Advances in Agronomy, Volume 58

A DVANCES IN

Advisory Board Martin Alexander

Eugene J. Kamprath

Cornell University

North Carolina State University

Kenneth J. Frey

Larry P. Wilding

Iowa State University

Texas A&M University

Prepared in cooperation with the American Society of Agronomy Monographs Committee P. S. Baenziger Jon Bartels Jerry M. Bigham M. B. Kirkham

William T. Frankenberger, Jr. Chairman David H. Kral Dennis E. Rolston Sarah E. Lingle Diane E. Stott Joseph W. Stucki Kenneth J. Moore Gary A. Peterson

D V A N C E S I N

onomy VOLUME 58 Edited by

Donald L. Sparks Department of Plant and Soil Sciences University of Delaware Newark, Delaware

ACADEMIC PRESS San Diego London Boston NewYork

Sydney Tokyo Toronto

This book is printed on acid-free paper.

@

Copyright 0 1997 by ACADEMIC PRESS All Rights Reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher.

Academic Press, Inc. 525 B Street, Suite 1900, San Diego, California 92101-4495, USA

http://www.apnet.com Academic Press Limited 24-28 Oval Road, London NW 1 7DX, UK http://www.hbuk.co.uk/ap/ International Standard Serial Number: 0065-21 13 International Standard Book Number: 0-12-000758-4

PRINTED IN THE UNITEDSTATES OF AMERICA 96 97 9 8 9 9 00 01 BB 9 8 7 6

5

4

3 2 1

Contents CONTRIBUI-ORS. ............................................ PREFACE. .................................................

ix xi

EFFECTOF SORPTIONON BIODEGRADATION OF SOILPOLLUTANTS Kate M. Scow and Carol R. Johnson I. Introduction. .............................................. 11. Overview of Sorption.. ..................................... 111. Biodegradation of Sorbed Chemicals. ......................... N. Analogies between Sorbed Chemical Degradation and Carbon Flow in Soil.. .................................. V. Factors Controlling Coupled Biodegradation and Sorption.. . . . . . VI. Bioremediation of Sorbed Chemicals. ......................... VII. Future Research Needs and Directions ........................ References ................................................

1 3 12 30 32 42 47 48

HERBICIDE RESISTANCE:IMPACT AND MANAGEMENT S. B. Powles, C. Preston, I. B. Bryan, and A. R. Jutsum I. Introduction: Major Crops and Herbicide Markets.. ............ 11. The Threat from Herbicide-Resistant Weeds. ..................

111. Managing Herbicide Resistance .............................. IV. Conclusions ............................................... References ................................................

57 64 71 83 84

PHYSICAL NONEQUILIBFUUM MODELING APPROACHES TO SOLUTETRANSPORT IN SOILS Liwang Ma and H. M. Selim I. Introduction. ..............................................

11. Mobile-Immobile Two-Region Models. ....................... 111. Two-Flow Domain Models. .................................

IV Capillary Bundle Models,. .................................. V. Multiple-Flow Domain Models

..............................

VI. Coupled Physical and Chemical Nonequilibrium Models . . . . . . . . V

95 101 116 122 124 126

CONTENTS

vi

w. Field Applications .......................................... VIII. Summary and Conclusion ................................... Appendix: Nomenclature .................................... References ................................................

I. I1. 111.

lv.

SILICON MANAGEMENT AND SUSTAINABLE RICEPRODUCTION N . K . Savant. G. H. Snyder. and L. E . Damoff Introduction ............................................... Silicon Nutrition in Rice .................................... Silicon in Soil and Water.................................... Silicon Management Agenda ................................. Potential Benefits of Silicon Management .....................

V. VI . Agronomic Essentiality of Silicon Management ................ VII . Determining Need for Silicon Fertilization .................... VIII. Suggestions for Research .................................... Ix. Summary ................................................. References ................................................

138 140 142 144

151 152 155 158 165 177 179 185 188 189

TISSUE CULTURE-INDUCED VARIATION AND CROP k R O V E M E N T

R . R. Duncan

I. Introduction............................................... I1. Causes and Range of Variation ............................... III. Methodological Basis for Variation ........................... n? Rate of Variation ........................................... V. In Vim Selection .......................................... VI. Conclusions ............................................... References ................................................

201 202 207 214 215 222 224

GEOSTATISTICAL ANALYSISOF A SOILSALINITYDATASET G. Bourgault. A. G. Journel. J . D . Rhoades. D . L . Corwin. and S. M. Lesch

I. I1. I11. TV. V.

Introduction ............................................... Exploratory Data Analysis ................................... Mapping the EC. Distribution ............................... Filtering Structures ......................................... Spatial Cluster Analysis .....................................

241 245 254 269 276

CONTENTS

VI . VII. VIII. IX.

Stochastic Imaging ......................................... Assessment of Spatial Uncertainty ............................ Ranking of Stochastic Images ................................ Conclusions ............................................... References ................................................

vii 280 287 291 291 292

FURTHER PROGRESS IN CROPWATERRELATIONS Neil C. Turner

I . Introduction ............................................... I1. Measurement of Water Deficits ..............................

111. “Sensing” Water Deficits .................................... n! Water Deficits and Yield .................................... V. Use of Reserves to Maintain Yields under Water Deficits ........ VI. Water Use Efficiency ....................................... VII. Drought Resistance ......................................... VIII . Concluding Remarks ....................................... References ................................................

293 294 296 302 305 307 314 324 325

INDEX.....................................................

339

This Page Intentionally Left Blank

Contributors Numbers in parentheses indicate the pages on which the authors' contributions begin.

G. BOURGAULT (241), Geological and Environmental Sciences Department, Stanford University, Stanford, California 94305 I. B. BRYAN (57), Zeneca Agrochemicals, Jealotts Hill Research Station, Bracknell, Berkshire, United Kingdom D. L. CORWIN (241), USDA-ARS, U. S. Salinity Laboratory, Riverside, California 92507 L. E. DATNOFF (1 5 I), Department of Plant Pathology, Everglades Research and Education Center, University of Florida, Belle Glade, Florida 33430 R. R. DUNCAN (201), Department of Crop and Soil Sciences, University of Georgia, Grzfin, Georgia 30223 CAROL R. JOHNSON (l), Department of Land, Air, and Water Resources, University of California, Davis, Davis, California 9561 6 A. G. JOURNEL (241), Geological and Environmental Sciences Department, Stanford University, Stanford, California 94305 A. R. JUTSUM (57), Zeneca Agrochemicals, Jealotts Hill Research Station, Bracknell, Berkshire, United Kingdom S. M. LESCH (241), USDA-ARS, U. S. Salinity Laboratory, Riverside, California 92507 LrWANG MA ( 9 9 , Department of Agronomy, Agricultural Center, Louisiana State University, Baton Rouge, Louisiana 70803 S. B. POWLES (57), C. R. C. for WeedManagement Systems, Waite Campus, University of Adelaide, Glen Osmond, Australia C. PRESTON (57), C. R. C. for Weed Management Systems, Waite Campus, University of Adelaide, Glen Osmond, Australia J. D. RHOADES (241), USDA-ARS, U. S. Salinity Laboratory, Riverside, California 92507 N. K. SAVANT (1 5 l), StaSav International, Florence, Alabama 35630 KATE M. SCOW (I), Department of Land, Air, and Water Resources, University of California, Davis, Davis, California 9561 6 H. M. SELIM (95),Department of Agronomy, Agricultural Center, Louisiana State University, Baton Rouge, Louisiana 70803 G. H. SNYDER (1 5 l), Departments of Soil- Water Science and Plant Pathology, Everglades Research and Education Center, University of Florida, Belle Glade, Florida 33430 NEIL C. TURNER (293), CSIRO Division of Plant Industry, Centrefor Mediterranean Agriczlltural Research, Wmblq, (Perth), W A. 6014, Australia ix

This Page Intentionally Left Blank

Preface Volume 58 consists of seven comprehensive and cutting-edge reviews on various aspects of plant and soil sciences. Chapter 1 is a contemporary review of the effect of sorption on biodegradation of soil pollutants. Sorption equilibrium and kinetics are covered, and extensive discussions on biodegradation of sorbed chemicals, factors controlling coupled biodegradation and sorption, and bioremediation of sorbed chemicals are presented. Chapter 2 deals with the impact and management of herbicide resistance, a major area of research in plant biotechnology. Chapter 3 provides a comprehensive discussion of physical nonequilibrium models that can be used to predict solute transport in soils. Various mobileimmobile, two and multiple flow, and coupled physical and chemical nonequilibrium models are described and applied to field settings. Chapter 4 thoroughly covers aspects of silicon in plants, soil, and water and its use and management in rice production. Chapter 5 provides a timely review on tissue-culture-induced variation and crop management. Topics that are discussed include causes and range of variation, the methodological basis for variation, the rate of variation, and in v i m selection. Chapter 6 is an extensive geostatistical analysis of a soil salinity data set. The final review, Chapter 7, discusses advances in crop water relations, including measurement and sensing of water deficits, deficit effects on crop yields, use of reserves to maintain yields under water deficits, water use efficiency, and drought resistance. Many thanks to the authors for their fine contributions.

1 h ~ ~ i . L. 1 ) SPARKS

Xi

This Page Intentionally Left Blank

EFFECTOF SORPTIONON BIODEGRADATION OF SOIL POLLUTANTS Kate M. Scow and Carol R.Johnson Department of Land, Air and Water Resources, University of California, Davis, California 95616

I. Introduction 11. Overview of Sorption A. Sorption Thermodynamics B. Sorption Kinetics 111. Biodegradation of Sorbed Chemicals A. Coupled Process Models under Batch Conditions B. Coupling Advection, Sorption, and Biodegradation- Column Studies C. Extrapolation of Microbial Rate Parameters across Experimental Systems IV. Analogies between Sorbed Chemical Degradation and Carbon Flow in Soil V. Factors Controlling Coupled Biodegradation and Sorption A. Sorption Partition Coefficient B. Diffusion C. Rate of Metabolism D. Spatial Distribution at the Pore and Aggregate Scale E. Effect of Soil Moisture F. Temporal Changes in Bioavailability VI. Bioremediation of Sorbed Chemicals A. Inoculation B. Decreasing Aggregate Size by Mixing and/or Crushing C. Addition of Surfactants D. Evaluation of Rate-Limiting Steps in Bioremediation E. Regulatory Implications of Reduced Bioavailability VII. Future Research Needs and Directions References

I. INTRODUCTION Without the existence in soil of natural processes that act to preserve organic compounds from microbial activity, there would be little organic matter, and thus 1 Adurnires m A p n m n y . Volume 58 Copyright 0 1997 hy Academic Press. Inc. All rights nf reproduction in any form reserved.

2

K. M. SCOW AND C. R. JOHNSON

life, left in soil. The major, relatively labile reservoirs of organic carbon and energy in ecosystems, such as cellulose, hemicellulose, and lignin, are highly insoluble and over time become increasingly sequestered through associations with mineral components. Consequently, these compounds and their products (e.g., soil humus) cannot be readily degraded by microorganisms. Many organic pollutants also become associated with mineral and organic surfaces or enmeshed in the three-dimensional organic-mineral complex of soil organic matter. These physical processes also reduce the biodegradation rates of pollutants. The term bioavailability designates the state of that fraction of a chemical that is available for uptake and/or transformation by living organisms. Although associated primarily with ecotoxicology, and usually in reference to organic and metallic pollutants, the term bioavailability is also relevant to native organic material. Thus, the “problem” of bioavailability has existed for microorganisms far longer than has the presence of xenobiotic chemicals in the environment. Sorption, insolubility, and related processes are largely responsible for controlling the bioavailability of many pollutants to microorganisms in soils and sediments. The turnover rates of forms of a chemical that are not bioavailable are usually slower than those for the same chemical form in solution (Scow, 1993; Rijnaarts et al., 1991; Ogram et al., 1985). In addition, for many chemicals with limited bioavailabilities, the rate-limiting step in their biodegradation is mass transfer from an unavailable to available form. Consequently, because multiple processes are involved, prediction of these chemicals’ biodegradation rates is far more complex than is the case for readily available chemicals. The practical issues associated with the reduced bioavailability of pollutants are many. How strongly a pollutant is sorbed, as indicated by its sorption partition coefficient, and how rapidly it biodegrades are criteria used for screening its potential to leach to groundwater (Gustafson, 1989; Jury et al., 1987). However, pollutants that should be readily biodegradable (and are not strongly sorbed) have been found to persist for decades and eventually find their way to groundwater, such as has been observed for ethylene dibromide (Pignatello et al., 1987). Soils polluted with organic contaminants become increasingly difficult to remediate, by any method, the longer the pollutant has been in contact with the soil. Recently, the issue of what is an acceptable treatment endpoint in contaminated soil and groundwater has been discussed with respect to the issue of bioavailability (Alexander, 1995; Beck et al., 1995). Some researchers challenge whether a strongly sorbed pollutant poses a significant environmental problem if the pollutant is not available for uptake by sensitive populations, even if its limited bioavailability also means it cannot be taken up by microorganisms that can potentially degrade it. Much of the debate over this issue is fueled by uncertainty in our understanding of coupled sorption/desorptionand biodegradation processes and of their impact on bioavailability. The objectives of this chapter are to demonstrate the importance of physical

EFFECT OF SORPTION ON BIODEGRADATION

3

mass transfer processes-including nonequilibrium sorption, diffusion and other pollutant-soil interactions-in governing rates of biodegradation of organic pollutants in soil. To understand the coupling of sorption/diffusion and biodegradation kinetics, one must answer four questions: (i) of the total pollutant entering the soil, which fraction is available, in time and space, to microorganisms?; (ii) what are the mechanisms of exchange between the available and unavailable pools?; (iii) what controls the kinetics of metabolism of the available portion of the pollutant?; and (iv) what interactions between mass transfer and metabolism must be considered in describing the biodegradation of sorbed pollutants? For each question, one must identify the forces and processes involved, develop mathematical descriptions of key processes, and conduct well-designed experiments to test and modify hypotheses about the phenomena. We discuss several approaches, of increasing complexity, for describing coupled mass transferbiodegradation reactions in soil and present data to illustrate important concepts. We also discuss the implications of this knowledge for the bioremediation of sorbed chemicals. This chapter differs from previous reviews discussing impacts of sorption on biodegradation. Sims et al. (1991) reviewed the effect of sorption on pesticide biodegradation and implications for pollutant transport. Scow ( 1993) provided a broad overview of sorption and biodegradation of organic pollutants in soil, including discussion of indirect effects of pollutant-sorbent interactions on microorganisms and methodological limitations. Mihelcic et al. (1993) reviewed the impact of phase partitioning, with respect to both sorption and solubilization, on the biodegradation of hydrophobic pollutants. Brusseau et al. (1992b) provided an overview of coupled chemical-biological process models for solute transport. Beck et al. (1995) reviewed the implications of mass-transfer limitations on biodegradation in the context of soil quality limits.

11. OVERVIEW OF SORPTION

A. SORPTIONTHERMODYNAMICS Sorption refers to the association of a chemical with soil solids and includes a wide variety of mechanisms (Hamaker and Thompson, 1972; Koskinen and Harper, 1990). Which mechanism(s) dominate the interaction of a particular compound with soil mineral and/or organic matter depends on the chemical’s properties and the soil composition. Ionic compounds can bond via coulombic forces to charged soil sites (ion-exchange reactions). Some functional groups such as benzylic amines are capable of forming covalent bonds with soil organic matter. Basic compounds such as amines with available lone pair electrons can

4

K M. SCOW AND C. R. JOHNSON

complex with positive metal centers displacing the water or inorganic hydroxyl previously attached to the metal (ligand exchange). The mechanisms listed previously result in strong bonds to soil solids. Weaker bonds may be formed by cation bridging in which an anionic or polar functional group on the organic compound forms an inner sphere complex (one with no intervening molecules) with an exchangeable cation on the soil surface. If the organic compound cannot displace the waters hydrating the exchangeable cation, an outer sphere complex may form (water bridging). Other weak sorption mechanisms include charge transfer (partial election orbital overlap and exchange of electron density between electron-rich and electron-poor compounds) and hydrogen bonding. Finally, the weakest forces, but ones that apply to all organic compounds, are Van der Waals forces resulting from attractions between fixed and/or fluctuating molecular dipoles. This chapter will focus primarily on hydrophobic compounds; that is, compounds that are nonionic, contain no polar functional groups, and are relatively insoluble in water. Many environmental pollutants are hydrophobic compounds. Chemicals such as alkanes and chlorinated alkanes, alkylated and chlorinated benzenes, and polyaromatic hydrocarbons belong in this class. In general, the lower a compound’s water solubility, the more it sorbs to soil solids (Chiou et al., 1983). Hydrophobic compounds interact with soil solids primarily through nonspecific low-energy Van der Waals forces. Biodegradation studies involving ionic and polar organic compounds will be discussed in this chapter only to illustrate important general principles. Organic matter is the primary soil sorbent in surface soils for nonpolar organic chemicals. The extent of soil sorption for a particular chemical may be expressed as an equilibrium partition coefficient, Kp, for the chemical between soil solids and water.

kp =

(ng chemical sorbed/g soil solid) (ng chemical dissolved/ml soil solution)

K p values may vary by several orders of magnitude for the same compound on different soils. These differences are reduced when the Kp value is divided by the fractional organic carbon content of these soils to produce the organic carbon partition coefficient, K, (Lyman, 1982). KO, values for the same chemical on different soils have been estimated to vary by a factor of three to five for nonpolar organic chemicals (Rutherford et al., 1992; Curtis et al., 1986). Mineral surfaces, especially dry surfaces, also sorb organic compounds. In moist soils, however, water generally displaces organic compounds from mineral sorption sites, and the small remaining contribution of mineral sorption is overshadowed by organic matter sorption (Chiou, 1990). Contributions of mineral components must be considered in subsurface soils in which organic matter

