellschaft Deutscher Metallhiitten- und Bergleute. Nr. 36, Weinheim, Deerfield Beach, Basel: Verlag Chemie. Pottkamp, F., 1980, Probenahme von Elei und Zink, Probenahme - Theorie und Praxis, Schriftenreihe der Gesellschaft Deutscher Metallhiitten- und Bergleute, N r . 36, Weinheim, Deerfield Beach. Basel: Verlag Chemie. Reynolds, S. G., 197.5. Soil property variahility in slope studies: suggested stimpling xheme.Y and typical required sample size, Z . Geomorph. N . F. 19, 191 -208. Ristenpart, E., Gitzel, R., Uhl, M., 1992, Examination qf sediment samplers, Watcr Sci. Technol. 25 (8), 63-9. Rhodc, C. A., 1979, Bafch, Bulk, and composite sampling. in: Cormack, R., Patil, G. P., Robson, D. S., Sampling biological populations, Burtonsville, Maryland: International Co-operative Publishing House. Saxena, S. K., 1985, Updating suhsurjace sampling qf soils und rocks and their in situ tcsting, Confcrence January 1982, New York : Engineering Foundation. Smith, R., James, G. V., 1981, The sampling qf bulk materials, Analytical Science Monographs, Vol. 8, Dorchester: Dorset Press. Soil Sampling International Society for Soil Mechanics and Foundation Engineering, 198 I , International manual f o r the sampling of soft cohesive soils, Tokyo: Tokai University Press. Starr, J. L., Parkin, T. B., Meisinger, J. J., 1992, Sample size consideration in the determination of’ soil nitrufe, Soil Sci. Soc. Am. .I.56 (6), 1824-30. Stein, A,, Wopereis, M. C. S., 1992, Sumpling strategies and geostatisticul techniques, in: Manual for soil hydraulic measurements, field monitoring techniques and sampling strategies for rice based cropping systems, The Winand Staring Centre for Integrated Land, Soil and Water Research. Urban, M. J.. Smith, J . S., Schultz, E. K., Dickinson, R. K., 1992. Volatile organic sample preservation f o r ci soil, sedimenr or waste, ASTM, STP 1075, I70 - 8. Webster, R., Brugess, T. M., 1984, Sumpling and bulking strategies ,fbr estimuting soil properties in small regions, Journal of Soil Science 35. 127- 140.

27.2.6 Biota Ahl. W.. Wcber, A,, 1981, A simple monitoring technique to determine the heavy metal load if’ algae in crquutic ecosystems, Environ. Technol. Letters 2, 3 17 - 322. Albert, R., Horwitz, W., 1988, Coping with sampling variability in biota: percentiles and other strategies. in : Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C.: American Chemical Society. Benton Jones, J., Large, R. L., Pfleiderer, D. B., Klosky, H. S., 1971, How toproperl~isamnplr.,f~~r a plant analysis, Corps Soils 23, 15 - 18. Bourke, J . B., Spittler, T. D., Young, S. J., 1988, Sample size: relation to analytical and yuality ussurance und quality control requirements, in : Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society.

Overview of References f o r sampling and Related Topics

505

Brian, M. V., 1953, Speciesfreyuencies in randoni sample.s,fiom animal populations, J. Anim. Ecol. 22, 51-64.

Brown, D. H., Brown, R. M.. 1990, Reproducibility of sampling ,for element analysis using Bryophytes, in: Lieth, H., Markert, B., Element concentration cadasters in ecosystems, Weinheim: VCH Verlagsgesellschaft mbH. Camerlynck, R., Kiekens, L., 1986, Problems encountered with sampling of plants f o r I C P trace element analysis, in: Gomez, A,, Leschber, R., L'Hermite, P. (eds.), Sampling problems for the chemical analysis of sludge, soils and plants, London, New York: Elsevier Applied Science Publishers. Caroli, S., 1992, The accuracy syndrome in trace element analysis of biological samples, Microchim. J. 45, 257-271. Connor, E. F., McCoy, E. D., 1979, The statistics and biology q f t h e species-area relationship, Am. Nat. 113, 791-833. Cornelis, R., 1992, Quality control in truce element analysis of clinical and biological samples. How good are your data?, J. Tr. El. Electr. Hlth. Dis. 6, 129 - 135. Duijsings, J. J. H. M., Verstraten, J. M., Bouten, W., 1986, Spatial variability in nutrient deposition under an oaklheech canopj', Zeitschrift fur Pflanzenernahrung und Bodenkunde 149, 718 - 721. Ellenberg, H., Mayer, R., Schauermann, J., 1986, iikosystemforschung, Ergebnisse des Sollingprojekts 1966 - 1986, Stuttgart: Ulmer Verlag. Elliot, J. M., 1977, Some methods ,for the statistical unalysis of benthic invertebrates, Freshwater Biological Association Scientific Publication 25, Ambleside: The Ferry House, England. Elliot, J. M., Tullette. P. A,, 1978, A bibliography ofsamplers,for benthic invertebrates, Ableside: Occ. Publ. Freshwater Biology Assoc.. Elliot, J. M., Tullette, P. A,, 1983, A supplement to a bibliography ofsamplers,for benthic invertebrates, Ambleside: Occ. Publ. Freshwater Biology Assoc. Emst, W., 1975. Variation in the mineral contents of leaves o trees in Miombo Woodland in South Central Africa, J. Ecol. 63, 801 -808. Ernst, W. H. 0.. 1990, Element allocation und (re)translocation in plants and its impact on repre.s~~ntative.samp/ing, in: Lieth, H., Markert, B., Element concentration cadasters in ecosystems, Weinheim: VCH Verlagsgesellschaft GmbH. Forster, H., Schimmack, W., Kreutzer, K. E.. 1991, Die horizontale Verteilung uon Rndiocasium im Waldboden unter Fichte und Buche, Zeitschrift fur Pflanzenerniihrung und Bodenkunde 154, 87 - 92. Garner, F. C.. Stapanian, M. A,, Williams, L. R., 1988, Composite sampling f o r enuironmental monitoring, in: Kcith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C.: American Chemical Society. Hall, A,, 1981. Copper accumulation in copper-tolerant and non-tolerant populations of the marine ,fouling alga, Ectocarpus siliculosus (Dillw.) Lynbye. Bot. Mar. 24, 223 -228. Herzig, R., Urech. M., Liebendorfer, L, Ammann, K., Guecheve, M., Landolt, W., 1991, Lichens as biologicul indicators of air pollution in Switzerland :passive biomonitoring as a part of an integratedmeasuring system,for monitoring uirpollution, in: Lieth, H., Markert, B. (eds.), Element concentration cadasters in ecosystems, Weinheim : VCH Verlagsgesellschaft GmbH. Hertz, J., Angehrn-Bettinazzi, C., 1989, Ungenauigkeit von Schwermetallanalysen - ein Homogenitiitsproblem, Bulletin - Bodenkundliche Gesellschaft der Schweiz 13, 81 -92. Iyengar, G. V., 1982, Presampling factors in the elemental composition of biological systems, Anal. Chem. 54, 554A - 560A. Kempton, R. A,, Taylor, L. R., 1976, Models and statistics ,for species diversity, Nature 262, 818 - 820. Klink, A,, 1993, Die Begleitfauna auf Mytilus chilensis (HupP 1854) - Kulturen in Yaldad, Insel Chilot, Siidchile und ihre Dynanzik, Dissertation, Universitat Konstanz. Krivan, V., Schaldach. G., 1986. Untersuchungen zur Prohenahme und -vorbehandlung von Baumnadeln zur Elementunalyse, Fresenius J. Anal. Chem. 324, 158 - 167. Lewis, T., Taylor, L. R., 1967, Introduction to experimental ecology, London, New York: Academic Prcss.