EFFECT OF SORPTION ON BIODEGRADATION

5

contents are very low or in soils with large ratios of high surface area clays (such as montmorillonite) to organic carbon (Karickhoff, 1984). Interactions between the organic and mineral components of a soil may influence sorption. Murphy et al. (1990) found the K , values of several hydrophobic pollutants to vary with the type of clay mineral present in humic-mineral complexes to which the pollutants were sorbed. Nonpolar organic compounds, containing no strongly charged sites or reactive functional groups, interact with organic matter only through weak bonding mechanisms. Water’s free energy gain upon expulsion of the organic compound drives the sorption reaction. Chiou et al. (1983) described sorption of hydrophobic organic compounds as a partitioning process between water and soil organic matter that is depicted as an amorphous polymeric substance. Evidence supporting the partitioning model includes (i) the dependence of sorption extent on soil organic carbon content, (ii) sorption isotherm linearity to relatively high sorbate concentrations, and (iii) low sorption enthalpies for nonpolar organic chemicals from water to soil. The linear sorption isotherms predicted by partitioning theory have been widely observed. For linear isotherms, all sorption behavior is contained in a single concentration independent parameter, Kp or K,, which can be estimated from chemical properties. Due to their mathematical simplicity, Kp and KO, values have been widely used to predict sorption equilibrium in pollutant fate models. Not all experimental data support a linear partitioning model, however. Nonlinear isotherms have been reported for a variety of compounds: tetrachloroethylene, 1,4-dichlorobenzene, 1,2,4-trichlorobenzene (McGinley et al., 1993), phenanthrene (Young and Weber, 1995), phenylurea herbicides (Spurlock and Biggar, 1994b), and pesticides and other compounds (Hamaker and Thompson, 1972; Mingelgrin and Gerstl, 1983). Spurlock and Biggar (1994a,b) have proposed that for nonionic compounds that contain polar functional groups, such as phenylurea herbicides, specific organic matter-sorbate interactions lead to changes in sorption energy with the amount of chemical sorbed resulting in nonlinear isotherms. This theory does not explain, however, the nonlinear isotherms observed for molecules with no polar functional groups such as phenanthrene. Organic matter between soils or within the same soil has been reported to differ in polarity, elemental composition, aromaticity, condensation, and degree of diagenetic evolution from a loose polymer to condensed coal like structures (Grathwohl, 1990; Rutherford et al., 1992; Garbarini and Lion, 1986; Gauthier et al., 1987; Weber et al., 1992a; McGinley et al., 1993; Young and Weber, 1995; Preston and Newman, 1992). A second theory, the distributed reactivity model, views soil as an assemblage of discrete components each of which has a distinct sorption capacity and linearity (Weber et al., 1992a; McGinley et al., 1993; Young and Weber, 1995). Observed sorption behavior is the sum of the

6

K. M. SCOW AND C. R. JOHNSON

behaviors of each of the individual sorbing units. The degree of diagenetic alteration of the organic matter associated with each soil component is considered the main determinant of that component's sorptive behavior. Older, more condensed organic matter, as found in kerogens, may display sorptive characteristics closer to those of a surface adsorptive model (nonlinear isotherms, competition among sorbates for specific sorption sites). Sorptive processes in younger, more amorphous organic matter will be well described by a partitioning model (linear isotherms and no competition effects). In general, nonlinear soil isotherms can be described by the Freundlich equation: If considered over a narrow concentration range, all sorption isotherms appear linear and, as stated previously, many authors have reported that observed soil isotherms fit a linear model. Most environmental fate models do not consider the effect of sorption nonlinearity on pollutant transport and degradation. Simulations and experimental results of soil column studies show that isotherm nonlinearity can affect the transport of solutes in soil and obfuscate the interpretation of soil column studies (Spurlock et af., 1995; Brusseau and Rao, 1989b; Weber et af., 1992). The effect of equilibrium nonlinearity on coupled transport and first-order transformation kinetics has been evaluated by Brusseau (1995). He concluded that for Freundlich exponents less than 0.95, nonlinear sorption equilibrium can alter transport kinetics and retard biodegradation rates.

B. SORPTION KINETICS 1. Processes

Partitioning of a chemical between soil solids and soil water can require hours or months to reach equilibrium. For a given soil, sorption and desorption kinetics slow as the sorbate K p value increases (Karickhoff and Morris, 1985; Brusseau and Rao, 1989a). For nonpolar organic compounds in soil, sorption (and desorption) kinetics generally exhibit two stages. Roughly 1-50% of the chemical will be sorbed in a relatively short initial period (minutes to hours) with the remaining chemical sorbed slowly (days to months) (Karickhoff and Moms, 1985; Coates and Elzerman, 1986; Ball and Roberts, 1991). With increased residence time in the soil, the fraction of the sorbate exhibiting slow desorption kinetics increases (Steinberg er al., 1987; Pignatello, 1990a,b; Pavlostathis and Mathavan, 1992). The exact mechanism underlying sorption/desorption kinetics in soil is unknown. The reader is referred to Pignatello and Xing (1996) for an in-depth review of the subject. Slow sorption/desorption kinetics can be caused by the

EFFECT OF SORPTION ON BIODEGRADATION

7

activation energy of chemical bonds or physical transport processes. For hydrophobic compounds, which primarily form weak bonds with soil solids, ratelimited chemical reactions are usually not considered as important as physical rate-limiting processes (Brusseau and Rao, 1989a). Two transport process have been proposed to explain sorption/desorption rates: (i) diffusion through intraaggregate pores, and (ii) diffusion through the soil organic matter matrix itself (Pignatello, 1989; Brusseau and Rao. 1989a,b; Pignatello and Xing, 1996). In soil, however, a variety of mechanisms may occur simultaneously. For example, phosphate sorption, which does involve slow chemical reaction rates, has been found to be ultimately diffusion controlled (Aharoni et al., 1991). For hydrophobic chemicals in soil, it may be that both proposed diffusion mechanisms operate or that diffusion processes are combined with the slow release of a fraction of the sorbate bound to sites with high desorption activation energies (physical transport and chemical bonds) (Pignatello and Xing, 1996).

2. Mathematical Models for SorptiodDesorption Kinetics a. First-Order Mass-Transfer Equations The simplest first-order model postulates that the sorption/desorption rate is proportional to the concentration of dissolved/sorbed chemical. This model, however, cannot emulate the biphasic sorption rates often observed in the laboratory (Connaughton et al., 1993). A more useful model divides sorption sites into equilibrium and kinetic sites (also called labile and resistant sites or rapid and slow sites). Chemicals in equilibrium sorption sites maintain equilibrium partitioning between these sites and the aqueous phase at all times. Kinetic sorbed sites exchange chemical with the dissolved phase according to first-order kinetics. This formulation is mathematically simple and does not require detailed information about sorption mechanisms or soil microstructure. Two-site models (also called two-compartment models) can represent a variety of physical systems. Both chemical rate and diffusion-dependent explanations for sorption kinetics can be approximated by this approach. The two sorbed compartments may represent two classes of sorbing sites, a chemical reaction in series, interior and exterior sites in the soil aggregate, or interior and exterior sites on the organic matter itself (Wu and Gschwend, 1986; Brusseau, 1995). The parameters used in the two site model cannot be calculated a priori from soil and solute properties. Karickhoff (1980) successfully used a two-site sorption model (shown in Fig. 1.) to describe sorption/desorption kinetics for polyaromatic hydrocarbons in soil slumes Studies of solute transport in soil columns have used first-order sorption kinetics in a variety of models to describe coupled sorption-advection kinetics (see references in Brusseau and Rao, 1989b). The two-site and two-region models are examples of this approach (Fig. 2). The two-site model is very similar to Ka-

K. M. SCOW AND C . R.JOHNSON

8

Resistant Phase

Sorbed Phase

Phase

Figure 1 Two-site sorption model. Double-headed arrow links compartments that reach equilibrium instantaneously. Single-headed arrows link Compartments that transfer solute according to firstorder kinetics (from Karickhoff, 1980).

rickhoff’s model in that a fraction of the sorption sites is considered to be in equilibrium with dissolved chemical in the mobile phase (the solution phase that moves down the soil column), whereas the remaining sorption sites exchange sorbate with the mobile phase according to first-order kinetics. The two-region model divides water in the soil column into mobile and immobile regions. Firstorder kinetics describe solute transfer between these two regions. Sorption sites are divided between mobile and immobile regions, but sorbed solute is always at equilibrium with respect to dissolved chemical in its region. The nature of the immobile solution is not specified. It could include dead-end pores, intraaggregate pores, or pores formed by matrix packing. A combination of the two-site and two-region models, a multiple nonequilibrium process model, was proposed by Brusseau er af. (1992a) to simulate situations in which both sorption kinetics (e.g., intraorganic matter diffusion) and diffusion into immobile water regions (e.g., intraaggregate pores) affect sorption/diffusion kinetics in the column (Fig. 2). Although extremely useful, two compartment models can only be an approximation to the soil’s true complexity. More accurately, an infinite number of compartments exist in soil representing sites with slightly different diffusion pathlengths or sorption energies. Connaughton er al. (1993) modeled desorption kinetics using a continuous distribution of first-order rate constants. The shape of the rate constant distribution is described by the gamma function, a two-parameter probability distribution whose mathematical properties are well defined. The shape of the rate constant distribution function determines what fraction of the sorbed chemical experiences fast or slow desorption kinetics and thus controls observed sorption/desorption rates. Connaughton er al. (1993) successfully fit the model to data describing naphthalene desorption kinetics from soils in which the chemical had aged for different periods of time, ranging from 3 days to 30 years. b. Diffusion Equations A second approach to modeling sorption/desorption kinetics assumes that diffusion within the soil matrix controls chemical release to and removal from the

EFFECT OF SORPTION ON BIODEGRADATION

r Multiprocess

9

Nonequillbrlum Model

Mobile Aqueous Phase

Labile Sorbed Phase

Phase

7-F Immobile Aqueous Phase

Kpim>

Labile Sorbed Phase

“iml

4

Resist ant Sorbed Phase

aim2

Two lhgion Model

Mobile Aqueous Phase

Two Site Model

Labile Sorbed Phase

Mobile Aqueous Phase

Resist ant Sorbed Phase

Ffgure 2 Examples of coupled sorption/advection models used in soil column studies (adapted from van Genuchten and Wagenet, 1989; Brusseau et al., 1992a).

10

K. M. SCOW AND C. R.JOHNSON

soil solution. Fick’s law is used to describe diffusion rates. Models of this type generally postulate that solute moves through intraaggregate pores. Sorption to pore walls is instantaneous, reversible, and retards the diffusive movement of the compound. The diffusion equation approach requires a more exact hypothesis concerning the underlying processes controlling sorption equilibrium and kinetics as well as a picture of the soil microgeometry. For diffusion models, it is theoretically possible to measure all model parameters a priori because each parameter represents a measurable physical quantity. Given the complexity of soil, such information is difficult to obtain except for simplified well-defined systems. When working in real systems, one or more model parameters are often estimated from fitting the model equations to experimental data. Wu and Gschwend (1986) used a diffusion model to describe sorption kinetics of four chlorobenzene congeners in soil slurries. Soil and silt particles were modeled as porous spheres and only one model parameter, the effective intraaggregate diffusion coefficient, was unknown and thus obtained from fitting the model to experimental data (Fig. 3). Rao et al. (1980a) used a radial diffusion model to predict the transport of nonsorbing solutes in soil columns containing sand and artificial porous aggregates. All model parameters were measured in independent experiments. The intraaggregate diffusion model accurately predicted observed solute transport kinetics. In many cases, both the first-order model and radial diffusion model describe experimental data equally well and it is important to determine when the simpler

Figure 3 Radial diffusion model (from Schwarzenbach er al., 1993, with permission of the publisher).

EFFECT OF SORPTION ON BIODEGRADATION

11

model may be used. Goltz and Roberts (1986), working with transport data from a sandy aquifer, and Miller and Weber (1986), in column studies, compared transport models that utilized sorption kinetics described either by radial diffusion or first-order rate models. Both models described the data equally well. The first-order rate model can be used to approximate the diffusion model under some conditions. Rao et al. (1980b) and Wu and Gschwend (1988) obtained analytical expressions for the first-order rate constant of a two compartment model in terms of diffusion model parameters. The expressions obtained depended not only on the system’s physical characteristics but also on the duration of the sorptiondesorption experiment. The first-order rate constant must vary with time to account for the time-dependent concentration gradients within the diffusioncontrolled regime. In soil column experiments, the contact time between solute and sorbent depends on mobile phase velocity. Rao et a1 (1980b) predicted that first-order rate constants used to describe sorption in soil column studies should vary with mobile phase velocity and such results have been observed. The simplicity of first-order kinetics, the uncertainties found in experimental data, the applicability of the two-compartment model to a variety of mechanisms, and the inherent difficulty in defining soil microgeometry often make the first order model an acceptable alternative to describe sorption kinetics.

3. Irreversible Sorption: Aging and Bound Residue Formation The longer nonpolar organic compounds remain sorbed to soil, the more time and/or energy is required to fully remove them (Steinberg et al., 1987; Pignatello, 1990a,b; Pavlostathis and Mathavan, 1992). It has been suggested that this phenomenon, termed aging, may result from very slow desorption kinetics as discussed previously, bound residue formation, or physical trapping of nonpolar organic compounds in soil (Pignatello and Xing, 1996; Calderbank, 1989). Although different theories of sorption kinetics predict that the time for desorption will increase the longer a pollutant remains in the soil, ultimately all of the pollutant is assumed to desorb. Pollutants that are physically trapped or fixed by strong chemical bonds to soil particles will never, for all practical purposes, desorb without some change in soil structure or a chemical reaction. Most existing models describing sorption and desorption do not include an irreversibly sorbed pool. The best understood category of irreversible sorption is formation of strong ionic or covalent bonds, also known as bound residue formation. This topic has been reviewed by Calderbank (1989) and Khan (1982). Bound residues may be formed through ionic bonds between clay and cationic pesticides such as diquat and paraquat. Bound residues are formed through covalent bonds between organic matter and compounds such as amines and phenols, certain urea and anilide

12

K. M. SCOW AND C. R.JOHNSON

pesticides that are readily converted into aromatic amines, and phenolic and quinone residues that are created by transformation of phenoxy pesticides. Many commonly occurring extracellular oxidases and certain minerals in soil can catalyze these reactions (Scheunert and Mansour, 1992). The phenomenon of physical trapping is not well understood. Examples of chemicals subject to physical trapping include aliphatic halocarbons such as ethylene dibromide, trichloroethylene, and others (Pignatello, 1990a,b). Portions of these chemicals can be released and made bioavailable by pulverization of soil aggregates. Reduced bioavailability has also been observed for other pesticides and pollutants, such as simazine, phenanthrene, and styrene, that have had long-term contact with soil (Scribner et al., 1992; Hatzinger and Alexander, 1995; Scow et al., 1994). Many hypotheses for aging have been suggested (Pignatello and Xing, 1996);however, more research on the topic is needed. Aging, its link to natural carbon cycling, and its effect on pollutant biodegradation will be discussed in more detail in a later section. Although a theoretical distinction may be made between chemicals that are irreversibly sorbed to soil and those that have extremely slow release kinetics, experimentally it is difficult to distinguish between them. Generally, the issue is not addressed in the literature. A chemical is considered “bound” or “aged” if it is not removed from soil by a specified solvent extraction method or if a bioassay indicates that it is not bioavailable. Why the fraction remains in soil is usually not known. A second consideration is the practical relevance of the distinction. If the process of interest occurs on a time scale much shorter than a chemical’s desorption kinetics, it matters little if the chemical is irreversibly or reversibly bound.

111. BIODEGRADATION OF SORBED CHEMICALS The biodegradation of a sorbed chemical is actually a coupled process that includes a biological component, i.e., the metabolism of the chemical, and a physical/chemical component, i.e., the distribution and movement of the chemical in the physical environment in relation to the microbial population able to degrade it. The relative importance of these processes depends on how strongly sorbed and how rapidly degraded is the particular compound in a given soil. The biological component of biodegradation of sorbed chemicals is usually described very simply; however, this is only because of the limited number of kinetic expressions that have been developed to model metabolism or substrate disappearance. The most commonly used forms are the Monod (for growthlinked metabolism) and Michaelis-Menten (for non-growth-linked metabolism) equations, and the simple expressions of first- or zero-order kinetics that can be derived from both equations (Table I) (Simkins and Alexander, 1984; Alexander

EFFECT OF SORPTION ON BIODEGRADATION

13

Table I Monod-Derived Biodegradation Kinetics Model

Equation

Monod

Michaelis-Menten

dCldt =

v,,,,xc ~

(K,,, + C )

k,C

First order

dCldt

Zero order

dCldt = k,,

=

Note. C . substrate concentration (mglliter); 1. time (days); u,,,,,,maximum specific growth rate ( I /day); B. biomass concentration (mglliter biomass); Y. yield coefficient (mglliter biomass produced per mg/liter substrate degraded): K,, half saturation constant for growth (mg/liter); V,,,,,. maximum reaction velocity (mglliterl day); K,,,.Michaelis-Menten constant (mglliter); k , , first-order biodegradation rate constant ( I /day); k,,, zero-order biodegradation rate constant (mg/liter/day). Units of mglliter refer to substrate concentration unless otherwise specified.

and Scow, 1989). Whether these are the best expressions for describing metabolism has been questioned (Baveye and Valocchi, 1989); however, surprisingly few alternatives have been developed and these will not be discussed further in this chapter. It is important to recognize, however, that equations describing metabolism disregard a large number of the biological interactions that actually occur within the soil. Thus, the phenomena described are usually limited to the uptake of the pollutant (Alexander and Scow, 1989), uptake of other limiting nutrients (Celia ef al., 1989), and may include a decay rate or maintenance requirements of the microbial population (Celia et af.,1989). Not considered is the biodegradation of the same chemical by multiple microbial species, predation of biodegrading populations, competition between biodegrading and other organisms for the same resources, concentration dependency of growth and metabolism, and other issues (Schmidt, 1992). In this section, we explore different approaches for considering the effect of sorption on biodegradation rates. Effects of pollutant sorption on microbial ecology and metabolism will not be discussed except in terms of pollutant bioavailability. Models will be grouped according to how they represent sorption kinetics. This discussion will thus mirror the description of sorption kinetics

K. M. SCOW AND C . R.JOHNSON

14

Table I1 Coupled Process Models under Batch Conditions Physical model

Equilibrium sorption coupled to biodegradation Equilibrium sorption

Compartment models First-order mass-transfer rates Continuous distribution of firstorder rates Radial diffusion models Radial diffusion coupled with linear sorption

Biological model

Examples

First order or Michaelis-Menten

Mihelcic and Luthy (1991), Stem et al. (1980), Miller and Alexander (1991), Guerin and Boyd (1992), Ogram et al. (1985)

First order

Hamaker and Goring (1976), Scow et al. (1986) Gustafson and Holden (1990)

First order

First order or Michaelis-Menten

Scow and Hutson (1992). Chung et al. (1993), Mihelcic and Luthy (1991). Rijnaarts et al. (1991)

presented previously. Table I1 contains a summary of the different approaches used to model coupled sorption/desorption and biodegradation processes.

A. COUPLED PROCESS MODELSUNDER BATCHCONDITIONS 1. Equilibrium Sorption

If a constant equilibrium is maintained between a dissolved substrate available to microorganisms and its sorbed, unavailable form, sorption should affect biodegradation kinetics only by lowering the amount of substrate metabolized in each time step. This rule assumes, however, no concentration-dependent effects on enzyme induction or biodegradation rate constants. The same expressions utilized to describe biodegradation in well-mixed liquid culture (first-order, Michaelis-Menten, and Monod equations) apply in this case except that the dissolved concentration is reduced according to the applicable sorption isotherm expression. Kinetics resembling liquid culture biodegradation rates will also occur when sorption/desorption kinetics are very fast compared to microbial kinetics. Systems involving soluble compounds that exhibit relatively fast sorptioddesorption rates, sorbents with short diffusive path lengths, and low microbial metabolism rates could be expected to behave in this manner. Mihelcic and

EFFECT OF SORPTION ON BIODEGRADATION

1.5

Luthy ( I 99 I ) found that naphthalene biodegradation in anaerobic soil slurries was equally well described by a Michaelis-Menten equation, in which the solute phase was reduced according to a linear sorption isotherm, and by the more complex model coupling radial diffusion and biodegradation. Chemicals associated with colloidal organic matter or sorbed to mineral surfaces have a short diffusion pathway to reach the aqueous phase and, in such cases, their biodegradation can be well described if the solution concentration is corrected for equilibrium sorption (Steen er al., 1980; Miller and Alexander, 1991). It should be noted that chemicals bound to mineral surfaces can exhibit desorption-controlled biodegradation kinetics if chemical forces involved are sufficiently strong (Smith er al., 1992) An equilibrium sorption-biodegradation model for evaluating biodegradation kinetics in soil suspensions was developed by Guerin and Boyd (1992, 1993). Initial degradation rates were compared with predicted biodegradation rates based on the equilibrium aqueous naphthalene concentration in soil slurries. If sorbed naphthalene was unavailable to microbes and desorption was slow, initial degradation rates should equal those in soil-free cultures with the same aqueous naphthalene concentration. Rates higher than the predicted response were assumed to indicate the ability of organisms to use a portion of the sorbed pool of chemical. Another modeling approach that assumes equilibrium sorption of the pollutant is that of Ogram er al. (1985). They developed a family of simple models coupling linear sorption partition coefficients and first-order biodegradation kinetics to predict the biodegradation of a sorbed chemical under different conditions. With regard to the chemical, either the dissolved phase only or both the dissolved and sorbed phase was assumed to be metabolized. With respect to microbial population distribution, organisms were assumed to be either attached or not attached. The model that best described the biodegradation of 2,4-dichlorophenoxyacetic acid ( 2 , 4 - ~ by ) a pure culture of bacterium in soil suspensions assumed that only the dissolved phase was used and that both attached and suspended organisms were responsible for degradation.

2. First-Order Approximation to Sorption Kinetics and Two-Compartment Models As discussed previously, first-order mass-transfer expressions may be used to model sorption kinetics. In such models, the soil is divided conceptually into compartments that are considered homogeneous and well mixed. The model specifies which compartments may exchange chemicals and lists the equations controlling these exchanges. Models such as that described previously are well known in the fields of pharmacokinetics and environmental modeling and are referred to as compartment models (Godfrey, 1983). In the models considered in

16

K. M. SCOW AND C . R.JOHNSON

this chapter, transfer between compartments is either controlled by first-order rate expressions or is instantaneous if the compartments are assumed to maintain an equilibrium distribution. First-order compartment models usually describe the movement of solute between compartments by a set of differential equations (Fig. 4) for which standard methods exist to find solutions (Simmons, 1972). The resulting expression for solute concentration in any compartment as a function of time has the form of a sum of exponential terms. The parameters in the model solutions are actually complicated functions of the rate constants used in the model and the initial substrate concentrations in the various compartments. It is not always possible to solve for the model rate constants from the parameters obtained by fitting the model solution to experimental data. Figure 4 illustrates the equations and solution for the rate of product formation using the soil biodegradation model proposed by Hamaker and Goring (1976). Compartment sorption models coupled to first-order biodegradation kinetics have also been used to explain biodegradation kinetics in soil. Figure 5 presents several examples of compartment models that have been proposed to describe biodegradation kinetics in batch systems (Scow et al., 1986; Hamaker and Goring, 1976; C. R. Johnson and K. M. Scow, unpublished data). The solution for the rate of product formation has the same form for each of these models; thus, they are indistinguishable from observations of biodegradation kinetics alone. Model I (Fig. 5) couples the sorption kinetics model proposed by Karickhoff (1980) to first-order metabolism kinetics for solute in the dissolved phase. Hamaker and Goring (1976) used model I1 to describe the biodegradation rates of Triclopyr (2,3,5-trichloro-2-pyridyloxyacetic acid) in two soils. In Hamaker’s model, the dissolved and equilibrium sorbed phases of Karickhoff’s model are lumped into one labile compartment that exchanges solute with a resistant compartment according to first-order kinetics. According to Hamaker, rapid exchange between dissolved and equilibrium sorbed chemical makes all material in the labile pool available to microbes. “The rate of degradation for the labile pool is, therefore, considered as a single reaction, even though it is internally complex” (Hamaker and Goring, 1976). Hamaker and Goring chose first-order kinetics to describe transfer to and from the labile and resistant pools partially due to the mathematical simplicity of the form and partially because first-order kinetics describes systems that contain small amounts of substrate relative to large initial population densities of microorganisms. Figure 6 shows the effect on biodegradation kinetics for varying ratios of k , and k2 in the Hamaker model. If the initial pollutant concentrations in the labile and sorbed phases are known, all model rate constants may be determined by fitting the model to experimental data (Hamaker and Goring, 1976). Scow et al. (1986) used model III (Fig. 5) to describe mineralization kinetics for phenol, aniline, and nitriloacetic acid in soil. This model is similar to Hamaker’s except that biodegradation is possible from

EFFECT OF SORPTION ON BIODEGRADATION

17

Definina Eauat ions

a = klC2 - Cl(k2 + kb) dt = kzC1

-

klC2

dt Solution for m t e of Product Evolution

he = Aexp(mlt) + f3exp(m2t) dt

If at time t = 0 , P = 0 , C2 = 0 , and C1 = C, then: A = kbG(m1 + k l ) (m, - m 2 )

B = kb$(m, (m2

+ kt) - mi)

FFgure 4 Hamaker’s two-compartment biodegradation model (adapted from Hamaker and Goring, 1976). Concentrations expressed per total soil mass.

both Compartments. The compartments are not identified as labile or nonlabile prior to fitting the experimental data. As discussed previously, many different two-compartment models lead to biodegradation rates that have the form of a sum of two exponential terms. Biodegradation rates best described by this form have been reported by Hyzak and Zimdahl (1974) for three triazine herbicides at 35°C and one triazine herbicide at 20°C and by Zimdahl and Gwynn (1977) for three dinitroaniline pesti-

18

Ei-H.-.l-oduct K. M. SCOW AND C . R.JOHNSON

Aqueous

kb

Model 1, Johnson & Scow, unpublished data

Resist ant

Model II, Hamaker & Goring, 1976

-

Phase One

a1

Phase

Two

a,

Model 111, Scow et a/., 1986

Figure 5 Coupled sorption/biodegradationcompartment models (adapted from Hamaker and Goring, 1976; Scow et al., 1986, C . R. Johnson and K. M.Scow, unpublished data).

cides at 30°C. Interestingly, these pesticides exhibited first-order biodegradation kinetics at lower temperatures. The shift from first-order to two-compartment kinetics at higher temperatures (and higher degradation rates) is consistent with data generated in our lab that show a shift from first-order to two-compartment kinetics as phenanthrene degradation rates increase (C. R. Johnson and K. M. Scow, unpublished data). Smith et al. (1992) observed biphasic biodegradation kinetics for a bacterium using quinoline in a montmorillonite clay suspension. Mineralization kinetics could be described by dividing the mineralization curve into two sections each with its own first-order rate constant. Quinoline binds to clay surfaces via interactions of the molecule’s lone pair nitrogen electrons and TI elections in its

EFFECT OF SORPTION ON BIODEGRADATION

19

aromatic rings. Smith postulated that in this case chemical reaction rates, rather than diffusion kinetics, controlled desorption rates and limited biodegradation kinetics during the second section of the mineralization curve. Guerin and Boyd, (1992,1993)compared naphthalene degradation kinetics for two bacterial stains, Afcafigenes NP-ALK and Pseudomonas purida (ATCC 17484). As discussed previously, a coupled equilibrium partitioning-first-order biodegradation model was used to evaluate differences in the strains’ ability to uti-

100

k (decamp.) = 0.0152 (tl,2decomp. = 4 5 . 6 d a y s )

70

50

40

30

20

10

I

50

100

\ I

150

I

200

\I

250

I

300

J

350

Rgure 6 Effect of varying k , and k, on Hamaker model biodegradation kinetics. R = k, (binding) / k , (unbinding). k - and k , in figure equal k , and k, in text (from Hamaker and Goring, 1976, with permission of the publisher).

,

20

K. M. SCOW AND C. R.JOHNSON

lize sorbed naphthalene. Strain 17484 could utilize sorbed naphthalene, whereas strain NP-ALK could not and differences were observed in the shape of the biodegradation curves for the two species. Naphthalene mineralization in soil slurries followed first-order kinetics for strain NP-ALK, whereas a biphasic pattern was observed for strain 17484. Guerin and Boyd interpreted these results in terms of Karickhoff’s two-compartment sorption model. Strain 17484 was hypothesized to have access to naphthalene in the labile sorbed compartment, whereas strain NP-ALK did not. Utilization of the labile sorbed naphthalene stimulated desorption of resistant sorbed naphthalene that could then be utilized by strain 17484. Initial biodegradation rates for strain 17484 were also measured as a function of naphthalene-soil contact time prior to inoculation with the bacterium and these results were also consistent with a two-site sorption model. Initial rates decreased as naphthalene incubation time prior to inoculation increased; however, eventually most of the aged naphthalene was degraded. During incubation prior to inoculation, chemical was transfered from the labile to resistant sorbed phase after which, being unavailable to stain 17484, it lowered the initial mineralization rate. A consequence of the utilization of the remaining labile sorbed naphthalene by strain 17484 was the transfer of naphthalene from the resistant sorbed to the labile compartment. Guerin and Boyd (1993) also pointed out that the behavior of strains NP-ALK and 17484 is consistent with the intraparticle and intraorganic matter diffusion theories proposed to explain observed sorption kinetics in soil. In a treatment very similar to that of Connaughton et al. (1993),Gustafson and Holden ( 1990) represented pesticide biodegradation rates by assuming soil contained a continuum of spatially segregated compartments each exhibiting firstorder dissipation kinetics. A probability function, the gamma function, was used to describe the distribution of compartments having the same first-order dissipation rate constant. Specifying the two parameters controlling the shape of the gamma function determined the distribution of fast and slow degrading compartments and thus the observed dissipation kinetics. Analysis of 45 data sets of pesticide dissipation, in lab and field studies, indicated that the gamma function model usually fit better than a first-order decay model. The model reduces to a first-order decay equation when the rate constants approach uniformity. The relative variability in the rate constant was similar for data describing pesticide disappearance at the scale of both the flask and the field. Thus, the authors speculated that the length scale of the variability was at the pore scale and perhaps due to sorption and/or diffusion processes affecting the availability of the pesticide for degradation. The homogeneous well-mixed boxes portrayed in compartment models may seem to have little relationship to the heterogeneous soil environment. To relate the model to reality, one needs some method to experimentally quantitate the

EFFECT OF SORPTION ON BIODEGRADATION

21

amount of chemical in labile and resistant compartments. The measurement method itself then serves as an operational definition of these pools. Cheng ( 1990) has suggested developing chemical fractionation techniques to characterize soil organic matter and pollutant residues into pools of differing bioreactivity or bioavailability. Burford et al. (1993) suggested that differences, with regard to extraction kinetics and efficiencies using supercritical fluid extraction, between recently added and aged PAHs might be useful in exploring the locations and interactions of sorbed chemicals within the soil matrix. Several studies have attempted to relate different pools of a pollutant in soil, as defined by chemical extraction methods, to their biodegradability. For styrene in soil, pools defined included a readily desorbed, not readily desorbed but extractable, and not desorbed nor readily extractable fraction (Fu er al., 1994). Robinson et al. (1990) found that greater than 90% of toluene in soil slumes was readily extractable by water and the remaining 10% was not extractable by water and partially extractable by an organic solvent. These two pools appeared to correspond to the biodegradation kinetics with removal of an initially rapidly degraded, large fraction of the toluene, followed by removal of a slowly degradable, smaller fraction of toluene. Weissenfels er al. (1992) used solvent extraction to remove the nonbiodegradable fraction of PAHs from a high organic carbon soil. They were then able to show that the fraction was actually bioavailable, in the absence of the mass-transfer limitation imposed by the soil, as evidenced by its metabolism by a mixed culture of microorganisms in solution culture. The use of chemical fractionation techniques to measure the amount of pollutant in different soil compartments is flawed because these methods often alter organic matter structure and thus the results are difficult to interpret. Relating conceptual compartment models to measurable quantities in soil is also a challenge faced by scientists studying natural carbon and nutrient flows in soil systems. Similar methods, curve fitting and chemical fractionation, as well as density and size fractionation have been used to define natural soil organic carbon pools. Results of these studies are discussed under Section 1V.

3. Coupling Radial Diffusion with Sorption and Biodegradation Several investigations have combined radial diffusion models describing the kinetics of sorption and desorption, such as that of Wu and Gschwend (1986), with biodegradation kinetics. This approach has been used to describe the metabolism of chemicals limited by mass transfer in artificial aggregates (Scow and Hutson, 1992) and soil (Mihelcic and Luthy, 1991; Rijnaarts et al., 1991). All studies made some attempt to compare predictions using the radial diffusion and less complex models, such as those described previously, to identify conditions when the more complex approach was necessary.

K. M. SCOW AND C. R.JOHNSON

22

Coupled radial diffusion-biodegradation models (Scow and Hutson, 1992; Mihelcic and Luthy, 1991) describe biodegradation kinetics in a saturated slurry containing spherical porous aggregates. Figure 7 depicts the processes and their spatial locations within the aggregate-solution system (Mihelcic and Luthy, 1991). The assumptions about diffusion and sorption have been described previously. Microorganisms, and thus biodegradation, are assumed to occur only outside of aggregates in a well-mixed outer solution; evidence for this assumption is discussed below. Biodegradation is described by the Michaelis-Menten or Monod equations, which can reduce to first-order kinetics. Simulations using coupled radial diffusion-biodegradation models have successfully described experimental data in several cases (Scow and Alexander, 1992). The biodegradation of phenol by a pure culture of Pseudomonas sp. in the absence and presence of two sizes of clay aggregates was measured in an experimental system in which all input parameters could be determined by measure-

metabolism of aqueous-phase

associated solute;

solute sorbed along micropore,

solute in micropore

solute in mocropore

I Rgure 7 Multiple processes occurring during the biodegradation of a chemical distributed between soil solution and an aggregate. Microorganisms are excluded from the pores inside the aggregate and a local equilibrium exists between the solute in the aggregate and that sorbed onto micropore surfaces (from Mihelcic and Luthy, 1991, with permission of the publisher).

EFFECT OF SORPTION ON BIODEGRADATION

23

ment. Using measurements of the mass-transfer coefficient, first-order rate constant, and the physical system, model simulations compared well to experimental data for both sizes of clay aggregates and at two initial population densities. Model simulations, again using independently measured input parameters, also compared reasonably well to data describing phenol biodegradation by a pure culture of Pseudomonas sp. in the presence of different amounts of polyacrylamide gel exclusion beads that sequestered part of the solution from bacteria. Model simulations were performed to identify conditions under which Scow er a l . 3 (1986) two-compartment model or a first-order model could adequately describe biodegradation data generated by the more complex coupled radial diffusion-biodegradation model (Scow and Hutson, 1992). With a small aggregate radius (0.05 cm) and low sorption partition coefficient ( 5 1 0 dm3 kg-I), first-order kinetics provided a reasonable fit to the biodegradation data in the presence of aggregates. In the presence of larger aggregates (0.25 cm), even at a low sorption partition coefficient, biodegradation curves were clearly biphasic and better fit by the two-compartment rather than the first-order model. Biodegradation was overestimated when long-term predictions of the chemical’s persistence in 0.25-cm aggregates were calculated based on a first-order half-life fit to the first part of the biodegradation curve. Half-lives in the presence of aggregates were substantially greater than equivalent half-lives in the absence of aggregates when the sorption partition coefficient was 2 3 . 2 and the aggregate radius was 20.32 cm. The radial diffusion-biodegradation model developed by Scow and Hutson (1992) was adapted to include nonlinear conditions for adsorption and a masstransfer term across the surface of the aggregate (Chung et al., 1993). Also, for use in describing the behavior of the model under different conditions, four dimensionless groups were derived from model equations and are presented in Table 111. Sensitivity analyses were performed using the various dimensionless groups to determine the dependence of a chemical’s biodegradation kinetics on the chemical, physical, and biological parameters comprising the model (Chung et al., 1993). The overall rate of biodegradation of a chemical is a function of (i) its rate of mass transfer across the surface of the aggregate, as determined by the number of aggregates, their surface area, their volume, and the mass-transfer coefficient; (ii) its diffusion within the aggregate, as determined by its sorption partition coefficient, diffusion coefficient, the aggregate’s radius and internal porosity, and the tortuosity term; and (iii) its intrinsic rate of biodegradation. Different shapes of biodegradation time-course curves were observed depending on whether the reaction was controlled by mass transfer or the reaction rate as well as on whether the chemical started out sorbed (inside the aggregate) or not sorbed (in the outside solution) (Fig. 8). As seen from Table 111, an increase in $ can result from an increase in sorption

K. M. SCOW AND C. R. JOHNSON

24

Table 111 Dimensionless Parameters Describing Coupled Radial Diffusion/Biodegradation Model(’ Dimensionless group

Definition J[E

+ K(lD;

Rk,

Dc U

kRZ

Diffusive resistance Biological rxn Transfer velocity into aggregate Biological rxn

11

Bi

E)]

Comment

Film diffusion lntraaggregate diffusion Initial chemical distribution within system

Note. E , aggregate porosity; K. adsorption partition coefficient (cm3/cm3 solid); R . aggregate radius (cm); k. biodegradation first-order rate constant; D,. effective diffusion coefficient in aggregate pores (cm2/min); n . number of aggregates per volume liquid (cm-3); v p , aggregate volume (cm2);a,,, surface area per volume aggregate (cm- 1); k,., mass-transfer coefficient from aggregate to external fluid (cmlhr); C,,, initial chemical concentration inside aggregate ( g i c d ) ; C,,,, initial chemical concentration outside aggregate (g/cm3). “From Chung e r a / . (1993).

partition coefficient or aggregate radius, or from a decrease in the diffusion coefficient (Fig. 8). If biodegradation is rapid relative to the mass-transfer rate (q = O.l), there is little impact of sorption on the rate of degradation of a chemical starting out in the solution phase (u = a);however, sorption is more important if the chemical is initially completely sorbed (u = 0). For chemicals starting out in solution and with slower degradation rates relative to the mass-transfer rate (q = l.O), initially there is a brief and rapid rate of biodegradation of the chemical available in solution. Once most of the solution concentration has been depleted by biodegradation or diffusion into the aggregate, there is a much slower rate of biodegradation controlled by mass transfer of the chemical back out of the aggregate. These kinetics are similar to the biphasic mineralization curves described by two-compartment kinetics, as discussed previously. Scow and Hutson (1992) found that curves of this shape could be fit by a two-compartment model. When the chemical starts out entirely in the sorbed phase, an increase in the sorption partition coefficient changes the kinetics of biodegradation from first order to zero order, at least within the time period shown on the graph.