506

S. Hannappel

Lieth, H., Markert, B., 1990, Element concentration cadasters in ecosystems. State of the art and plans f o r the further development of an international research program till 1990, in: Lieth, H., Markert, B. (eds.), Element concentration cadasters in ecosystems: methods and assessment, Weinheim: VCH Verlagsgesellschaft GmbH. Markert, B., 1988, The distribution of some chemical elements in dgferent herds of‘ Vacciniurn vitis idaea (red whortleberry) - a contribution to the representative sampling procedure qf‘ terrestriul plants f o r multi-element analysis, Angewandte Botanik 62, 343 - 353. Markert, B., Steinbeck. R., 1988, Some usperts of element distribution in Betulu albu, a contribution to rrJpresentativesampling of’ terrestrial plants,for multi-element analysis, Fresenius J. Anal. Chem. 331, 616-619. Markert, B., 1989, Distribution of chemical elements in Vaccinium myrtillus (blueberry) - hasic prohlems for representative sampling of plunts f o r multi-elemen f analysis in ecosystems, Fresenius J. Anal. Chem. 333. 11 - 14. Markert, B., 1989, Multi-element analysis in ecosystems: basic conditions,fbr representative sumpling of plant materials, Fresenius J . Anal. Chem. 335, 562- 565. Markert, B., Klausmeyer. N., 1990, Variations in the elemental composition of plants and computer aided sampling in ecosystems, Toxicological and Environmental Chemistry 25, 201 -212. Markert, B., 1991, Die repra.!entutive Prohenahme pflanzlicher Matrizes als Grundvoraussetzung f u r Stoff~u~betraclitungen in O k ~ ~ y s t e m ein: n , Riewenherm, S., Lieth, H . (ed.), Osnabriick: Verhandlungen der Gesellschaft fur Okologie, Band XIXjIII, 349-3361. McArthur, B., 1992, Dermal measurement and wipe sampling method,y: a review, Appl. Occup. Environ. Hyg. 7 (9), 599-606. Minderhoud, A., 1986, Problems of sampling soil fauna f o r terrestrial ecological studies, in : Gomez, A,, Leschber, R., L’Hermite, P. (eds.), Sampling problems for the chemical analysis of sludge, soils and plants, London, New York: Elsevier Applied Science Publishers. Paterson, C. C., Settle, D. M., 1976, The reduction of orders of magnitude errors in lead anal.vsis of’ natural waters, Fresenius J. Anal. Chem. 299, 97- 102. Patil, G. P., Taillie, C., 1977, Diversity as a concept and its implications f o r random communities, Bull. Int. Statist. 47, 497-515. Prepas, E. E., 1971, Sorwe stutistical method.7 f o r the design of’ experiments and analysis ofsumples, in: Downing, J. A., Rigler, F. H., A manual on methods for the assessment of secondary productivity in fresh waters, Oxford: Blackwell Scientific Publications. Rennie, P. J., 1966, A forest sumpling procedure ,for nutrient-uptake studies, Commun. For. Rev. 45, 119-228. Robson, D. S., 1979, Approximations to some mark-recapture sampling distrihutions, in: Cormack, R., Patil, G. P., Robson, D. S., Sampling biological populations, Burtonsville, Maryland; International Co-operative Publishing House. Sansoni, B., Iyemgar, V., 1978, Sumpling and sample preparation met had.^ .for the unalysiu of’ trace elements in biologicul materials, Forschungszentrum Jiilich, Jiil. Spez., 13. Spittler, T. D., Bourke, J. B., 1988, Consideration.sforpre.servingbioticsamples, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society. Tokeshi, M., 1993, Species abundance patterns and community structure, in: Begon, M., Fitter, A. H., Advances in Ecological Research, Vol. 24, London: Harcourt Brace & Company, Publishers. Vollenweider, R. A., 1969, Primary production in aquatic environments, A manual on methods for measuring, Oxford: Blackwell Scientific Publications. Wagner, G., 1987, Entwirklung einer Methode zur grojlraumigen Uberwnchung inittels stundardisierter Pappelhlattproben von Pyramidenpappeln (Populus nigra “Itulica”) am Beispiel von BIei, Cudmium und Zink, in: Stoeppler, M., Diirbeck, H. W. (eds.), Beitrkgc zur Umweltprobenbank, 5 , Jiil. Spez., 412. Wagner, G., 1990, Variuhility af element concentrations in tree leaves depending on sampling paramcters, in: Lieth, H., Markert, B., Element concentration cadasters in ecosystems, Weinheim: VCH Verlagsgesellschaft mbH.