1 .o

0.8 0.6 0.4

0.2

0.0

0 , dimensionless time Figure 8 Dimensionless mineralization (@) vs dimensionless time (8)for different values of dimensionless parameters Q.q, and u. Bi value fixed. Coupled radial diffusion/biodegradationmodel with an additional diffusive resistance at the aggregateiouter solution boundary. @, fraction substrate mineralized; 8 = kr; k, first-order biodegradation rate constant (l/day). See Table I1 for dimensionless parameter definitions (from Chung et al.. 1993, with permission of the publisher).

26

K. M. SCOW AND C . R.JOHNSON

Mihelcic and Luthy (1991) used a coupled radial diffusion-biodegradation model to describe the biodegradation of naphthalene under denitrification conditions in soil-water suspensions under well-defined conditions. As mentioned previously, predictions of the more complex model were compared to a simple model coupling equilibrium desorption with Michaelis-Menten kinetics. Both models could describe the experimental data and this indicated that intraparticle diffusion of naphthalene was rapid compared to the rate of microbial degradation. Rijnaarts et al. (1991) used a radial diffusion model coupled to first-order biodegradation kinetics to evaluate data describing the effect of desorption and intraparticle mass transfer in a well-mixed slurry containing soil from a waste site that had been contaminated with a-hexachlorocyclohexane 20 years previously. Desorption and biodegradation were measured in a stirred (well-mixed) bioreactor and end-over-end (less well-mixed) bioreactor. Because the starting conditions for this study were unknown, unllke the more controlled studies described previously, it was not possible to directly measure many of the model input parameters. Therefore, some values were obtained by curve fitting or calculated from published values. Desorption kinetics in the stirred reactor could be described equally well by the radial diffusion model or a simple mass-transfer equation; however, desorption in the less well-mixed system could only be described by the radial diffusion model. The coupled radial diffusion-biodegradation model described biodegradation data in both systems; however, suprisingly, a first-order decay equation fit the biodegradation data equally well. Penetration of bacteria into the aggregates may have occurred that would decrease the diffusion path length, lessening the importance of mass-transfer limitations, and possibly resulting in first-order kinetics being sufficient to describe the phenomenon. To better understand interrelationships between soil structure and microbial processes in soil, Priesack (1991) derived a radial diffusion model simulating solute diffusion and biodegradation in which microorganisms are present inside but not outside of the aggregates. The model equations could be solved analytically for the case of linear adsorption and logarithmic growth of microorganisms. In model simulations, he found that the diffusing chemical did not reach the centers of aggregates because it was depleted in the outer layer by biodegradation. Also, the sizes of the unaffected centers increased as the microbial growth rate increased, as the diffusion coefficient decreased, or as the sorption partition coefficient was decreased. Priesack and Kisser-Priesack (1993) compared model simulations to experimental data describing glucose mineralization in synthetic soil aggregates of 1.8-cm diameter. Patterns of chemical disappearance as well as biomass distributions supported the idea that the inner zone of the aggregate was an area of low activity and population density, which the model explained as being due to depletion of glucose before it reached the center.

EFFECT OF SORPTION ON BIODEGRADATION

27

Dhawan et a f . (1991, 1993) developed a radial diffusion model coupling diffusion, sorption, and biodegradation in soil for use in site characterization and screening of bioremediation treatment strategies. The model assumed that microorganisms are present inside of the aggregates; however, there was no evidence provided to support this assumption. Biodegradation was described by Monod kinetics in which the reaction was dependent on chemical, biomass, and oxygen concentrations. Model simulations for the case in which the chemical was initially distributed inside the aggregate indicated that most of the chemical was degraded before it could diffuse outside of the aggregate. The region within the aggregate sustaining substantial microbial activity coincided with regions containing sufficient concentrations of oxygen and pollutant substrate. This active region moved toward the center of the aggregate over the course of biodegradation.

4. Dissolution from Solids and NAPLS Similar to the approaches used to model sorption-limited and other types of mass-transfer-limited biodegradation, coupled process models have also been formulated to describe biodegradation of pollutants in solid form or dissolved in nonaqueous phases such as organic solvents. The transfer of phenanthrene from crystalline solid form, or from silicone oil or heptamethylnonane, to a solution in which it could be degraded by Pseudomonas sp. was measured and compared to predictions using a coupled model (Bouchez et a f . , 1995). Growth was measured by oxygen consumption that was found to be representative of phenanthrene disappearance, biomass increase, and carbon dioxide evolution. There was an initial phase of exponential growth with the same specific growth rate for the bacteria in all three systems. The second phase of growth was system dependent and virtually identical in rate to the masstransfer rate determined under abiotic conditions for each system (Bouchez et al., 1995). Volkering et al. (1992) developed a model describing biodegradation of PAHs in solid form or in a sorbed phase. They described growth by a Monod expression and dissolution by a first-order mass-transfer equation, as described previously. The model predicted that with a low initial cell density, growth would first be logarithmic until PAH availability became limited by mass transfer from the solid to solution phase. When the cell density was high (e.g., after growth), the model predicted linear growth that could not exceed the mass-transfer rate. Experiments with a mixed bacterial culture in batch systems growing on crystalline naphthalene found that bacterial growth was initially exponential and then linear, as the model predicted. Unfortunately, mass transfer of the chemical from the solid to solution phase was not measured to test whether the linear rate was proportional to the mass-transfer rate. The rate of the second phase of growth

28

K. M. SCOW AND C. R. JOHNSON

was found to be inversely proportional to the diameter of naphthalene particle and thus appeared to be related to surface area; however, this was not tested using model calculations. For all three PAHs, the maximum specific growth rate was proportional to the saturation concentration (in mineral salts medium).

B. COUPLING ADVECTION, SORPTION, AND BIODEGRADATION COLUMNSTUDIES The studies described previously concern biodegradation and sorption occurring in batch systems in which it is often difficult to measure biodegradation and abiotic processes independently. In column studies, it is possible to determine the mass-transfer kinetics independently of biodegradation by measuring breakthrough curves in sterile soil columns and by observing the behavior of nonsorbing tracer chemicals (typically 3H,O and C1-) in the system. An in-depth discussion of coupled advection/sorption/degradationmodels is beyond the scope of this review. In the same manner that first-order biodegradation kinetics is coupled to abiotic sorption models, most models used to predict biodegradation in soil column studies couple first-order biodegradation kinetics to existing two-site and two-region models described previously. Table IV presents a list of some studies that have compared experimental results to coupled

Table IV Column Studies Utilizing Coupled Advection/Sorption/BiodegradationModels Chemical used 2.4.5-T"

Sorption model Two region

Biodegradation model First order

2,4,5-T"

Multiprocess First order nonequilibrium Alkylbenzenes Two site First order

Quinoline

Nonequilibrium"

First order

2.4-0'

Two site

First order or Monod

Comment

Reference

Biodegradation rate constant fitted No parameters fitted

Gamerdinger er a/. (1990) Brusseau er a / . ( 1992) Biodegradation parame- Angely er a / . ters from mass( 1992) balance approach Biodegradation parame- McBride er a / . ters from mass bal(1992) ance approach Saturated and unsatuEstrella er a / . rated systems ( 1992)

<'Data obtained from van Genuchten ( 1974). 2.43-T, 2,4,5-trichlorophenoxyaceticacid. "References indicate authors used eigher a two-site or two-region model. ' 2.4-0. 2.4-dichlorophenoxyacetic acid.

EFFECT OF SORPTION ON BIODEGRADATION

29

models to predict biodegradation in soil columns. Rao et al. (1993) and Brusseau et al. (1992b) contain more extensive lists of model formulations for column and field scale simulations. Closely related are models coupling abiotic pollutant transformation kinetics with sorption and advection processes. The reader is referred to Gamerdinger et al. (1993) and references therein for more information. The investigations listed in Table IV illustrate or discuss several problems with modeling studies that may apply to batch as well as column experiments: 1. The same data can be fit by several different conceptual models; thus, model validation requires input parameters to be determined in experiments independent of the data being simulated (Brusseau et al., 1992; Gamerdinger et al., 1990). 2. Fitted biodegradation parameter values may be dependent on the equations and processes used to simulate sorption-diffusion kinetics (Brusseau et af., 1992; Gamerdinger er al., 1990; Estrella et al., 1993). 3. Biological parameters may be system dependent (batch vs column, fast mobile phase vs slow mobile phase) if experimental conditions unexpectedly alter microbial physiology or, stated more generally, the physical or biological models used do not adequately approximate the systems under study (Angley et al., 1992; Estrella et al., 1993; McBride et al., 1992).

C. EXTRAPOLATION OF MICROBIAL RATE PARAMETERS ACROSS EXPERLMENTAL SYSTEMS Important biological considerations in the modeling of coupled processes are (i) whether biological properties measured in solution culture, in which such measurements can be easily obtained, are applicable to soil; and (ii) whether parameters measured in batch systems can be extrapolated to column systems. Van Loosdrecht et al. (1990) reviewed literature comparing microbial parameters measured in the absence (e.g., solution culture) and the presence of surfaces (e.g., artificial surfaces, sediment, and soil). The authors concluded that indirect effects (e.g., on solution chemistry) resulting from microenvironmental changes near surfaces, rather than direct effects of surfaces on microorganisms, explained kinetic parameter differences between microorganisms that are unattached versus those that are attached to surfaces. Most of the studies reviewed by van Loosdrecht er al. (1990) were conducted in artificial or aquatic systems. Similar comparative studies are needed for surface-free versus soil systems; however, current methods are neither sufficiently sensitive nor nondestructive to permit measurements of microbial processes at the scale of the soil pore, where substrate mass-transfer limitations are likely to be minor.

30

K. M. SCOW AND C. R. JOHNSON

With regard to trying to estimate microbial parameters from batch systems for use in coupled process models, both Harms and Zehnder (1994) and Focht and Shelton (1987) found that the maximum growth rate was similar for organisms growing in solution or attached to surfaces (e.g., beads or soil, respectively). The apparent half-saturation constant, however, was higher for the attached organisms, which both studies attributed to substrate mass-transfer limitations. Scow (1993) calculated, based on model simulations using a radial diffusion model, how much the apparent half-saturation constant (K,) increased as a function of diffusion path length or sorption partition coefficient. Thus, in a well-defined physical system, it may be possible to correct the biological parameters for the influence of substrate mass transfer. For a soil system, it is not easy to isolate the biological from the physical processes governing the overall rate of biodegradation; thus, batch-derived biodegradation rates must be used with caution.

IV. ANALOGIES BETWEEN SORBED CHEMICAL DEGRADATION AND CARBON FLOW IN SOIL It is well known that, in soil, organic compounds are present in multiple fractions varying in their bioavailability, and these fractions range in persistence from weeks to thousands of years and in form fiom newly added plant material to soil humus (Jenkinson and Rayner, 1977; Stevenson, 1994). Inasmuch as the same fundamental processes govern the chemistry of all molecules in soil, regardless of whether the molecules are indigenous or xenobiotic, consideration of native carbon cycling in soil may provide insights into the bioavailability of pollutants. A large portion of native carbon has a low bioavailability due to its chemical complexity and, more important, physical protection in the soil matrix (Adu and Oades, 1978; Anderson et al., 1981; Elliot, 1986). As with sorbed pollutants, a major issue for native carbon is whether the various rate-defined compartments can be measured and what factors determine the size of each compartment (Stevenson, 1994). In carbon flow studies, researchers have quantified the amount of carbon in different compartments using a number of methods. These methods include modeling approaches that use curve fitting to obtain unknown parameters, chemical fractionation based on solvent extractions used either sequentially or independently, and physical fractionation by size or density (Stevenson and Elliot, 1989). Models describing carbon flow in soil bear a strong resemblance to the compartment models used to describe biodegradation of sorbed chemicals. As is the case for sorbed chemicals, multicompartment models, with a first-order decay expression for each compartment, are used to simulate carbon turnover in soil

EFFECT OF SORPTION ON BIODEGRADATION

31

(Woodruff, 1949). Two-compartment models, which consider a readily degradable and slowly available pool of carbon, are commonly used to estimate shortterm (at the scale of a growing season) mineralizable elements such as nitrogen (Jenkinson and Rayner, 1977). Models consisting of four to five compartments describe carbon Aow in soil over long time periods-on the order of decades (Jenkinson and Rayner, 1977; Parton et al., 1988; Van Veen and Paul, 1981). These compartments include readily degradable pools, such as decomposable plant material and microbial biomass, as well as physically stabilized and chemically stabilized organic matter. Direct chemical attachment of native organic compounds to reactive sites on colloidal organic surfaces and incorporation into structures of newly formed humic and fulvic acids occur during the humification process (Stevenson, 1976). Many pesticides and their decomposition products may become part of the pool of precursor molecules involved in humus formation. In doing so, these pesticides may lose their chemical identity. Many are basic and can form a chemical linkage with the C=O groups of organic matter, e.g., pesticides with amino groups. If they have a C = O group themselves, then they can react with amino groups in soil (Stevenson, 1976). Dec and Bollag (1988), using a process that they term “incorporation,” have synthesized humic polymers from phenolic precursors, including chlorophenol. They distinguish incorporated from surfacebound pollutants-the latter resulting from bonds formed after incubation of the pollutant in the presence of previously synthesized humic material. Dec et al. ( 1990) found dichlorophenol that was surface bound rather than incorporated into humus was mineralized more rapidly, although rates were slow for both pools. Dichlorophenol was mineralized more rapidly when it was bound or incorporated into fulvic compared to in humic acids. They attributed the higher mineralization rates to the greater propensity for microorganisms to degrade fulvic compared to humic acids (Dec et al., 1990). Yet another similarity in the behavior of indigenous and xenobiotic chemicals is that pollutants may become physically trapped by the same mechanisms responsible for trapping indigenous organic constituents between clay, other mineral constituents, and organic matter (Calderbank, 1989). Physical fractionation approaches are common in studies of carbon Bow in soil and include measurement of carbon associated with aggregate sizes of different diameters or, on a smaller scale, associated with sand-, silt-, and clay-sized particles. Older and more biologically resistant organic matter is largely associated with silt-sized fractions, whereas more rapidly turning over material is associated with sandand clay-sized fractions. Also, microbial populations appear to be most strongly associated with the clay-sized fraction. Densitometric methods also suggest that part of the reason for the recalcitrance of certain organic matter components is their physical association with the mineral fraction of the soil (Stevenson and Elliot, 1989).

32

K. M. SCOW AND C . R.JOHNSON

V. FACTORS CONTROLLING COUPLED BIODEGRADATION AND SORPTION Most studies concerning the biodegradation of sorbed chemicals are empirical and do not independently consider the individual processes involved. The purpose of this section is to examine published data in the context of the theoretical concepts discussed under Section III,A,3. We will consider how changes in the sorption partition coefficient, diffusion rate coefficient, the intrinsic biodegradation rate, spatial distribution, temporal patterns, and water content affect the biodegradation of sorbed chemicals. In most cases, data are not sufficiently quantitative nor is there enough information provided to evaluate data within existing conceptual frameworks (e.g., radial diffusion models coupled with sorption and biodegradation). Thus, we will consider general trends in data rather than specific quantities.

A. SORPTIONPARTITION COEFFICIENT There are several ways in which the effect of changes in the sorption partition coefficient can be tested. One can artificially manipulate the sorbent concentration for the same soil and chemical or vary the soil type for a single chemical. The chemical itself can also be varied to give different sorption partition coefficients; however, it is difficult to isolate sorption effects from changes in metabolism kinetics when different chemicals are compared. Although artificial sorbents obviously differ from soil organic matter, their sorption properties are defined and sorbent quantity and affinity can be controlled if they are added to soil in sorption studies. Activated carbon has been used as an artificial sorbent. Biodegradation rates of maleic hydrazide (Helweg, 1975), diallate (Anderson, 1981), and atrazine and chlorthiamid (Moyer et af.,1972) in soil decreased with increases in the amount of added activated carbon. Chromatography beads are well-defined model sorbents that provide a wide choice of sorbent composition and sorption partition coefficients. Organic chemicals mixed with hydrophobic SM-7 chromatography beads (Bio-Rad) exhibit Kp’s that are two or three orders of magnitude greater than those in soil (K. M. Scow, unpublished results). Addition of SM-7 beads to a silt loam soil did not change the initial rate of mineralization of p-nitrophenol; however, it led to a substantial decease in the second phase of mineralization and in the percentage of chemical mineralized at 50 h (Scow, 1993). The curve shapes were consistent with the results predicted when the sorption partition coefficient is increased in a system in which the chemical is added outside of the aggregates and there is a high intrinsic biodegradation rate (shown in Fig. 8).

EFFECT OF SORPTION ON BIODEGRADATION

33

Studies in which biodegradation of a chemical is measured in soils differing in their sorption properties are more difficult to evaluate because of the many other factors that vary among the soils. Weissenfels et al. (1992) found that biodegradation rates and the biotoxicity (assayed using Microtox) of PAHs in soil that had been contaminated for some time were far greater in a soil with 1% than 13.6% organic carbon. The biotoxicity of the low organic carbon soil could be reduced in proportion to amounts of activated carbon added to the soil, suggesting that bioavailability may have some role in reducing the toxicity. PAHs extracted by soxhlet from the high organic carbon soil could be biodegraded by a mixed culture of PAH-degrading organisms and thus indicated that the chemical in the soil was not intrinsically refractile. The hypothesis that there is a relationship between sorption and biodegradation was tested in six soils (Scow et al., 1994). The biodegradation of a low concentration (50 ng per gram) of phenanthrene by indigenous microbial populations was measured in soils with organic carbon contents ranging from 3.6 to 9.5%. A low concentration was selected to minimize the influence of growth on phenanthrene biodegradation kinetics and high organic matter soils were selected to ensure substantial sorption of the chemical. The rates of biodegradation varied by soil type, with the highest rates observed in the two soils with the lowest and highest organic carbon content. Measured K, values for phenanthrene in the six soils varied by a factor of 10, whereas K , values varied by a factor of 2. Linear regression analysis showed no relationship between the biodegradation rate and either K , or K, for the soils tested. Other factors, particularly differences in the intrinsic biodegradation activity of the indigenous populations, appear to contribute to the observed differences in rate (C. R. Johnson and K. M. Scow, unpublished results).

B. DIFFUSION The diffusion characteristics of a system can be varied by changing the diffusion coefficient (e.g., by changing the chemical species) or by changing the diffusion path length (e.g., by manipulating aggregate size or tortuosity constant). The intrinsic chemical diffusion coefficient does not vary much among organic chemicals, except for very large polymers that are also slower to degrade, so the diffusion coefficient is not likely to be important in controlling biodegradation among most chemicals. Where size may be important is in diffusion within organic matter. Brusseau and Rao (1991) found an inverse relationship between molecular size and diffusion coefficient. The diffusion path length can influence biodegradation rate. In a defined experimental system in which the mineralization of phenol by a pure culture of Pseudomonas sp. was measured in the presence of kaolinite beads, Scow and

34

K. M. SCOW AND C. R. JOHNSON

Alexander (1992) found that increasing the diameter of clay beads led to a decrease in the phenol biodegradation rate. The decrease in rate could be directly attributed to a reduction in the mass transfer of phenol from the beads to the solution. Additional studies in which the diffusion path length is manipulated by altering the aggregate diameter are described under Section VI.

C. RATE OF METABOLISM The intrinsic biodegradation rate can affect the overall biodegradation of sorbed chemicals when its magnitude is high relative to the rate of chemical transfer across the aggregate surface or to the rate of chemical diffusion within the aggregate (Table 111). As seen in the curves shown in Fig. 8, the degree to which the intrinsic biodegradation rate can impact biodegradation depends on whether the chemical is already inside the aggregate and on how strongly the chemical is sorbed. Increases in the intrinsic biodegradation rate can be achieved directly by increasing the population density of the biodegrading organism, and, in some cases, indirectly by other means. Indirect means may be the addition of nutrients and electron acceptors when they limit the reaction, providing cosubstrates for cometabolized substrates, or by adding a microbial isolate with a higher rate of biodegradation than the organisms already present. There is a strong temporal interaction between biodegradation and desorption. Enhancement of total biodegradation can occur if there is an opportunity for substantial biodegradation before a large portion of a chemical becomes sorbed. For chemicals that have been present in soil for long time periods, a higher microbial population density may not stimulate biodegradation. 1. Microbial Population Density

There can be a strong feedback between growth of microbial populations and chemical desorption rates. Greer and Shelton (1992) found that concentrationsof 2 , 4 - ~a, relatively soluble chemical, in soil pore water were 20 times lower in a high organic compared to a low organic soil due to higher sorption in organic and greater dilution in the organic soil because it holds more water at field capacity. They attributed slower biodegradation of 2 , 4 - ~in part to a limited pool of soluble chemicals preventing rapid growth of the biodegrading population. This would in turn result in a lower concentration gradient between the sorbed and soluble phase and therefore desorption would not be very great. They concluded that inoculation may be unsuccessful for degradation of more strongly sorbed chemicals because the initial concentration of the pollutant in the soil solution may be too low to support growth or growth rates may not exceed death rates. High population densities can affect biodegradation of sorbed chemicals by

EFFECT OF SORPTION ON BIODEGRADATION

35

forming large substrate depletion zones in areas of high activity that, in turn, can steepen the diffusion gradient (Harms and Zehnder, 1994). In a study of the biodegradation of p-nitrophenol, dichlorophenol, and pentachlorophenol in granular activated carbon systems, Speitel et d. (1989) attributed the slow degradation of pentachlorophenol to its slow rate of desorption because of its inability to support much microbial growth. The compound, p-nitrophenol, however, supported high microbial populations. The high populations rapidly consumed the solution phase of the chemical leading to a steep concentration gradient between the activated carbon and bulk solution. Thus, chemical desorption rates were high, which, in turn, could provide enough chemical to support high rates of biodegradation. Similarly, in model simulations, Dhawan et al. (1991) found that the time for bioremediation of pollutant contained inside soil aggregates increased with decreasing microbial biomass largely due to the impacts of low biomass on mass-transfer rates. Population density is important in controlling the biodegradation rate of phenanthrene in soil, even in the presence of strong mass-transfer limitations. There was a direct relationship between population density and biodegradation rate of a pure culture of Arthrobacter sp. in solution culture, as well as when inoculated into soil slurries, sterile, and nonsterile unsaturated soil (data not shown). In Forbes soil, with a very low rate of intrinsic biodegradation and high sorption partition coefficient (Scow et al., 1994), there was a substantial increase (approximately 200 times) in the maximum rate of biodegradation when inoculated with Arthrobacter sp. at a density of 107 cells per gram of soil. This enhancement suggested that reaction rate rather than sorption was limiting biodegradation in Forbes soil. In contrast to Forbes soil, Yolo and Tinker soils have relatively high rates of intrinsic biodegradation (Scow et al., 1994). Neither Yolo nor Tinker soils showed enhanced phenanthrene biodegradation rates when inoculated with Arthrobacter sp. at a population density of 107 cells per gram soil, but did exhibit increased rates when inoculated with 109 cells per gram (Fig. 9). The population density of biodegrading organisms may influence the biodegradation rates of even chemicals that form bound residues. You and Bartha ( 1982) found that freshly added radiolabeled 3,4-dichloroaniline, which forms bound residues in soil, showed stimulated I4CO2 evolution when aniline was added. The stimulation was attributed to enhancement of the dichloroanilinedegrading populations that presumably could metabolize the aniline; however, this was not explicitly demonstrated. The shapes of the curves changed from a first-order shape (without aniline) to the S-shaped curves characteristic of growth (with aniline), and the stimulation was proportional to aniline concentration. A possible kinetic interpretation of the results is that the initially rapid growth of the aniline-stimulated microbial population enabled them to intercept a larger portion of the dichloroaniline (before it formed bound residues) than could the presumably lower population in the unstimulated soil. However, this may not be the

K. M. SCOW AND C . R.JOHNSON

36 60

40

20

0

0

200

400

600

800

Hour Figure 9 Effect of bacterial cell density on phenanthrene biodegradation in Tinker (0) and Yo10 (0)soils. Numbers refer to the number of cells initially added per gram of soil.

only mechanism involved. Stimulation of dichloroaniline biodegradation from added aniline was also observed in ' T O 2 evolution from a mixed bacterial culture using radiolabeled dichloroaniline that had been previously complexed with humic acid (You and Bartha, 1982).

2. Differences among Organisms The importance of differences in the ability of microorganisms to access and degrade sorbed chemicals has been largely overlooked until recently (Guerin and Boyd, 1992, 1993). Guerin and Boyd (1992, 1993) showed that Pseudomonas putida (ATCC 17484) and an Alcaligenes sp. (strain NP-Alk) differed in their ability to utilize naphthalene sorbed to soil particles as measured by the organisms' mineralization kinetics in the presence of different amounts of soil in suspension cultures. The P. putida could use naphthalene rapidly regardless of the amount of soil present. The Alcaligenes sp., on the other hand, showed decreases in its rate and extent of naphthalene mineralization with increases in the amount of added soil. Plots of initial mineralization rates against equilibrium aqueous concentrations of naphthalene indicated that Alcalignes was mineralizing naphthalene at a rate below what the aqueous concentration could potentially support, whereas P. putida was mineralizing at a rate above what the aqueous concentration alone could support. This suggested that the pseudomonad had access to some of the sorbed phase of

EFFECT OF SORPTION ON BIODEGRADATION

37

naphthalene. Although there were differences in physiological and ecological traits of the two organisms, no specific mechanism was provided for the observed difference. Measured values of K,,, for naphthalene were 21 1 and 91 ng ml-1 for the Pseudomonas and Alcaligenes, respectively (Guerin and Boyd, 1993). The Alcaligenes sp. strongly attached to soil and the authors speculated that this may have resulted in blocking of the organism’s membrane-associated transport sites for naphthalene. In a later paper, the pseudomonad was found to maintain naphthalene degradation activity at low levels for long periods of time in the absence of naphthalene and could rapidly respond to new additions of naphthalene to culture (Guerin and Boyd, 1995). The Alcaligenes sp., on the other hand, was unable to maintain naphthalene activity in the long-term absence of naphthalene and needed many hours after exposure to naphthalene before degradation activity was induced. The implications of these behaviors for the utilization of sorbed chemicals were not discussed. Two strains of Alcaligenes sp. with similar maximum growth rates but with a 10-fold difference in their K , constants were compared with respect to their ability to degrade 2 , 4 - ~in soil (Greer and Shelton, 1992). At concentrations of 60 and 600 pg 2 , 4 - ~ml-1 of soil solution, there was no difference in the two organisms’ rates of biodegradation. However, at 8 pg 2 , 4 - ~ml-1, strain MI, which had the lower K , of the two strains, could use the pesticide more rapidly than the other organism (strain 155). This low 2 , 4 - ~concentration was below the K , value for strain 155 but still above that of strain MI. The authors speculated that an organism’s kinetic parameters, particularly the half-saturation constant, may influence its ability to use the very low solution concentrations of sorbed chemicals in soil (Greer and Shelton, 1992). There may be direct feedback of reduced substrate availability on microbial physiology. Under nutrient-limited conditions, organisms may induce high-affinity uptake mechanisms for the limiting nutrient, which can be energetically costly to the organism (Tempest and Neijssel, 1978). Examples of high-affinity uptake systems include synthesis of binding polymers, chelating agents, and extracellular enzymes. Other changes can include dramatic alteration of metabolic pathways and their regulation, e.g., enzyme induction, that may be manifested in changes in biodegradation kinetics. Efroymson and Alexander (1995) found unexpectedly low mineralization rates for a Pseudomonas sp. using phenanthrene in the presence of an organic solvent. The authors speculated that the solvent phase may have lowered the dissolved phenanthrene to a concentration so low that enzyme induction was impaired. Also, the particular maintenance costs of an organism may determine how long it can survive for long periods of time on very low substrate concentrations. The physiology of microorganisms that metabolize very low concentrations of pollutants has been largely unexplored. Attachment may be important in the biodegradation of sorbed chemicals and

38

K. M. SCOW AND C. R. JOHNSON

microorganisms differ in their ability to attach to solid phases (van Loosdrecht et al., 1990). Attachment is clearly necessary for many cellulolytic organisms that need close proximities with their substrate for the functioning of membraneassociated, extracellular enzyme systems. Attachment of cells directly to the solid phase of insoluble pollutants may also be essential for biodegradation of these substances (Rosenberg and Rosenberg, 1981; Efroysom and Alexander, 199I). The mechanism of attachment may be through hydrophobic interactions between the organism and its substrate, production of extracellular emulsifying agents (Rosenberg and Rosenberg, 1981), or via specific organs such as pili (van Loosdrecht et a l . , 1990). The majority of microorganisms living in soil are known to be attached to soil particles as evidenced by the difficulty in extracting cells from soil (Ramsay, 1984). It is not known whether biodegrading organisms can selectively attach to pollutants that are sorbed to the surfaces of soil particles. The hydrophobicity of microorganisms may impact not only their ability to attach to solid phases but also may influence their uptake of pollutants. Hydrophobic chemicals can partition into the cell membranes directly from the solution phase (Button et al., 1992; Harms and Zehnder, 1994). Differences among organisms in their propensity to act as sorbents for pollutants may affect their ability to utilized hydrophobic chemicals. For a Sphingornonas sp., Harms and Zehnder (1994) calculated a partition coefficient between its membrane and the outside solution of 25,000 ml g-I, a value that exceeded the molecule’s octanolwater partition coefficient. Partitioning into the cell membrane can actually result in a reduction in the apparent half-saturation constant (lu,)as was observed for bacteria using toluene in a well-mixed system (Button et al., 1992). Actively metabolizing cells are surrounded by zones in which substrate concentration is reduced relative to the bulk solution (Harms and Zehnder, 1994). These zones result when diffusion cannot support substrate at the rate at which the microorganism can degrade it. The concentration gradients established by these depletion zones drive, in turn, diffusive transport. Another potentially important difference among organisms is their ability to produce emulsifiers and surfactants, which may affect their ability to attach to the free phase of the chemical they are degrading (Rosenberg and Rosenberg, 1981) or may aid in uptake of the chemical. Goswami and Singh (1991) and Reddy et al. (1983) have found certain pseudomonads produce emulsifying and solubilizing factors that can increase the uptake and metabolism of hydrocarbons. The process responsible, called “pseudosolubilization,” involves the formation of small micelles of the pollutant by the action of surface-active compounds produced by the organisms. Koch et al. (1991) found that extracellular rhamnolipids were responsible for the ability of a pseudomonad to grow on hexadecane as a sole carbon source.

EFFECT OF SORPTION ON BIODEGRADATION

39

D. SPATIALDISTRIBUTION AT THEPORE AND AGGREGATE SCALE 1. Chemical Distribution The microscale spatial distribution of chemicals in soil is largely unknown. The distributed reactivity model of Young and Weber (1995) suggests that organic matter is nonuniform with respect to chemical sorption. Several studies have measured sorption of chemicals in different particle size fractions of soil and sediment. Nkedi-Kizza et af. (1983) found that Freundlich adsorption coefficients for diuron and 2,4,5-trichlorophenoxyaceticacid (2,4,5-T) were much greater for the fine, clay-sized fraction than for the coarser, sand-sized fraction. Similar results were reported for polyaromatic hydrocarbons and methoxychlor in sediments (Karickhoff et af.,1979). To improve sorption model predictions, it has been suggested that different K , values be assigned for different particle-size fractions (Nkedi-Kizza et af., 1983). Some of the differences among particle size fractions are associated with differences in organic carbon content (Nkedi-Kizza et af., 1983). At the aggregate scale, Ghadiri and Rose (1991) found higher pesticide concentrations in the outer compared to the inner regions of soil aggregates. However, Novak et af. (1994) found little impact on the sorption of atrazine among intact soil aggregates of different size ranges. 2. Microbial Distribution

One of the greatest unknowns in our understanding of biodegradation processes in soil is the spatial distribution of microbial populations. Our knowledge is greatly limited by our poor understanding of the physical architecture of soil organic matter and its mineral associations. Another major roadblock is the difficulty in enumerating populations able to degrade specific pollutants. It is hoped that with the rapid evolution of molecular microbial ecology, it will soon be possible to quantify organisms based on DNA and eventually RNA associated with specific metabolic pathways. Currently, most information on population distribution is not specifically for biodegrading populations. Only a small fraction of the surface area of soil is occupied by microorganisms (Hissett and Gray, 1976; Foster, 1988). At the particle scale, microbial biomass is concentrated in particle size fractions of 2-50 pm, with the majority in sizes of 2-20 pm (Monrozier et al., 1991). Particle sizes define pore sizes in soil, and pore size also controls the spatial distribution of microbial populations. Kilbertus (1980) found soil bacteria residing primarily in pores of 2 or 3 pm. Observed microbial distribution patterns suggest that a significant portion of the solution retained in pores is inaccessible to many microorganisms in soils at field

40

K. M. SCOW AND C. R. JOHNSON

moisture levels (Alexander and Scow, 1989). There are differences in microbial distribution as a function of aggregate size too. Kanazawa and Filip (1986) found that the majority of organisms as indicated by plate counts, ATP, and enzyme activity in organic particles are greater than 5 mm in diameter and in the silt-clay fraction less than 0.05 mm in diameter. Under aerobic conditions, carbon dioxide production and microbial biomass C was found to be greatest in microaggregates (<0.25 mm) and biomass declined with increasing aggregate diameters up to 20 mm (Seech and Beauchamp, 1988). Nishiyama et al. (1992) found that association with soil aggregates of an inoculated strain of Sphingomonus paucimobilis able to degrade y- 1,2,3,4,5,6-hexachlorocyclohexane(HCH) increased its survival in soil. Unfortunately, inoculation studies may not always provide relevant data regarding the distribution of indigenous populations responsible for biodegradation. There were distinct differences in the spatial distribution of indigenous and inoculated organisms, with most indigenous organisms associated with the smallest fractions (C0.025mm) of soil (Nishiyama er al.,

1992). With respect to distribution within a soil aggregate, the majority of the microbial biomass appears to concentrate at the surface or in the outer layer of soil aggregates (Priesack and Kisser-Priesack, 1993). Using microscopy, Kilbertus (1980)found that only 6% of the inner volume of soil aggregates was colonized by microorganisms. The major factors controlling microbial distribution within aggregates may be similar to those controlling distribution in engineered systems. Microorganisms immobilized in alginate or polyacrylamide-hydrazide beads grew only within the outer layer of the beads due to limited diffusion of oxygen or carbon sources into the interior of the bead. Microorganisms in the outer layer or at the surface of the beads consumed these resources before they could diffuse into the bead interior (Bettmann and Rehm, 1984; White and Thomas, 1990).

E. EFFECTOF SOILMOISTURE Soil moisture has a strong impact on both sorption and biodegradation. Sorption of many organic chemicals increases with large decreases in soil moisture. For volatile chemicals, such as trichloroethylene and toluene, sorption partition coefficients may be as much as two to four orders of magnitude higher for very dry than for the moist soils in which sorption measurements are usually made (Petersen ef al., 1995). This phenomenon could have a profound impact on bioavailability; however, an important question is how much biodegradation activity occurs at such low moisture contents. There was no biodegradation of toluene and trichloroethylene in one of the soils (Yolo) measured in the Petersen et al. (1995)study at or below a moisture content of 5% (Fan and Scow, 1993).