Overview of References f o r Sampling and Related Topics

507

Warren, W. G., 1979, Trends in the sampling offorest populations, in: Cormack, R., Patil, G. P., Robson, D. S., Sampling biological populations, Burtonsville, Maryland: International Cooperative Publishing House. Zimmerman, R. D.. Plankenhorn, W. E., 1986, Methodik der Blattprobenentnahme an der Rotbuche unter emissions-iikologischem Aspekt, Allg. Forstz. 41, 33 - 35.

27.2.7 Waste, Sewage, Sludges Bone, L. I., 1988, Preservation Techniquesfor samples qf solids, sludges and nonaqueous liquids, in: Keith, L. H. (ed.), Principles of environmental sampling. ACS Professional Reference Book, Washington D.C.: American Chemical Society. Bohnke, B., 1990, Analytik von Feststojfen - Abfhll - Altlasten (Prohenahme - Prohenvorhereitung - Bewertung), Aachen : Gesellschaft zur Forderung der Siedlungswasserwirtschaf~an der RWTH. Gomez, A,, 1984, Utilisation des houes de station d’kpurution en agriculture: problPmes posks par IMchantillonnage pour la determination des Plkments-traces, Dordrecht, Boston, Lancaster : Reidel Publ. Co.. Harkness, N., 1986, Sampling and analysis of sludges and soils in England and Wales f o r the munagement qf agrirultural utilisation ofsewage sludge, in: Gomez, A,, Leschber, R., L’Hermite, P. (eds.), Sampling problems for the chemical analysis of sludge, soils and plants, London, New York: Elsevier Applied Science Publishers. Huesemann, M. H., Guidelines .for the development of etfective statistical soil sampling strategies j b r environniental applications, Houston : Shell Development Company. Jackson, L. P., 1988, Sampling and analysis of hazardous and industrial wastes: special quality assurance and quality control considerations, in : Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society. Landesamt fur Wasser und Abfall Nordrhein-Westfalen, 1989, Prohenahme bei Altlasten, Diisseldorf: Landesamt fur Wasser und Abfall Nordrhein-Westfalen. Petersen, T. A., 1983, Link actuated sampler ,for sampling lagoon sludges, Industrial Wastes, Sept.-Oct., 16- 17. Schulze, G., Fronia, J., Schumann, H., 1991, Autqmatische Abwawrprobenahme - Grenzen, Trends, Aushlick, Teil 1: Uberblick, TUE Technische Uberwachung 32 ( 9 , 200 - 204. Schulze, G., Fronia, J., Schumann, H., 199I , Autornatische Abwasserprobenahme - Grenzen, Trends, Ausblick, Stand, Miiglichkeiten, Korrespondenz Abwasser 38 (4), 499 - 509. Smith, W., 1979, A n oil spill sampling strategy, in: Cormack. R., Patil, G. P., Robson, D. S., Sampling biological populations, Burtonsville, Maryland: International Co-operative Publishing House. Triegel, E. K., 1988, Sampling variability in soils and solid wastes, in: Keith, L. H. (ed.), Principles of environmental sampling, ACS Professional Reference Book, Washington D.C. : American Chemical Society.

27.3 Norms by the International Organization of Standardization in Geneva 27.3.1 Air I S 0 42 19, 1979,

I S 0 TR 1108, 1983, I S 0 8760, 1990,

Air quality - Determination of gaseous sulphur compounds in ambient air - sampling equipment. Air quality - Particle sizefraction definitionsfor health-relatedsampling. Work-place air - Determination of mass concentration ofcarhon monoxide - Method using detector tubesfor short-term sampling with direct indication.

508

S. Hannuppel

I S 0 8761, 1989, I S 0 9359. 1989,

Work-place air - Determination of mass concentration q f nitrogen dioxide - Method using detector tubes ,for short-term sampling with dircct indication. Air quality - Stratified sampling method ,for assessment of ambient uir quality.

27.3.2 Water I S 0 5667, 1991,

Water quality sampling, Part I , Guidance to the design of sampling programmes. Part 2, Guidance on .sampling techniques. Part 3, Guidance on the preservation and handling of samples. Purl 4 , Guidance on .rampling from lakes, natural and man-made. Part 5, Guidance on sampling of drinking water and water usedfbr.fbod and beverage processing. Part 6, Guidance on sampling of rivers iind streams. Part 7, Guidance on sampling o f water and steam in boiler plants. Part 8, Guidance on the sampling of wet deposition. Part 9, Guidance on sampling from marine waters. Part 10, Guidance on sampling of wastewaters. Part 11, Guidance on sampling of groundwaters. Part 12, Guidance on sampling of industrial cooling water. Part 13, Guidance on sampling of sludges and sediments.