EFFECT OF SORPTION ON BIODEGRADATION

41

This moisture content was still higher than the levels at which the very high sorption partition coefficients were observed (Petersen et al., 1995). In a study of the effect of soil moisture on the biodegradation and sorption of carbofuran, Shelton and Parkin (1991) measured sorbed and soluble pools of carbofuran and mineralization at initial soil moisture potentials of -0.4, -0.8, - 1.5, -3.4, and -7.0 bar. Concentrations of soluble carbofuran initially decreased concomitant with increases in sorbed carbofuran, and then levels of both soluble and sorbed carbofuran decreased due to biodegradation. Mineralization rates were the same for the two higher water potentials and declined with decreases in the initial water contents. The sorption partition coefficients for carbofuran increased over the time of incubation, with greater increases seen in wet compared to dry soils. Carbofuran mineralization rates in the soil solution appeared to exceed desorption rates in the wetter soils but were slower than desorption rates in the dryer soils. Low soil moisture was thought to decrease rates both by directly inhibiting microbial activity and by reducing chemical bioavailability.

F. TEMPORAL CHANGES INBIOAVAILABILITY Temporal changes in a chemical’s bioavailability, such as irreversible aging and bound residue formation, are not accounted for in existing kinetic approaches that couple biodegradation and sorption. Although some mechanisms for bound residue formation are known, it is still not easy to be predict the quantitative impact of bound residue formation on biodegradation rates. Although a large fraction of a bound residue is not bioavailable, there is some evidence that some portion can be biodegraded (Calderbank, 1989). Although much of the biodegradation evidence is based on evolution of radiolabeled carbon dioxide from radiolabeled pesticides that have formed bound residues, there is evidence that parent compounds and their metabolites can be released from soil or taken up by biota (Kahn and Ivarson, 1982; Helling and Knvonak, 1978). In addition, there may be differences depending on the type of organic matter to which the residue is bound. Scheunert et al. (1991) found differences in the uptake by plants and microorganisms of bound residues of chlorinated anilines and phenols, depending on whether the residues were associated with fulvic or humic acid (Scheunert et al., 199 I ) . Because aging is such a new area of investigation, most of the research has focused on demonstrating that the phenomenon exists and does not illuminate the mechanisms involved. Phenanthrene and p-nitrophenol were found to have decreased rates of mineralization by inoculated organisms the longer the chemicals were in contact with sterilized soil (Hatzinger and Alexander, 1995). The longer that styrene was incubated in two soils before inoculation with bacteria, the slower the rate and lower percentage of the chemical was mineralized (Fu et al.,

K. M. SCOW AND C. R. JOHNSON

42

6 0

5 0 4 0

30 20 10

0 0

100

200

300

400

500

600

700

Hour Figure 10 Effect of aging and bacterial cell density on phenanthrene biodegradation in sterile Yolo soil. Numbers refer to the number of cells initially added per gram of soil. Fresh refers to soil that phenanthrene was added at the time of inoculation and aged refers to soil incubated with phenanthrene for 10 weeks before inoculation.

1994). In both organic and mineral soils, incubation for 30 days before inoculation decreased the amount of styrene mineralized to less than 25%of that mineralized for styrene recently added to soil. The interaction of aging and initial population density was investigated in sterile Yolo soil that was aged for 10 weeks with phenanthrene and then incubated under unsaturated conditions with different initial population densities In aged soil, inoculation with 5 X lo5 cells per gram of sterile soil did not result in any mineralization of phenanthrene. Inoculation with I X 108 cells per gram resulted in more mineralization than the lower inoculum densities; however, the initial rate and final percentage of chemical mineralized were significantly lower than levels measured in inoculated sterile soil that had been freshly amended with phenanthrene (Fig. 10).

VI. BIOREMEDIATION OF SORBED CHEMICALS There is widespread recognition of the challenges associated with treating soils contaminated by strongly sorbed chemicals using bioremediation techniques (Blackburn and Hafker, 1993; Stroo et al., 1992). The biodegradation of sorbed chemicals is influenced by the intrinsic biodegradation rate, soil aggregate diam-

EFFECT OF SORPTION ON BIODEGRADATION

43

eter, and sorption partition coefficient (Chung et al., 1993). In bioremediation, each of these parameters can be manipulated to various degrees, e.g., by inoculation with microbial strains, physical disruption of soil, or addition of surfactants, respectively. Studies relevant to the bioremediation of contaminated soil, particularly studies in which the previously listed parameters were manipulated, are discussed.

A. INOCULATION Most studies evaluating the efficacy of inoculating contaminated soil with microorganisms have, as their objective, enhancement of biodegradation rather than testing whether increasing the intrinsic biodegradation rate can overcome mass-transfer limitations. Weissenfels et a f . (1992) found that inoculation of PAH-contaminated soil from a coking plant had no impact on PAH biodegradation. Biodegradation of freshly added phenanthrene to a sandy loam soil with 3.3% organic matter was greatly stimulated by inoculation with an Alcaligenes sp. and the rate of phenanthrene degradation was similar for inoculation densities ranging from 5 x 105 to 5 X 108 cells g-' soil (Moller and Ingvorsen, 1993).

B. DECREASING AGGREGATE SIZEBY MIXING AND/OR CRUSHING Soil aggregate size can affect how rapidly a chemical reaches an equilibrium (if it does so) with respect to its distribution between regions to which it is sorbed and the soil solution (Brusseau and Rao, 1989b). This can have obvious impacts on the biodegradation rate of chemicals limited by mass transfer, particularly if microbial populations are primarily associated with the outer layers and surfaces of the aggregates. Dhawan et al. (1991) suggested, based on simulations using a radial diffusion model and parameters from the literature, that bioremediation of chemicals with sorption partition coefficients even as low as 1.5 cm3 per gram, which are sorbed to aggregates with radii of larger than 1 cm, would be greatly enhanced by reduction of aggregate diameter through crushing. Rijnaarts et a f . (1991) found that treatment conditions in a bioreactor containing a slurry of soil aggregates contaminated with a-hexachlorocyclohexane, when switched from an end-over-end mix to a well-stirred suspension, resulted in an increased biodegradation rate. The consequence of stirring was a decrease in the number-mean diameter of aggregates, assuming a log-normal distribution, from 41 to 26 p,m. The decrease in aggregate sizes reduced the average diffusion path length and thus enhanced desorption kinetics. Increased desorption rates, in

44

K. M. SCOW AND C. R.JOHNSON

turn, caused a reduction in the biodegradation lag period of 3 to 4 to less than 1 day and a 1.5-fold increase in the first-orderbiodegradation rate constant (assuming first-order kinetics). Ju and Devinny (1993) found substantial increases in soil respiration rates when oil-contaminated soil was pulverized in laboratory microcosms. They applied this technique in a landfarming application in which the simple ploughing commonly used is often insufficient to break up large soil clods enough to increase oil bioavailability. The field application of the pulverization techniques supported their laboratory findings. Pulverized soil (so that aggregates <0.5 cm diameter remained) had a 25% reduction in its treatment time compared to a soil that had been only ploughed. Increases in both the rate and the total amount of carbon mineralization from soil contaminated with oily waste were achieved after dispersion of saturated suspensions of soil by ultrasonication (Rasiah et al., 1992). Although sonication appeared to partially reduce microbial activity, its negative impact was small relative to its stimulatory effect in increasing the availability of the waste. The authors hypothesized that a portion of the oil was physically stabilized, as is native organic matter, through its association with soil particles, and that sonication released and made bioavailable a portion of this material.

C. ADDITIONOF SURFACTANTS Although, in theory, the addition of a surfactant should increase the fraction of a pollutant present in the solution phase and thus enhance biodegradation rates, the behavior of surfactants in soil systems is far more complex than this. Use of surfactants to enhance the biodegradation of sorbed pollutants is far from predictable and has been of mixed success depending on the type and amount of surfactant used, the soil type, the microorganisms involved, and the type and concentration of the pollutant being treated. Surfactant addition can result in stimulation of, repression of, and no effect on the biodegradation rate, as summarized in Liu et al. (1995). Stimulation can result from a decrease in the apparent sorption partition coefficient or use of the surfactant as a cosubstrate with an enhancement of the biodegradation of the pollutant. Repression can result from preferential use of the surfactant at the expense of the pollutant, toxicity of the surfactant, or pollutant association with a sorbed surfactant phase rendering it unavailable to microorganisms. Mihelcic et al. (1993) conclude that a major problem in the interpretation of the published literature stems from an insufficient description of the biological components of the systems. The microorganisms involved are rarely characterized with respect to their population densities, rates of biodegradation of the pollutant or the surfactant, or how they are impacted by the surfactant (e.g., by toxicity).

EFFECT OF SORPTION ON BIODEGRADATION

45

D. EVALUATION OF RATE-LIMITING STEPSIN BIOREMEDIATION Several researchers have developed approaches, similar to that of Chung et al. (1993) described under Section III,A,3, to screen for conditions when masstransfer limitations may substantially interfere with the bioremediation of contaminated material. Goshal and Luthy (1996) developed a conceptual framework, using dimensionless rate parameters, for identifying the rate-controlling processes for bioremediation in a bioreactor system. This framework was specifically developed for assessing bioremediation of naphthalene associated with free-flowing coal tar; however, the approach has a far more general application. The decision flowchart they developed includes the Biot number (ratio of flux at particle surface to intraparticle diffusion), the Thiele number (ratio of biodegradation to intraparticle diffusion), and the Damkohler number (ratio of biodegradation to flux at particle surface) (Fig. 11). This framework permits conclusions to be made on whether bioremediation is likely to be limited by the biodegradation rate, diffusion within particles, or transport through the bulk phase. Estimates of the rate coefficients needed for evaluation of naphthalene biodegradation were made from literature values or measurements in slurry systems containing coal tar as large globules or coated onto small, microporous beads that presented a high surface area. Goshal and Luthy (1996) concluded that biodegradation was the rate limiting step in the microporous beads, whereas mass transfer of material out of the globules was rate limiting in the bead-free system. Fry and Istok (1994) developed, as a screening tool, an analytical solution to the solute transport equation that includes rate-limited desorption and first-order biodegradation. For a wide range of published sorption partition coefficients, desorption rates, and biodegradation rates, they performed sensitivity analyses to define when desorption of biodegradation was the rate-limiting step for in situ

rnax biokinetic rate rnax diffusive rate

max biokinetic rate max flux at part. surf.

I

I

>1 pore sorption I diff. control

biokinetic control

>1

bulk phase transpt. control

Figure 11 Flowchart to identify processes controlling biodegradation kinetics in soil. Reprinted with permission from Environ. Tox. Chem., 1996. “Bioavailability of Hydrophobic Organic Compounds from NAPLs: The Biodegradation of Naphthalene from Coal Tar,”by S . Ghoshal and R. G. Luthy, Vol. 15(11). Copyright 1996 SETAC.

46

K. M. SCOW AND C. R. JOHNSON

bioremediation . They concluded that when desorption is strongly rate limiting, an increase in biodegradation rate has no impact on pollutant removal rates.

E. REGULATORY IMPLICATIONS OF REDUCEDBIOAVAILA~IL~~Y Current approaches evaluate the potential hazard of a pollutant in soil based on its extractabile concentration rather than on considerations of its bioavailability. However, as discussed previously, it is evident that over time many soil pollutants show marked decreases in their availability for uptake by receptor populations, as well as by biodegrading populations (Alexander, 1995). Whereas bioavailability has been recognized as a variable in aquatic toxicity studies (Spacie, 1994), this concept is relatively new in soil and groundwater research. To evaluate the biological impacts of contaminated soil, toxicity assays have been performed on soils directly rather than on the pollutant extracted from the soil. Some pollutants, although present at concentrations toxic in the absence of soil, show no toxic effects on plants, macrofauna, or bacterial assays when the pollutant is associated with soil (Weissenfels et al., 1992; Hatzinger and Alexander, 1995). Given economic constraints, evidence of reduced bioavailability undoubtably will become a factor in regulatory decisions involving contaminated soils. These impacts will be discussed only briefly here and the reader is refered to the reviews listed below for additional information. It is likely that reduced bioavailability will be used as one criterion in the prioritization of contaminated sites for remediation and setting site-specific cleanup goals (Beck et al., 1995). In making these decisions, it is important to distinguish between two qualitatively different mechanisms governing reduced bioavailability: (i) incorporation of pollutants into the soil organic matrix by strong chemical bonds, and (ii) sorption or aging of pollutants that may undergo release, abeit slowly, and for which the mechanisms involved are not well understood. The regulatory implications of bound residue formation, the first mechanism, are discussed by Kovacs (1986) and Calderbank (1989). For aged chemicals not subject to bound residue formation, the implications of reduced bioavailability are more complex. For example, the fact that sorption and aging are kinetic, rather than equilibrium, phenomena complicates the setting of specific soil quality criteria. Beck et al. (1995) discuss the implications of the biphasic, or two-compartment, biodegradation kinetics typically observed for sorbed chemicals in soil. They propose a “kinetically constrained soil quality limit” set so that the mass of chemical left in the soil is released at such a slow rate as to be practically insignificant. Many of the unresolved issues about the bioavailability of sorbed chemicals relate to our poor understanding of soil organic matter, sorption mechanisms, and microbial populations. Obviously, any changes in how sorbed chemicals are regulated should be based on a solid understanding of the processes involved and, in some cases, additional research is much needed.

EFFECT OF SORPTION ON BIODEGRADATION

47

VII. FUTURE RESEARCH NEEDS AND DIRECTIONS Further investigations of the biodegradation of sorbed chemicals will be beneficial not only for their practical applications to bioremediation but also for their potential to contribute fundamental knowledge about processes involving soil microorganisms, organic matter, and carbon flow. The following section discusses what we have identified as some of the most pressing research needs. Need for quantitative conceptual models. Even with the strong scientific and regulatory interest in the bioavailability of sorbed chemicals, few experimental studies integrate the biological and physicaVchemica1 components of this phenomenon in such a way that permits testing of quantitative hypotheses. Many common beliefs about sorbed chemicals (e.g., only the sorbed phase is metabolized) are based on simple correspondences between experimental data and model simulations for a limited set of data. These beliefs, which are actually only hypotheses, need to be tested for a broader set of conditions. As is true of any newly emerging field, it is difficult to draw general conclusions from many studies because the experiments are descriptive and not very quantitative. It is quite difficult to understand a phenomenon as complex as the biodegradation of sorbed chemicals without utilizing mathematical models to integrate the contributions of the underlying physical and biological processes. New data should be evaluated using existing or improved general conceptual frameworks that couple reactions of both mass transfer and biodegradation. The time-dependency of bioavailubilify. In many cases, the longer the amount of time that a pollutant remains in contact with soil, the smaller the fraction that is available for biodegradation or, for that matter, chemical extraction. Conceivably, a continuum in aging phenomena exists, with irreversibly bound chemicals at one end and substances that can eventually be completely recovered from soil at the other end. Better understanding of both kinetics and underlying mechanisms for aging phenomena throughout this continuum is required. More longterm studies are needed. Our difficulty in identifying aging mechanisms is hindered by fundamental gaps in our knowledge of the in situ structure and composition of soil organic matter. There is also a pressing need for more soil organic chemists and physical chemists to become involved in these basic questions and to collaborate in research with soil microbiologists. Differences among microorganisms in ability to use sorbed chemicals. Most models coupling biodegradation and mass transfer do not account for differences in the ability of microbial species to utilize sorbed chemicals. Experiments without soil, but in which the availability to microorganisms of the chemical substrate is limited (e.g., solid and NAPL forms of organics), indicate that differences among species can indeed matter with respect to rates of metabolism. Biological traits important in biodegrading sorbed chemicals may include the ability to attach to the surface of a chemical-sorbent complex, to penetrate into the solid phase of organic matter via hyphal growth or motility, or to survive on

K. M. SCOW AND C. R.JOHNSON

low solution concentrations of a substrate through possession of a low K,,,or low maintenance energy needs. Physiological studies of soil organisms successful in degrading sorbed chemicals are likely to reveal important information on this subject. Spatial heterogeneity of soil. The immense spatial heterogeneity of soil seriously complicates the study of sorbed chemical biodegradation. Heterogeneity is found in the pore- and aggregate-scale distribution of the pollutant as well as in the microbial populations able to degrade the pollutant. It is important to know whether a pollutant’s distribution is patchy, e.g., confined to areas of high hydrophobicity, or is relatively more uniform. Are microbial populations initially randomly distributed or enriched in areas where pollutants are likely to concentrate? Do distributions of pollutants and populations change with time? To what degree can microorganisms move in soil under realistic moisture conditions? Questions such as these need to be explored in relatively undisturbed soil systems. Significant progress may occur through further development of microscope technologies, such as atomic force and confocal microscopy, which have been successfully used to study simple systems. On a larger scale, X-ray tomography has been successful in the nondestructive examination of structure in simplified soil systems. All these techniques are currently limited in their ability to be utilized in real soils and require more development. Nonetheless, the most challenging problems in soil science will continue to elude us until better techniques are developed to permit microscale visualization of soil without its destruction.

ACKNOWLEDGMENTS This work was supported by the National Institute of Environmental Health Science’s Superfund Basic Research Program P42 ESO 4699, the U.C. Toxic Substances Program in Ecotoxicology, Hatch Experiment Station Project 5108-H, the Kearney Foundation of Soil Science, U.S. EPA (R819658) Center for Ecological Health Research at U.C. Davis, and the USDA Western Region Sustainable Agriculture Research and Education (SARE) Program. Although the information in this document has been funded in part by the United States Environmental Protection Agency, it may not necessarily reflect the views of the Agency and no official endorsement should be inferred. Carol Johnson was supported by a U.C. Davis Crosby Fellowship in Environmental Chemistry and a Superfund traineeship. We express our appreciation to Sandra Uesugi and Venece Tom for help in preparing the manuscript. We also thank Tom Young for his comments on the manuscript.

REFERENCES Adu. J. K., and Oades, J. M. (1978). Physical factors influencing decomposition of organic materials in soil aggregates. Soil Biol. Biochem. 10, 109-115. Aharoni, C., Sparks, D. L., Levinson, S., and Ravina, I. (1991). Kinetics of soil chemical reactions: Relationships between empirical equations and diffusion models. Soil Sci. SOC. Am. J . 55, 1307- 1312.

EFFECT OF SORPTION ON BIODEGRADATION

49

Alexander. M. (1995). How toxic are toxic chemicals in soil? Environ. Sci. Technol. 29,2713-2717. Alexander, M., and Scow, K. M. (1989). Kinetics of biodegradation in soil. I n “Reactions and Movement of Organic Chemicals in Soil” (B. L. Sawhney and K. Brown, eds.), pp. 243-269. SSSA Spec. Publ. 22, SSSA, ASA, Madison, WI. Anderson, D. W., Saggar, S . , Bettany, J. R., and Stewart, J. W. B. (1981). Particle size fractions and their use in studies of soil organic matter: I. The nature and distribution of forms of carbon, nitrogen, and sulfur. Soil Sci. SOC. Am. J . 45, 767-772. Anderson, J. P. E. (198 I). Soil moisture and the rates of biodegradation of diallate and triallate. Soil Biol. Biochem. 13, 155-161. Angley, J. T., Brusseau, M. L., Miller, W. L., and Delfino, J. J. (1992). Nonequilibrium sorption and aerobic biodegradation of dissolved alkylbenzenes during transport in aquifer material: Column experiments and evaluation of a coupled-process model. Environ. Sci. Technol. 26, 1404- 1410. Ball, W. P., and Roberts, P. V. (1991). Long-term sorption of halogenated organic chemicals by aquifer material. 2. Intraparticle diffusion. Environ. Sci. Technol. 25, 1237- 1249. Baveye, P., and Valocchi, A. (1989). An evaluation of mathematical models of the transport of biologically reacting solutes in saturated soils and aquifers. Water Resour. Res. 25, 1413- 1421. Beck, A. J.. Wilson, S. C., Alcock, R. E., and Jones, K. C. (1995). Kinetic constraints on the loss of organic chemicals from contaminated soils: Implications for soil-quality limits. Crir. Rev. Environ. Sci. Technol. 25, 1-43. Bettmann, H.,and Rehm, H. J. (1984). Degradation of phenol by polymer entrapped microorganisms. Appl. Microbiol. Biotechnol. 20, 285-290. Blackburn. J. W., and Hatlcer, W. R. (1993). The impact of biochemistry, bioavailability and bioactivity on the selection of bioremediation techniques. Trends Biotechnol. 11, 328-333. Bouchez, M., Blanchet, D., Vandecasteele, J.-P. (1995). Substrate availability in phenanthrene biodegradation: transfer mechanism and influence on metabolism. Appl. Microbiol. Biotechnol. 43, 952-960. Brusseau, M. L. (1995). The effect of nonlinear sorption on transformation of contaminants during transport in porous media. J. Contam. Hydrol. 17, 277-291. Brusseau, M. L., and Rao, P. S. C. (1989a). The influence of sorbate-organic matter interactions on sorption nonequilibrium. Chemosphere 18, 1691-1706. Brusseau, M. L., and Rao, P. S. C. (1989b). Sorption nonideality during organic contaminant transport in porous media. Crit. Rev. Environ. Control 19, 33-98. Brusseau, M. L., and Rao. P. S. C. (1991). Influence of sorbate structure on nonequilibrium sorption of organic compounds. Environ. Sci. Technol. 25, 1501-1506. Brusseau, M. L., Jessup, R. E., and Rao, P. S. C. (1992a). Modeling solute transport influenced by multiprocess nonequilibrium and transformation reactions. Water Resour. Res. 28, 175- 182. Brusseau, M. L, Rao, P. S. C., and Bellin, C. A. (1992b). Modeling coupled processes in porous media: Sorption, transformation, and transport of organic solutes. In “Interacting Processes in Soil Science” (R. J. Wagenet, P. Baveye, and B. A. Stewart, eds.), pp. 147-184. Lewis, Boca Raton, FL. Burford, M. D., Hawthorne, S. B., and Miller, D. J. (1993). Extraction rates of spiked versus native PAHs from heterogeneous environmental samples using supercritical fluid extraction and sonication in methylene chloride. Anal. Chem. 65, 1497-1505 Button, D. K., Robertson, B. R., McIntosh, D.. and Jiittner, F. (1992). Interactions between marine bacteria and dissolved-phase and beached hydrocarbons after the Exxon Valdez oil spill. Appl. Environ. Microhiol. 58, 243-25 I . Calderbank, A. (1989). The occurrence and significance of bound pesticide residues in soil. Rev. Environ. Contam. Toxicol. 108, 71-103. Celia, M. A,. Kindred, J. S.. and Herrera, I. (1989). Contaminant transport and biodegradation I . A numerical model for reactive transport in porous media. Water Resour. Res. 25, 11411148.

K. M. SCOW AND C.R. JOHNSON Cheng, H. H. (1990).Organic residues in soils: Mechanisms of retention and extractability. Int. J. Eviron. Anal. Chem. 39, 165-171. Chiou, C. T. (1990). Roles of organic matter, minerals, and moisture in sorption of nonionic compounds and pesticides by soil. In “Humic Substances in Soil and Crop Sciences; Selected Readings” (P. MacCarthy, C. E. Clapp, R. L. Malcolm, and P. R. Bloom, eds.), pp. 111-160. SSSA, ASA, Madison, WI. Chiou, C. T., Peters, L. J., and Freed, V. H. (1979).A physical concept of soil-water equilibria for nonionic organic compounds. Science 206, 831-832. Chiou, C. T.,Porter, P. E., and Schmedding, D. W. (1983).Partition equilibria of nonionic organic compounds between soil organic matter and water. Environ. Sci. Technol. 17, 227-231. Chung, G-Y., McCoy, B. J., and Scow, K. M. (1993).Criteria to assess when biodegradation is kinetically limited by intraparticle diffusion and sorption. Eiotechnol. Eioeng. 41, 625-632. Coates, J. T.,and Elzerman, A. W. (1986).Desorption kinetics for selected PCB congeners from river sediments. J. Contam. Hydrol. 1, 191-210. Connaughton, D. F., Stedinger, J. R., Lion, L. W., and Shuler, M. L. (1993).Description of timevarying desorption kinetics: Release of naphthalene from contaminated soils. Environ. Sci. Technol. 21, 2397-2403. Curtis, G. P., Reinhard, M., and Roberts, P. V. (1986).Sorption of hydrophobic organic compounds by sediments. In “Geochemical Processes at Mineral Surfaces” (J. A. Davis and K. F. Hayes, eds.), pp. 192-216. American Chemical Society, Washington, DC. Dec, J.. and Bollag, J.-M. (1988).Microbial release and degradation of catechol and chlorophenols bound to synthetic humic acid. Soil Sci. SOC. Am. J . 52, 1366-1371. Dec, J., Shuttleworth, K. L., and Bollag, J.-M. (1990). Microbial release of 2,4-dichlorophenol bound to humic acid or incorporated during humification. J. Environ. Qual. 19, 546-551. Dhawan, S . , Fan, L. T., Erickson, L. E., and ’hitemwong, P. (1991). Modeling, analysis, and simulation of bioremediation of soil aggregates. Environ. Prog. 10, 25 1-260. Dhawan, S., Erickson, L. E., and Fan, L. T. (1993).Model development and simulation of bioremediation in soil beds with aggregates. Ground Water 31, 271-284. Efroymson, R. A., and Alexander, M. (1991).Biodegradation by an Arthrobacter species of hydrocarbons partitioned into an organic solvent. Appl. Environ. Microbiol. 57, 1441-1447. Efroymson, R. A., and Alexander, M. (1994).Role of partitioning in biodegradation of phenanthrene dissolved in nonaqueous-phase liquids. Environ. Sci. Technol. 28, 117-1 19. Efroymson, R. A., and Alexander, M. (1995). Reduced mineralization of low concentration of phenanthrene because of sequestering in nonaqueous-phase liquids. Environ. Sci. Technol. 29, 515-521. Elliott, E. T. (1986). Aggregate structure and carbon, nitrogen, and phosphorus in native and cultivated soils. Soil Sci. SOC. Am. J. 50, 627-633. Estrella, M. R., Brusseau, M. L., Maier, R. S., Pepper, I. L., Wierenga, P. J., and Miller, R. M. (1993).Biodegradation, sorption, and transport of 2,4-dichlorophenoxyacetic acid in saturated and unsaturated soils. Appl. Environ. Microbiol. 59, 4266-4273. Fan, S . , and Scow, K. M. (1993).Biodegradation of tricholorethylene and toluene by indigenous microbial populations in soil. Appl. Environ. Microbiol. 59, 191 1-1918. Focht, D.,and Shelton, D. (1987).Growth kinetics of Reudomonas alcaligenes C.O. Relative to inoculation and 3-chlorobenzoate metabolism in soil. Appl. Environ. Microbiol. 53, 18461849. Foster, R. C. (1988).Microenvironments of soil microorganisms. Eiol. Fertil. Soils 6, 189-203. Fry, V. A., and Istok, J. D. (1994).Effects of rate-limited desorption on the feasibility of in situ bioremediation. Water Resour. Res. 30, 2413-2422. Fu, M. H., Mayton, H., and Alexander, M. (1994).Desorption and biodegradation of sorbed styrene in soil and aquifer solids. Environ. Toxicol. Chem. 13, 749-753.

EFFECT OF SORPTION ON BIODEGRADATION

51

Gamerdinger, A. P., Wagenet, R. J,. and van Genuchten, M. Th. (1990). Application of two-site/two region models for studying simultaneous nonequilibrium transport and degradation of pesticides. Soil Sci. Soc. Am. J . 54, 957-963. Gamerdinger, A. P., Lemley, A. T., and Wagenet, R. J. (1991). Nonequilibrium sorption and degradation of three 2-choloro-s-triazine herbicides in soil-water systems. J. Environ. Qual. 20, 8 15-822. Gamerdinger, A. P., Dowling, K. C., and Lemley, A. T. (1993). Miscible displacement and theoretical techniques for simultaneous study of pesticide sorption and degradation during transport. In “Sorption and Degradation of Pesticides and Organic Chemicals in Soil” (D. M. Linn, T. H. Carski, M. L. Brusseau, and F-H. Chang, eds.), pp. 115-123. SSSA Spec. Publ. 32. SSSA, ASA, Madison, WI. Garbarini, D. R., and Lion, L. W. (1986). Influence of the nature of soil organics on the sorption of toluene and trichloroethylene. Environ. Sci. Technol. 20, 1263- 1269. Gauthier, T. D., Seltz, W. R., and Grant, C. L. (1987). Effects of structural and compositional variations of dissolved humic materials on pyrene Koc values. Environ. Sci. Technol. 21, 243248. Ghadiri, H., and Rose, C. W. (1991). Sorbed chemical transport in overland flow: I . A nutrient and pesticide enrichment mechanism. J. Environ. Qual. 20, 628-633. Ghoshal, S . , and Luthy, R. G. (1996). Bioavailability of hydrophobic organic compounds from NAPLS: The biodegradation of naphthalene from coal tar. Submitted for publication. Godfrey, K. (1983). “Compartmental Models and Their Application.” Academic Press, New York. Goltz, M. N., and Roberts, P. V. (1986). Interpreting organic solute transport data from a field experiment using physical nonequilibrium models. J . Confum. Hydrol. 1, 77-93. Goswami, P., and Singh. H. D. (1991). Different modes of hydrocarbon uptake by two different Pseudomoms sp. Biorechnol. Bioeng. 37, I - I 1. Grathwohl, P. (1990). Influence of organic matter from soil and sediments from various origins on the sorption of some chlorinated aliphatic hydrocarbons: Implications on Koc correlations. Environ. Sci. Technol. 24, 1687-1693. Greer, L. A,, and Shelton, D. R. (1992). Effect of inoculant strain and organic matter content on kinetics of 2,4-dicholorophenoxyaceticacid degradation in soil. Appl. Environ. Microbiol. 58, 1459- 1465. Guerin, W. F., and Boyd, S. A. (1992). Differential bioavailability of soil-sorbed naphthalene to two bacterial species. Appl. Environ. Microbiol. 58, 1142-1 152. Guerin, W. F., and Boyd., S. A. (1993). Bioavailability of sorbed naphthalene to bacteria: Influence of contaminant aging and soil organic carbon content. In “Sorption and Degradation of Pesticides and Organic Chemicals in Soil” (D. M. Linn, T. H. Carski, M. L. Brusseau, and F-H. Chang, Eds.), pp. 197-208. SSSA Spec. Publ. 32, SSSA, ASA, Madison, WI. Guerin, W. F., and Boyd, S. A. (1995). Maintenance and induction of naphthalene degradation activity in Pseudomonas purida and an Alcaligenes sp. under different culture conditions. Appl. Environ. Microbiol. 61, 4061-4068. Gustafson, D. (1989). Ground water ubiquity score: A simple method for assessing pesticide leachability. Environ. Toxicol. Chem. 8, 339-357. Gustafson, D. I., and Holden, L. R. (1990). Nonlinear pesticide dissipation in soil: A new model based on spatial variability. Environ. Sci. Technol. 24, 1023- 1038. Hamaker, J. W., and Goring, C. A. I. (1976). Turnover of pesticide residues in soil. In “Bound and Conjugated Pesticide Residues”(D. D. Kaufman, G. G. Still, G. D. Paulson, and S. K. Bandal, eds.), pp. 219-243. ACS Symposium Series 29, American Chemical Society, Washington, DC. Hamaker, J. W., and Thompson, J. M. (1972). Adsorption. In “Organic Chemicals in the Soil Environment, Vol. I” (C. A. I. Goring and J. W. Hamaker, eds.), pp. 49-143. Dekker, New York .

K. M. SCOW AND C. R.JOHNSON

52

Harms, H., and Zehnder, A. J. B. (1994).Influence of substrate diffusion on degradation of dibenzofuran and 3-cholorodibenzofuran by attached and suspended bacteria. Appl. Environ. Microbiol. 60,2726-2745. Harms, H., and Zehnder, A. J. B. (1995).Bioavailability of sorbed 3-cholorodibenzofuran. Appl. Environ. Microbiul. 61,27-33. Hatzinger, P. B., and Alexander, M. (1995).Effect of aging of chemicals in soil on their biodegradability and extractability. Environ. Sci. Technol. 29, 537-545. Helling, C. S., and Krivonak, A. E. (1978). Biological characteristics of bound dinitroaniline herbicides in soils. J. Agric. Food Chem. 26, 1164-1 172. Helweg, A. (1975).Degradation of 14C-maleic hydrazide in soil as influenced by adsorption on activated carbon. Weed Res. 15, 129-133. Hissett, R., and Gray, T. R. G. (1976).Microsites and time changes in soil microbe ecology. In “The Role of Terrestrial and Aquatic Organisms in Decomposition Processes” (J. M. Anderson and A. MacFadyen, eds.), pp. 23-39. Blackwell, Oxford, UK. Hyzak, D. L., and Zimdahl, R. L. (1974).Rate of degradation of metribuzin and two analogs in soil. Weed Sci. 22, 75-79. Jenkinson, D. S., and Rayner, J. H. (1977).The turnover of soil organic matter in some of the Rothamsted classical experiments. Soil Sci. 123, 298-305. Ju, M., and Devinny, J. S. (1993).Effects of pulverization on soil bioremediation. Hazard. Waste Hazard. Muter. 10, 357-364. Jury, W.A., Focht, D. D., and Farmer, W. J. (1987).Evaluation of pesticide groundwater pollution potential from standard indices of soil-chemical adsorption and biodegradation. J. Environ. Q u d . 16, 422-428. Kanazawa, S . , and Filip, Z. (1986).Distribution of microorganisms, total biomass, and enzyme activities in different particles of brown soil. Microbiol. Ecol. 12, 205-215. Karickhoff, S. W. (1980).Sorption kinetics of hydrophobic pollutants in natural sediments. In “Contaminants and Sediments, Vol. 11” (R. A. Baker, ed.), pp. 193-204. Ann Arbor Science, Ann Arbor, MI. Karickhoff, S. W. (1984).Organic pollutant sorption in aquatic systems. J . Hydraulic Eng. 110,

707-735.

Karickhoff, S. W.,and Moms, K. R. (1985). Sorption dynamics of hydrophobic pollutants in sediment suspensions. Environ. Toxicol. Chem. 4, 469-479. Karickhoff, S.W.,Brown, D. S., and Scott, T. (1979).Sorption of hydrophobic pollutantson natural sediments. Water Res. 13, 241-248. Khan, S. U . (1982).Bound pesticide residues in soil and plants. Residue Rev. 84, 1-24. Khan, S. U.,and Ivarson, K. C. (1982). Release of bound (nonextractable) residues by various physiological groups of microorganisms. J . Environ. Sci. Health 17, 737-749. Kilbertus, G. (1980).Etude des microhabitats contenus dans les aggrkgats du sol. Leur relation avec la biomass bacteriknne et la taille des procaryotes presents. Rev. Ecol. Eiol. Sol. 17, 543-557. Koch, A. K., Kappeli, O., Fiechter, A., and Reiser, A. (1991). Hydrocarbon assimilation and biosurfactant production in Pseudomonas aeruginosa mutants. J. Eacteriol. 173, 4212-4219. Koskinen, W.C., and Harper, S. S. (1990).The retention process: Mechanisms. In “Pesticides in the Soil Environment: Processes, Impacts, and Modeling” (H. H. Cheng, ed.), pp. 51-77. SSSA, Madison, WI. Kovacs, M. F. (1986).Regulatory aspects of bound residues (chemistry). Residue Rev. 97, 1-17. Liu, Z., Jacobson, A. M., and Luthy, R. 0. (1995).Biodegradation of naphthalene in aqueous nonionic surfactant systems. Appl. Environ. Microbiol. 61, 145-151. Lyman, W. J. (1982).Adsorption coefficient for soils and sediments. In “Handbook of Chemical Property Estimation Methods” (W. J. Lyman, W. F. Reehl, and D. H. Rosenblatt, eds.), pp. 4-1-4-5.McGraw-Hill, New York.