27.3.3 Soil and Mining IS0 1213-2, 1992,

I S 0 1988, 1975, I S 0 3081, 1986, I S 0 3082, 1987, I S 0 3083, 1986, IS0 3085, 1986, I S 0 3086, 1986, I S 0 4296 - 1, 1984, IS 0 4296-2, 1983, I S 0 8530, 1986, I S 0 8541, 1986,

I S 0 8542, 1986,

Solid mineral juels - vocabulary - part 2 : terms relating to sampling, testing and analysis. Hard coal - sampling. Iron ores - increment sampling - manual method. Iron ores - increment sampling andsample preparation-mechanicalmethod. Iron ores - preparation of samples - manual method. Iron ores - experimental methods ,for checking the precision of sampling. Iron ores - experimental methods ,for checking the bias of sampling. Manganese ares - sampling - part 1: increment sampling. Manganese ores - sampling - part 2: preparation qf sump[rrs. Manganese and chromium ores - experimental methods f o r checking the precision of sample division. Manganese and chromium ores - experimental methods for checking the bias i~fsamplingand sample preparation. Manganese and chromium ores- experimental methods f o r checking the precision of sampling.

I~. S0

working document, 1993, Soil quality - Sample pretreatment for determination of organic contaminants in soil, - second draft, Document ISO/TC I90jSC 3/WG 9N45 Soil quality - pretreatment of samples f o r physico-chemical ana1.vsi.r I S 0 11464, 1992,

Overview of References f o r Sampling and Related Topics

509

27.3.4 Biota I S 0 7828, 1985, I S 0 8199, 1988,

I S 0 8265, 1988, I S 0 9998, 1991,

Water quality - methods f o r biological sampling. Guidance on handnet sampling of aquatic benthic macro-invertebrates. Water quality - general guide to the enumeration of micro-organisms by culture. Water quality - design and use of quantitative samplers ,for benthic macro-invertebrates on stony substrata in shallow ,freshwaters. Water quality - practices ,for evaluating and controlling microbiological colony count media used in water quality tests.

Environmental Sampling for Trace Analysis

Edited by Bernd Markert 0 VCH Verlagsgesellschaft mbH, 1994

Index

AA see atomic absorption spectrophotometry AC see automated colorimetry accuracy 94 - in sampling and analysis 106f Aceri tatarico-Quercetum - element concentration cadasters 44 1 - - of free leaves 441 - element contents 436,438 - - in leaves of tress 436f - - in soil 438f - rank of order for plants 440 acetate - determination - - effect of peroxyacetyl nitrate 175 adsorbents - for organic gas sampling 168ff - - capacity 170 - - cleaning 171 - - desorption behavior 170 - - DNPH-coated 173f - - multi-bed- 171 - - volume 170 aerial photography 254 aerosol analysis 126ff - flow diagram 151 - sources of error 152 aerosol samples see air filter samples AES see atomic emission spectroscopy air filter samples - carbon determination 132f - elemental composition 129f - mass measurements I26ff - water soluble ions 13Off algae - samples - - of benthic- 384 - - of epiphytic- 385

of planktonic- 384 analysis of variance (ANOVA) 95,97ff - Robust ANOVA 98, l 0 l f analytical error - sources 243 animals - element concentrations 83f - influence of stemflow 454 antagonism - of trace elements 82, 87 atomic absorption spectrophotometry (AAS) - elemental measurement - - of aerosol samples 127ff - sensitivity 74 atomic emission spectroscopy (AES) - sensitivity 74 automated colorimetry (AC) 127f - measurement of aerosol ions 13 1 autopsy analysis 75 - -

bailers 282,298,438 barley - element concentration cadasters - - of leaves 441 - element contents 438 beneficial elements 78ff, 383 beneficiality - criteria 80 P-HCH (beta-hexachlorocyclohexane) - foodstuff contamination 342f bias - absolute 94 - acceptable levels 106f - estimation of analytical- 103ff - - certified reference materials (CRM) 104f

512

Index

- - comparative analysis - - spikerecovery 105 - estimation of sampling- - spikerecovery 106 - relative 94

105 105f

bi oavailability of trace elements 75 bioindication 4 15 see also biomonitoring - of air pollution - - by cryptogams 415ff biomonitoring 415 see also bioindication - by red wood ants 465ff - of heavy metal pollution - - by plants 395ff blanks 104f - equipment- 193 - field- 193 - sampling- 193 - sediment sampling 361 blood 76f - trace element contents 77 breakthrough volume - of adsorbents 170 -

cadmium feeding of red wood ants 483ff - - effects 484ff cadmium pollution - of agricultural soils 368ff - of irrigation water 369f cambisols 309ff - dystric 309 - - representative sampling site 310 - eutric 309 - - representative sampling site 3 10 carbon - analysis - - thermal manganese oxidation (TMO) method 133 - - thermal-optical reflectance (TOR) method 133 - - thermal-optical transmission (TOT) method 133 -

classes 132 carbonyl compounds - sample storage 175 - selective preconcentration - - DNPH-based 172ff certified reference materials (CRM) 104f Chernozem brown forest soil 435ff - element contents 438 - - arable land 438f - - forest 438f chlorophyll values - in lakes 207ff Cladonia convoluta 4 17 - samples 418 - trace element contents 419ff - - effects of washing 430f - - in the soil substrate 426,428f - - vertical variability 42 1ff Cladonia furcata 4 17 - samples 418 - trace element contents 419ff - - horizontal variability 423 f - - vertical variability 420ff cleaning - of plant material 403ff - - deciduous tree leaves 406f, 445 - - herbage 407ff - - lichen thalli 429ff - - mosscushions 429ff - - needles 405 colloids - in groundwater - - particle size 291 - - reactivity 291 - - transport 290f conifers - element contents - - organ specific distributions 397f - - seasonal variation 402 corers see samplers crosstabulation - of soils 307 cryptogam bioindication - of air pollution 4 15ff cryptogam transplantation technique 417 -

Index

deciduous trees - element contents - - organ specific distributions - - seasonal variation 402 - washing of leaves 406f, 445

-

397f

desorption solvent extraction 170 thermal 170ff detection limits - below detection limit values - - estimation 109f - - organic compounds 1 13ff dew - chemical composition 199f - sampling 199ff DNPH - carbonyl compound sampling 173ff - - effect of ozone 174 duplicate analysis - analysis of variance (ANOVA) 97ff - regression method 97f -

earth’s crust chemical composition 83 Elbe - catchment area 223ff - - hydrological characteristics 225 - chemical analysis - - COD determination 241 f - - methods 240f - concentration cycle - - ofheavy metals 238f. - - ofsuspendedmatter 238f - cross-sectional distribution - - of organic substances 238 - macrozoobenthos colonization 244f - mathematical simulation models 246 - monitoring strategy 228ff - - automatic monitoring stations 231 f - - datacollection 237ff - - data handling 232f, 245f - - measured variables 230f - - measuring error 241 ff - - measuring frequency 23 1,235 ff - - trend analyses 245f - pollution loads 223 f, 226ff -

calculation 234 sampling points 224,229f - - representativeness 234f - sampling schedule 229 - sampling strategy 223 - sediment - - heavy metals 239 - sewage discharge 243ff - - industrial 226 - - municipal 226 - - plume formation 237 - water quality 227 - water sampling method 237 element concentration cadasters 435 - ofleaves - - crops 441 - - trees 441 - - weeds 441 f error estimation - in sampling 93ff essential elements 78ff, 382f essentiality - rules 80 estimation - with varying detection limits 109ff - - organic compounds 113ff eutric luvisols - representative sampling site 310 eutrophication control - algal bloom 220 - chlorophyll levels 207ff - data compilation 218ff - hypothetical monitoring budgets 216ff - in-lake phosphorus concentration 208 f - monitoring program - - calculated costs 213ff - oxygen depletion 208f - sampling program design 204 - water quality parameters 203ff - -

filter holders 146f in-line 147 - open-faced 147 filters -

513

514

lndrx

analysis methods 126ff blank concentrations 138f - Cost 138 - flow resistance 138 - for gas measurements 125ff - for groundwater samples 299 - for particle measurements 125ff - - types 135ff - particle sampling efficiency 138 - pretreatment 139f - stability - - chemical 134 - - mechanical 134 - - temperature 134 filtration - of groundwater 298f - of pore water 356f fish - sampling 250f flow control - for whole-air ambient VOC collection 165 - ofsamplers - - critical orifice- 148, 165 - - differential pressure- 148, 166 - - manual- 147 - - mass- 147, 165 fog - sampling 201 formaldehyde - determination - - ozone effect 174 Formica s. str. see red wood ants freshwater - sampling 187ff - - contamination sources 188f - - for eutrophication control 203ff fungi - influence of stemflow 453f - sampling 386 -

-

gaseous organic acids sample storage 175 selective preconcentration 174f grasses 402 - element contents -

organ specific distributions 398f seasonal variation 402 gravimetry 126, 128 ground water - filtration - - errors 299 - mobile colloids 290f - - particle size 291 - - reactivity 291 - monitoring wells 280f - - development 289f - - purging 292ff - - turbidity 290ff - quality monitoring 279ff - - costs 284 - sampling 279ff, 287ff - - -

hair sample treatment 76 - trace element contents 77 - trace metal storage 76 halogenated hydrocarbons - sorption - - by sampling tool materials 189f heavy metal pollution - biomonitoring - - by plants 395ff - of agricultural soils 365ff, 374ff - - cadmium 368ff - of irrigation water 365ff, 375ff - - cadmium 369ff - - lead 370 - of sediments 375ff - - cadmium 369ff - - lead 370 heavy metals 397 see also trace elements - contents in plants - - effects of rain 409f - - effects of washing 403ff - - organ specific distributions 397f - - seasonal variations 401 ff - in Quercus sp. 436f, 439 - in barley 437f - in Chemozem brown forest soil 438f - in Elbe -

Index

sediments 239 water 2384 red wood ants 48 1ff weeds 439 - in wheat 437,439 - interactions with essential elements 87 heparin 76f herbage - cleaning 407ff - element contents - - organ specific distributions 398f - - seasonal variation 402f - soil contamination 407 honeydew 465 - metal contents 467 - - - in - in

IC see ion chromatography ice - chemical composition 198f - sampling 198 ICP/AES see inductively coupled plasma with atomic emission spectroscopy IHEARU schema 343 f INAA see instrumental neutron activation analysis indicator organs - for trace elements - - blood 76f - - hair 76 - - toenail 76 inductively coupled plasma with atomic emission spectroscopy (ICP/AES) - elemental measurement - - of aerosol samples 127ff instrumental neutron activation analysis (INAA) - elemental measurement - - of aerosol samples 127ff iodine deficiency 85f ion chromatography (2C) 127f ion chromatography (IC) - measurement of aerosol ions 131 irrigation water 365 ff

heavy metal concentrations 370ff, 375 ff - - cadmium 370 - lead 370 - sampling 368, 373 -

judgement sampling 327 Keshan disease 85 lichens 385 as bioindicators - - of air pollution 415ff - influence of stemflow 454 - sampling 385f, 415ff likelihood equations 1lOff literature survey - general aspects of sampling 493 - sampling of - - air 498f - - biota 504ff - - mining 503f - - rocks 503f - - sediments 503f - - sewage 507 - - sludges 503f,507 - - soils 503f - - waste 507 - - water 499ff - statistical methodology 495ff -

macroelements 382f essential 78 maximum likelihood estimates 111 - ofthemean 115 - of variance 1 15 measurement errors 24 1ff - analytical error 95 - sampling error 95 micronutrients 383 mosses - as bioindicators - - of air pollution 415ff - influence of stemflow 454 - sampling 386,415ff -

515

516

Index

mud samples - measured variables 354 mud-water interface - sample collection 355 mycorrhizal fungi - ectomycorrhizal- 387 - vesicular-arbuscular mycorrhizal- 387 NAA see neutron activation analysis needles - washing procedures 405 neighborhood analysis - of soils 307ff nephelometric turbidity units (NTU) 292 neutron activation analysis (NAA) - sensitivity 74 non-contacting liquid presence detector 272f OECD load-response models 2 I8 ores - sampling approaches 3 ff orthic podzols - representative sampling site 310 oxygen depletion - inlakes 208f ozone 1 - effect on formaldehyde determination 174 PAH see polycyclic aromatic hydrocarbons parameter actuator loggers 276 particulate analysis - filter media 134ff - methods 126ff peroxyacetyl nitrate - effect on acetate determination 175 phosphorus - in lakes 208f PIXE see proton induced X-ray emission analysis plant sampling see sampling of plants plants

as biomonitors for heavy metal pollution 395ff chemical composition 83 element uptake systems 382 elemental distributions - agedependent- 389 - in conifers 397f - in deciduous trees 397f - in grasses 398f - in herbage 398f - individual differences 396 - organ specific- 388f, 397ff heavy metal contents - seasonal changes 40 1ff influence of stemflow 454 trace element accumulation 383f PM,o 125 - high-volume samplers 148 - monitoring - - Beta Attenuation Monitor (BAM) 149 - - Tapered Element Oscillating Microbalance (TEOM) 149 - reference methods 128 - sampling systems 140ff pol ycyclic aromatic hydrocarbons (PAW - partitioning 176 - vapor pressures 176 population equivalents (PE) - of pollution 224,226 precision - acceptable levels 102f - estimation of analytical- 96ff - - confidence intervals 98 - - duplicate analysis 97ff - estimation of sampling- - analysis of variance (ANOVA) 99ff - - duplicate samples 99ff - influence of concentration 94 preconcentration - on solid adsorbents - - organic gases 168ff - selective methods - - for carbonyl compounds 172f - - for gaseous organic acids 174f -

Index

proton induced X-ray emission analysis (PIXE) - elemental measurement - - of aerosol samples 127ff pumps - air 147 - for groundwater sampling - - bladder- 298 - - gas-driven 298 - - submersible 282,298 - - suction- 282 - for sample delivery - - bladder 271 f - - peristaltic- 269f - - vacuum- 270f purging - of groundwater wells 292ff - - lowflow- 294f rain chemical composition - sampling 199 red wood ants 375 - alarmpheromone 479 - aphid-ant interaction 466 - biomonitoring 465 ff - cadmium contents 484f - colonies 471 - - social structure 472ff - definition 468 - fish feeding 481 - food 465f - honey feeding 480 - - withcadmium 483ff - metal pollution - - effects on the enzyme balance 486 - natural Cd levels 48 1f - nest - - base 470 - - nest mound 469f - - relative humidity 474 - - temperature 474, 479,484 - - undergroundpart 470f - occurrence 469 - poison tolerance systems 485 - pollutant accumulation 466ff -

517

- pollution stress - - cadmium 483

samples 476ff foragers leaving the nest 477 foragers traveling to the nest 477 inside workers 478 - - larvae 478 - - nest material 478 - - pupae 478 - - reserve ants 478 - - surface workers 477 - sampling 474ff - - order 479 - sampling sites 470 - sexuals 472f - species 468f - supercolonies 471 f - workers - - inside workers 474f - - nurses 474 - - outside workers 475 - - reserve workers 473f rendzinas 309ff - representative sampling site 310 replication - ofsamples 190f -

- - -

sample 126 see also air filter samples 126 - composite- 328f - - flow proportional composite- 260 - - sequential composite- 260 - - time composite- 259 - conservation 19 - contamination 329 - - sources 17f, 188f - definition 93f - discrete- 259f - fractions 194 - gathering - - bladder pumps 271 f - - forced flow delivery systems 269 - - mechanical delivery systems 268f - - peristaltic pumps 269f - - suction lift 269 - - vacuumpumps 270f

518

Index

holding time 275 homogeneity 16f - likelihood - - bivariate case 112f - - one-dimensional case 1 lOff - losses - - of trace elements I8 - - VOC 252 - number 14f - - calculation 362 - - soils 328f - on-situ analysis - - instruments 276f - - parameters 276,282f - preparation - - water 240 - preservation - - fish 251 - - soils 329 - - water 194, 240, 268,275 - pretreatment 194 - - plant material 403ff, 429ff, 445 - - soils 331 - quantity - - soils 328 - replication 190f - size 14 - - water 191f - storage 19 - - carbonyl compounds 175 - - contaminations 168 - - gaseous organic acids 175 - - SUMMATM-treatedcanisters I67f - - water 194,266ff, 283 - - VOC stability 167 - types 12 - volume 14 - - accuracy 272f samplers - automated - - refrigerated 273f - automatic 257f - - components 264 - - composite 251f - - continuous 251 f - - electron controller 264f -

-

- - for VOC collection 272f - - portable 150,262f, 272f - - powersource 264 - - refrigerated 262f - - sample delivery systems 268ff - - sampleintake 265f - - sample transport line 265f - bailers 282 - core- 250

corers 348f alpinegravity- 353 benthosgravity- 353 - - box- 353 - - hand- 352 - - Kajak-Brinkhurst- 353 - - multiple- 354 - - Phleger- 352f - - Piston- 353f - filter monitors - - Beta Attenuation Monitor (BAM) 149 - - Tapered Element Oscillating Microbalance (TEOM) 149 - grab- 250, 348ff - - Birge-Ekman- 350f - - Petersen- 350f - - Ponar- 350f - - VanVeen- 350f - high-volume- 148 - instantaneous- 355 - integrating- 355 - low-volume dichotomous- I48f - mud-water interface- 355 - pumping- 355 - sequential filter- 149 - submersible pumps 282 - suctionpumps 282 sampling - bias estimation - - field blanks 105f - definition 11 - depth - - of waterbodies 196ff, 207 - - sediments 349 - - soils 327 - duration -

- - -

Index

- - wastewater 15 - error - - estimation 93ff - - sediments 359ff - - variance 5

- -

519

forgroundwater 287f

- - soils 323 - of aerosols 125ff - - filter types - of dew 199f - Of fish

135ff

frequency eutrophication control 209f - - techniques 250 groundwater 284 - offog 201 wastewater 15 - of freshwater 187ff water 192, 206 - - for eutrophication management progeneral formula 2 1 1 f grams 203ff - guidelines 19ff - offungi 386 - - for eutrophication management pro- of groundwater 279ff, 287ff grams 203ff - - springs 281 - - VDI guidelines 21 - - wells 280f - in stemflow areas - ofice 198f - - of forests 449ff - of irrigation water 347ff, 368, 373 - in the river Elbe 223ff - of ores - - sampling points 224,229f, 234f - - empirical approaches 3 - in throughfall areas - - history 3ff - - of forests 449ff - - influence of particle size 4 - literature - - influence of sample weight 4 - - air 498f - - mathematical-statistical formulati- - biota 504ff on 6 - - general aspects 493ff - - sample taking schemes 13f - - mining 503f - - theoretical approaches 3 ff - - rocks 503f - of organic gases 163ff - - sediments 503f - - common VOC classes 179f - - sewage 507 - - ozoneartifact 174 - - sludges 503f, 507 - - preconcentration 168ff - - soils 503f - of plants 395ff - - statistical methodology 495 ff - - age factor 389, 397 - - waste 507 - - epiphytes 385f - - water 499ff - - genera1 strategy 389f - location 13f - - height factor 397 - - eutrophication control 212 - - hygrophytes 384 - - groundwater 279ff,288 - - in an Aceri tatarico-Quercetum - - wastewater discharge 253f, 257 435f - - water 206 - - in tropical forests 443ff - norms 19ff - - on arable land 435f - - air 507f - - roots 387 - - biota 509 - - terricolous lichens 4 15ff - - mining 508 - - terricolous mosses 415ff - - soil 508 - of rain 199 - - water 508 - of red wood ants 475ff - objectives 256f - of sediments 368, 373 -

-

-

5 20 - -

Index

Elbe 239

- of sediment pore water - - methods 356f

ofsnow 199 of soils 367f, 373,435f - - international standards 321 f - - methods 335f - - representative types 305ff - - requirements 323ff - of wastewater 255ff - - receiving streams 249ff - of water 255ff - of whole air 163ff - passive techniques 177f, 296 - plants 381 ff - precision - - acceptable levels 102f - - estimation 99ff - preliminary investigation 325, 361 - quality control 20, 359 - randomization 444f - report 330 - spatial gradient technique 253f - standardized procedures 396f - time 15 - - eutrophication control 209f - types 12f, 260 sampling apparatus - market survey 23ff - - gaseous phase 50ff - - liquidphase 28ff - - solid phase 24ff sampling containers - adsorption - - of metals 190 - for water 267, 275 - for whole-air sampling - - cleaning 165 - - contamination 164 - soils 329f - water 283 sampling costs - calculation 213ff - fixed- 214 - variable- 214 sampling grids 305f, 338f, 361 -

-

bottle rack grid 339ff circular 326 - diagonal 339 - grid cells 326 - random 339 - rectangular 326f, 339, 341 - three-dimensional 327 - triangular 327 sampling history - ore mining 3ff sampling plan - criteria 337 - elements 256ff - fixed - - detection of a contamination crater 339ff - - grid plan 338f - hypothesis-guided - - detection of /3-HCH contamination 342ff sampling program - examples 261f - for eutrophication control 204 - groundwater monitoring 283, 293 sampling sites - of red wood ants 470 sampling systems - dedicated 295, 297 - for aerosols 140ff - - configurations 148ff - - filter holders 146f - - flow controllers 147f - - pumps 147 - - size-selective inlets 142ff - - surfaces 146 - for whole-air ambient VOC collection 165 - multiple-event canister- 166f - PMlomeasurement 140ff - portable 295,297 sampling tool materials - sorption - - of halogenated hydrocarbons 189f saprobic index 238 sea water 8 - element concentrations 83f -

Index

sediment penetrometer 354 sediment pore water - sampling - - centrifugation 356f - - dialysis 356f - - filtration 356f - - squeezers 357 sediments - heavy metal concentrations 375ff - - cadmium 370 - - lead 370 - oflakes - - sampling 347ff - sampling 368, 373 - - depth 349 - - Elbe 239 - subsampling 358 selenium deficiencies 85 semi-volatile organic compounds (SOC) 175ff - polycyclic aromatic hydrocarbons 176f sewage discharges - mathematical simulation models 246 - monitoring 243ff - plume formation 237 snow - chemical composition 199 - sampling 199 SOC see semi-volatile organic compounds soils - P-HCH contamination 342f - cambisols - - dyshic 307 - - eutric 307 - changes due to stemflow - - acidification 452f, 457f - - in Ranker 455ff - - in podsolized brown forest soil 455 ff - - point-source inputs 453 - chernozems 307f - - element contents 438 - - sampling 435f - determination of representative- 305ff - - crosstabulation 306f

521

- - neighborhood analysis 306ff -

for rice growing 365ff

- - heavy metal concentrations 368 ff, 374ff international sampling standards 321 ff - orthic podzols 307ff - phosphorus contents 3 18 - rendzinas 307ff - representative European- 306ff - - for sorption testing 309f - - sampling sites 310 - sample preservation 329 - sample pretreatment - - cooling 332 - - drying 332 - - grinding 331 - - international standards 331 ff - - milling 331f - sampling 367f, 373 - sampling depth 327f - sampling mehods 335f - - fixedplans 338ff - - hypothesis-guided plans 342ff - sampling objectives 323 - TNT contamination 339ff - trace element contents 426,428 - variability - - variogram analysis 310ff spike recovery 105f squeezers - for pore water sampling 357 stemflow areas - element contents - - beechroots 458f - - podsolized brown forest soil 456 - - Ranker soil 457 - pH values - - podsolized brown forest soil 456 - - Rankersoil 457 - sampling 449ff stemflow - amounts 449f - chemical composition - - elemental concentrations 45 I f - - pH-values 451f -

522

Index

influence on animals 454f influence on fungi 453 - influence on microorganisms 453 - influence on plants 454f - influence on soils 452f subsampling - of sediment samples 358 synergism - of trace elements 82, 87 -

-

Taraxacunz ofSicinale 3 elemental distributions 399 - - background regions 400 - - polluted regions 401 - leaves - - effects of washing 408ff - - raineffects 409f thermocline - determination 207 throughfall - amounts 449f - chemical composition 45 1 - - elemental concentrations 45 1 f - - pH-values 451 throughfall areas - element contents - - beech roots 458f - - podsolized brown forest soil 456 - - Rankersoil 457 - pH values - - podsolized brown forest soil 456 - - Rankersoil 457 - sampling 449ff TNT (trinitrotoluene) - soil contamination 339ff toenail - selenium 76 - trace element contents 77 Tortulu ruralis 417 - samples 418 - trace element contents 41 9ff - - effects of washing 430f - - horizontal variability 423f - - in the soil substrate 426,428f - - vertical variability 421 ff -

Total Suspended Particulate matter (TSP) 125 toxicity - parameters 81 f trace element contents - in blood 77 - influence of sample washing 429ff - in hair 77 - in lichens 419ff - - variability 426 - in mosses 419ff - - variability 426 - in soil 426,428 - in toenail 77 - in tropical tree leaves - - seasonal variations 445f trace elements 73 see ulso heavy metals - accumulation in sea animals 84 - analytical methods 73ff - - sensitivity 74 - antagonism 82, 87 - beneficial- 78f - bioavailability 75 - deficiencies - - iodine 85f - - selenium 85 - essential- 78f - generally essential- 78f - indicator organs 76f - in plants 382ff - - accumulation 383f - interdisciplinary research 88f - little known 78f - normal distribution 36 1,444 - pollution sources 85f - pollution trend monitoring 443 - skewed distribution 361,444 - supplementation by nutrition 86f - supply - - iodine 85f - - selenium 85 - synergism 82, 87 - toxicity - - parameters 81f tropical rain forest

Index

pollution trend monitoring 443 turbidity 290ff - units 292

-

ultra-trace elements 74, 78 variance analysis 95, 97ff - - water sampling 21 1 - analytical 97ff, l O l f f - geochemical 99, 101 - of sampling 95,99, 101ff variogram analysis - mathematical concept 3 I 1 f - model applications 3 15ff - of soil variability 310ff variogram - anisotropy 314 - axes 312 - bounded 313 - hole-effect 3 15 - models 313f - nested 315 - nugget-effect 3 14 - periodical 315 - P0:distribution 316f - range 312f - sill 312f - SO, distribution 3 I6 - unbounded 313 VOC see volatile organic compounds volatile organic compounds (VOC) - automatic sampler 272f - collection in whole air 164ff - losses - - from wastewater samples 252f - passive samplers 177f - - uptake rate 178 - preconcentration 168 - sampling methods 179f - storage 167 -

wastewater

- sampling 249ff -

-

pollution - of sediments 365ff

of soils 365ff of water 365ff - sampling 255ff water see also freshwater 187 see also groundwater 279 see also irrigation water 365 see also wastewater 255 - sampling 255ff waterbody - eutrophication status 205 - longitudinal water quality gradients 210ff - stratified - - sampling depths 198,207f - - sampling position 196 - trophic classification 209 - unstratified 196, 207 - - sampling position 197 water quality - automatic monitoring stations 231 f - diurnal variations - - in dissolved oxygen 209 - Elbe - - characteristics 227 - eutrophication-related - - gradients 208ff - - parameters 203ff, 209 - influencing factors 236 - longitudinal gradients 210ff - monitoring - - Elbe 223ff - of groundwater - - indicator parameter equilibration 292,295ff - - monitoring 279ff - parameters 255f - saprobic values 238 weeds - element concentration cadasters - - of leaves 441 wells - for groundwater sampling 280f - - development 289f - - installation 289 - - purging 292ff -

-

-

523

524

Index

- turbidity 290f wheat - element concentration cadasters - - of leaves 441 - element contents 437 Wilson’s disease 82 wooded steppe see Aceri tataricoQuercetum -

X-ray fluorescence analysis (XRF) - elemental measurement - - of aerosol samples I27ff

sensitivity 74 XRF see X-ray fluorescence analysis

-