EFFECT OF SORPTION ON BIODEGRADATION

53

McBride, J. F., Brockman, F. J., Szcsody, J. E., and Streile, G . P. (1992).Kinetics of quinoline biodegradation, sorption and desorption in a clay-coated model soil containing a quinolinedegrading bacterium. J. Contam. Hydrol. 9, 133-154. McGinley, P. M., Katz. L. E., and Weber, W. J. (1993).A distributed reactivity model for sorption by soils and sediments. 2. Multicomponent systems and competitive effects. Environ. Sci. Technol. 27, 1524- 1531. Mihelcic, J. R., and Luthy, R. G. (1991).Sorption and microbial degradation of naphthalene in soilwater suspensions under denitrification conditions. Environ. Sci. Technol. 25, 169-177. Mihelcic, J. R., Lueking, D. R., Mitzell, R. J.. and Stapelton, J. M. (1993).Bioavailability of sorbed- and separate-phase chemicals. Biodegradation 4, 14I - 153. Miller, C. T., and Weber. W. I., Jr. (1986).Sorption of hydrophobic organic pollutants in saturated soil systems J. Contam. Hydrol. 1, 243-261. Miller, M. E.,and Alexander, M. (1991).Kinetics of bacterial degradation of benzylamine in a montmorillonite suspension. Environ. Sci. Technol. 25, 240-245. Mingelgrin, U., and Gerstl, 2. (1983).Reevaluation of partitioning as a mechanism of nonionic chemicals adsorption in soils. J. Environ. Qual. 12, 1 - I 1. Mdler, J., and Ingvorsen, H. (1993).Biodegradation of phenanthrene in soil microcosms stimulated by an introduced Alcaligenes sp. FEMS Microbiol. Ecol. 102, 271-278. Monrozier, L. J., Ladd, J. N., Fitzpatrick, R. W., and Foster, R. C. (1991).Components and microbial biomass content of size fractions in soils of contrasting aggregation. Geoderma 50,

37-62. Moyer, J. R., Hance, R. J., and McKone, C. E. (1972).The effect of adsorbents on the rate of degradation of herbicides incubated with soil. Soil Biol. Biochem. 4, 307-31 1. Murphy, E. M., Zachara, J. M., and Smith, S. C. (1990). Influence of mineral-bound humic substances on the sorption of hydrophobic organic compounds. Environ. Sci. Technol. 24, 1507- I5 16. Nishiyama, M., Senoo, K., Wada, H., and Matsumoto, S. (1992).Identification of soil microhabitats for growth, death and survival of a bacterium, y- I ,2,3,4,5,6-hexachlorocyclohexaneassimilating Sphingomonas paucimobilis, by fractionation of soil. FEMS Microbiol. Ecol. 101, 145- 150. Nkedi-Kizza, P., Rao, P. S. C., and Johnson. J. W. (1983).Adsorption of diuron and 2.43-Ton soil particle-size separates. J. Environ. Qual. 12, 195- 197. Nkedi-Kizza, P.,Biggar, J. W., Selim, H. M., van Genuchten, M. Th., Wierenga, P. J., Davidson, J. M., and Nielsen, D. R. (1984).On the equivalence of two conceptual models for describing ion exchange during transport through an aggregated oxisol. Water Resour. Res. 20, 1 123- 1130. Novak, J. M., Moorman, T. B., and Karlen. D. L. (1994).Influence of soil aggregate size on atrazine sorption kinetics. J. Agric. Food Chem. 42, 1809-1812. Ogram, A. V.,Jessup, R. E., Ou, L. T.,and Rao, P. S. C. (1985).Effects of sorption on biological degradation rates of (2,4-dicholorophenoxy) acetic acid in soils. Appl. Environ. Microbiol. 49,

582-587. Parton. W. J., Stewart, S. W. B., and Cole, C. V. (1988).Dynamics of carbon, nitrogen, phosphorus and sulfur in cultivated soils. A model. Biogeochernistry 5, 109-131. Pavlostathis, S. G., and Mathavan, G. N. (1992).Desorption kinetics of selected volatile organic compounds from field contaminated soils. Environ. Sci. Technol. 26, 532-538. Petersen, L. W., Moldrup, P., El-Farhan, Y. H., Jacobsen, 0. H.,Yamaguchi, T.. and Rolston, D. E. (1995).The effect of moisture and soil texture on the adsorption of organic vapors. J. Environ. Q u l . 24, 752-759. Pignatello, 1. J. (1989).Sorption dynamics of organic compounds in soils and sediments. In “Reactions and Movement of Organic Chemicals in Soils” (B. L. Sawhney and K. Brown, eds.), pp. 45-80. SSSA Spec. Publ. 22, SSSA, ASA, Madison, WI.

54

K. M. SCOW AND C . R.JOHNSON

Pignatello, J. J. (1990a). Slowly reversible sorption of aliphatic halocarbons in soils. 1. Formation of residual fractions. Environ. Toxicol. Chem. 9, 1107-1 115. Pignatello, J. J. (1990b). Slowly reversible sorption of aliphatic halocarbons in soils. 11. Mechanistic aspects. Environ. Toxicol. Chem. 9, 11 17-1126. Pignatello, J. J. , and Xing, B. (1996). Mechanisms of slow sorption of organic chemicals to natural particles. Environ. Sci. Technol. 30, 1-11. Pignatello, J. J., Sawhney, B. L., and Frink, C. R. (1987). EDB: Persistence in soil. Science 236, 898. Preston, C. M., and Newman, R. H. (1992). Demonstration of spatial heterogeneity in the organic matter of de-ashed humin samples by solid-state 13C CPMAS NMR. Can. J . Soil Sci. 72, 1319. Priesack, E. (1991). Analytical solution of solute diffusion and biodegradation in spherical aggregates. Soil Sci. SOC. Am. J. 55, 1227-1230. Priesack, E., and Kisser-Priesack, G. M. (1993). Modeling diffusion and microbial uptake of 13C-glucose in soil aggregates. Geoderma 56, 561-573. Ramsay, A. J. (1984). Extraction of bacteria from soil: Efficiency of shaking or ultrasonication as indicated by direct counts and autoradiography. Soil Biol. Biochem. 16, 475-481. Rao, P. S. C., Jessup, R. E., Rolston, D. E., Davidson, J. M., and Kilcrease, D. P. (1980a). Experimental and mathematical description of nonadsorbed solute transfer by diffusion in spherical aggregates. Soil Sci. Soc. Am. J . 44, 684-688. Rao, P. S. C., Rolston, D. E., Jessup, R. E., and Davidson, J. M. (1980b). Solute transport in aggregated porous media: Theoretical and experimental evaluation. Soil Sci. SOC.Am. J . 44, 1139-1 146. Rao, P. S. C, Bellin, C. A,, and Brusseau, M. L. (1993). Coupling biodegradation of organic chemicals to sorption and transport in soils and aquifers: Paradigms and paradoxes. In “Sorption and Degradation of Pesticides and Organic Chemicals in Soil” (D. M. Linn, T. H. Carski, M. L. Brusseau, and F-H. Chang, eds.), pp. 1-26. SSSA Spec. Publ. 32, SSSA, ASA, Madison, WI. Rasiah, V., Voroney, R. P., and Kachanoski, R. G. (1992). Bioavailability of stabilized oily waste organics in ultrasonified soil aggregates. Water Air Soil Pollut. 63, 179-186. Reddy, P. G., Singh, H. D., Pathak, M. G . , Bhagat, S . D., and Baruah, J. N. (1983). Isolation and function characterization of hydrocarbon emulsifying and solubilizing factors produced by a Pseudomonus sp. Biotechnol. Bioeng. 25, 387-401. Rijnaarts, H.H. M., Bachmann, A,, Jumelet, J. C., and Zehnder, A. J. (1991). Effect of desorption and intraparticle mass transfer on the aerobic biomineralization of a-hexachlorocyclohexane in a contaminated calcareous soil. Environ. Sci. Technol. 24, 1349- 1354. Robinson, K. G., Fanner, W. S., and Novak, J. T. (1990). Availability of sorbed toluene in soils for biodegradation by acclimated bacteria. Water Res. 24, 345-350. Rosenberg, M., and Rosenberg, E. (1981). Role of adherence in growth of Acinerobucter culcoaceticus RAG-l on hexadecane. J. Bucteriol. 148, 51-57. Rutherford, D. W., Chiou, C. T., and Kile, D. E. (1992). Influence of soil organic matter composition on the partition of organic compounds. Environ. Sci. Technol. 26, 336-340. Scheunert, I., and Mansour, M. (1992). Binding of organic pollutants to soil organic matter. In?. J . Environ. Anal. Chem. 46, 189-199. Scheunert, I., Mansour, M.,and Adrian, P. (1991). Formation of conversion products and of bound residues of chlorinated anilines in soil. Envrion. Toxicol. Chem. 31/32, 107-1 12. Schmidt, S. K. (1992). Models for studying the population ecology of microorganisms in natural systems. In “Modeling the Metabolic and Physiologic Activities of Microorganisms” (C. J. Hurst, ed.), pp. 31-59. Wiley, New York. Schwarzenbach, R. P., Gschwend, P. M., and Imboden, D. M. (1993). “Environmental Organic Chemistry.” Wiley, New York. Scow, K. M. (1993). Effect of sorption-desorption and diffusion processes on the kinetics of biodeg-

EFFECT OF SORPTION ON BIODEGRADATION

55

radation of organic chemicals in soil. In “Sorption and Degradation of Pesticides and Organic Chemicals in Soil” (D. M. Linn, T. H. Carski, M. L. Brusseau, and F-H. Chang, eds.), pp. 126. SSSA Spec. Publ. 32, SSSA, ASA, Madison, WI. Scow, K. M., and Alexander, M. (1992). Effect of diffusion and sorption on the kinetics of biodegradation: Experimental results with synthetic aggregates. Soil Sci. Soc. Am. J. 56, 128134. SCOW,K. M., and Hutson, J. (1992). Effect of diffusion and sorption on the kinetics of biodegradation: Theoretical considerations. Soil Sci. Soc. Am. J . 56, 119-127. Scow, K. M., Simkins, S., and Alexander, M. (1986). Kinetics of mineralization of organic compounds at low concentrations in soil. Appl. Environ. Microbiol. 51, 1028-1035. Scow, K. M., Fan, S . , Johnson, C., and Ma, G. M. (1994). Biodegradation of sorbed chemicals in soil. Environ. Health Perspec. 103, 93-95. Scribner, S. L., Benzing, T. R., Sun, S., and Boyd, S. A. (1992). Organic chemicals in the environment: Desorption and bioavailability of aged simazine residues in soil from a continuous corn field. J. Environ. Qual. 21, 115-120. Seech, A. G . , and Beauchamp, E. G. (1988). Denitrification in soil aggregates of different sizes. Soil Sci. Soc. Am. J . 52, 1616-1621. Shelton, D. R., and Parkin, T. B. (1991). Effect of moisture on sorption and biodegradation of carbofuran in soil. J. Agric. Food Chem. 39, 2063-2068. Simkins, S., and Alexander, M. (1984). Models for mineralization kinetics with the variables of substrate concentration and population density. Appl. Environ. Microbiol. 47, 1299- 1306. Simmons, G. F. (1972). “Differential Equations.” McGraw-Hill, New York. Sims, G. K . , Radosevich, M., He, X. T., and Traina, S. J. (1991). The effects of sorption on the bioavailability of pesticides. In “Biodegradation: Natural and Synthetic Materials” (W. B. Betts, ed.) Springer-Verlag, New York. Smith, S. C., Ainsworth, C. C., Traina, S. J., and Hicks, R. J. (1992). Effect of sorption on the biodegradation of quinoline. Soil Sci. Soc. Am. J. 56, 737-746. Spacie, A. (1994). Interactions of organic pollutants with inorganic solid phases: Are they important to bioavailability? In “Bioavailability. Physical, Chemical, and Biological Interactions” (J. L. Hamelink, P. F. Landrum, H. L. Bergman., and W. H. Benson, eds.), pp. 73-82. SETAC Spec. Publ. Ser., Lewis, Boca Raton, FL. Speitel, G . E., Jr., Chih-Jen, L., ’lbrakhia, M., and Zhu, X.-J. (1989). Biodegradation of trace concentrations of substituted phenols in granular activated carbon columns. Environ. Sci. Technol. 23, 68-74. Spurlock, F. C., and Biggar, J. W. (1994a). Thermodynamics of organic chemical partition in soils. 1. Development of a general partition model and application to linear isotherms. Environ. Sci. Technol. 28, 989-995. Spurlock, F. C., and Biggar, J. W. (1994b). Thermodynamics of organic chemical partition in soils. 2. Nonlinear partition of substituted phenylureas from aqueous solution. Environ. Sci. Technol. 28, 996-1002. Spurlock, F. C., Haung, K., and Van Genuchten, M. T. (1995). Isotherm nonlinearity and nonequilibrium sorption effects on transport of fenuron and monuron in soil columns. Environ. Sci. Technol. 29, 1000-1007. Steen, W. C., Paris, D. F., and Baughman, G. L. (1980). Effects of sediment sorption on microbial degradation of toxic substances. In “Contaminants and Sediments” (R. A. Barker, ed.), pp, 477-482. Ann Arbor Science, Ann Arbor, MI. Steinberg, S. M., Pignatello, J. J., and Sawhney, B. L. (1987). Persistence of 1.2-dibromoethane in soils: Entrapment in intraparticle micropores. Environ. Sci. Technol. 21, 1201- 1208. Stevenson, F. J. (1976). Organic matter reactions involving pesticides in soil. In “Bound and Conjugated Pesticide Residues” (D. D. Kaufman, G. G . Still, G. D. Paulson, and S. K. Bandal, eds.), pp. 180-207. ACS Symposium Series 29, American Chemical Society, Washington, DC.

56

K. M. SCOW AND C. R. JOHNSON

Stevenson, F. J. (1994). “Humus Chemistry. Genesis, Composition, Reactions,” 2nd Ed. Wiley, New York. Stevenson, F. J., and Elliot, E. T. (1989). Methodologies for assessing quantity and quality of SOM. In “Dynamics of Soil Organic Matter in Tropical Ecosystems” (D. C. Coleman, J. M. Oades, and G. Uehara, eds.), pp. 173-199. Univ. of Hawaii Press, Honolulu. Stroo, H. F., Anderson, H. K., and Smith, J. R. (1992). Predicting biodegradation in hydrocarbon contaminated soils. in “Hydrocarbon Contaminated Soils, Vol. II” (P. T. Kostecki, E. J. Calabrese, and M. Bonazountas, eds.) pp. 227-241. Lewis, Boca Raton, E L Tempest, D. W., and Neijssel, 0. M. (1978). Eco-physiological aspects of microbial growth in aerobic nutrient-limited environments. Adv. Microb. Ecol. 2, 105-153. van Genuchten, M. T. (1974). Mass transfer studies in sorbing porous media. PhD dissertation, New Mexico State Univ., Las Cruces. van Cenuchten, M. Th., and Wagenet, R. J. (1989). Two-site/two-region models for pesticide transport and degradation: Theoretical development and analytical solutions. Soil Sci. SOC.Am. J. 53, 1303-1310. van Loosdrecht, M. C. M., Lyklema, J., Norde, W., and Zehnder, A. J. B. (1990). Influences of interfaces on microbial activity. Microbiol. Rev. 54, 75-87. van Veen, J. A,, and Paul, E. A. (1981). Organic carbon dynamics in grassland soils. 1. Background information and computer simulation. Can. J. Soil Sci. 61, 185-201. Volkering, F., Breure, A. M.,Sterkenburg, A., and van Andel, J. G. (1992). Microbial degradation of polycyclic aromatic hydrocarbons: Effect of substrate availability on bacterial growth kinetics. Appl. Microbiol. Biotechnol. 36, 548-552. Weber, W. J., McGinley, P. M., and Katz, L. E. (1992). A distributed reactivity model for sorption by soils and sediments. 1. Conceptual basis and equilibrium assessments. Environ. Sci. Technol. 26, 1955-1962. Weissenfels, W. D., Klewer, H.-J., and Langhoff, J. (1992). Adsorption of polycyclic aromatic hydrocarbons (PAHs) by soil particles: Influence on biodegradability and biotoxicity. Appl. Microbiot. Biotechnol. 36, 689-696. White, G. F., and Thomas, 0. (1990). Immobilization of the surfactant-degrading bacterium Pseudomonas C12B in polyacrylamide gel beads: I. Effect of immobilization on the primary and ultimate biodegradation of SDS, and redistribution of bacteria within beads after use. Enzyme Microb. Technol. 12, 697-705. Woodruff, C. M. (1949). Estimating the nitrogen delivery of soil from the organic matter determination as reflected by Sanborn field. Soil Sci. SOC. Am. Proc. 14, 208-212. Wu, S. C., and Gschwend, P. M. (1986). Sorption kinetics of hydrophobic compounds to natural sediments and soils. Environ. Sci. Technol. 20, 717-725. Wu, S . C., and Gschwend, P. M. (1988). Numerical modeling of sorption kinetics of organic compounds to soil and sediment particles. Water Resour. Res. 24, 1373-1383. You, 1. S., and Bartha, R. (1982). Stimulation of 3,4-dichloroanaline mineralization by aniline. Appl. Environ. Microbiol. 44, 678-68 I. Young, T.M., and Weber, W. J., Jr. (1995). A distributed reactivity model for sorption by soils and sediments. 3. Effects of diagenetic processes on sorption energetics. Environ. Sci. Technol. 29, 92-97. Zimdahl, R. L., and Gwynn, S. M. (1977). Soil degradation of three dinitroanalines. Weed Sci. 25, 247-25 I .

HERBICIDE RESISTANCE:IMPACT AND MANAGEMENT S. B. Powles,' C. Preston,' I. B. Bryan,2 and A. R. Jutsum* 'C.R.C. for Weed Management Systems, Waite Campus, University of Adelaide, Glen Osmond 5064 Australia *Zeneca Agrochemicals, Jealott's Hill Research Station, Bracknell, Berkshire, RCA2 6ET, United Kingdom

I. Introduction: Major Crops and Herbicide Markets A. Global Herbicide Use and Crop Production B. Trends in Herbicide Use 11. The Threat from Herbicide-Resistant Weeds A. Selection Pressure and the Evolution of Herbicide Resistance B. Resistance to Group A - Acetyl-Coenzyme A Carboxylase (ACCasej Herbicides C. Resistance to Group B - Acetolactate Synthase ( U S ) Herbicides D. Resistance to Group C - Photosystem I1 (PS 11) Herbicides E. Resistance to Other Major Herbicide Groups F. Multiple Herbicide Resistance 111. Managing Herbicide Resistance A. The Need for Integrated Weed Management (IWM) B. The Significance of Herbicide Mode of Action C. Use of Transgenic Herbicide-Resistant Crops D. IWM Strategies IV.Conclusions References

I. INTRODUCTION: MAJOR CROPS AND HERBICIDE MARKETS Sustainable management of weed populations in world agriculture is critical. Weeds infesting crops must be controlled or they reduce crop yields, hinder harvest operations, and contaminate produce. Successful farming systems maxi57 Aduenrer m &onmy, Volume f8 Copyright 0 IY91 by Academic Press. Inc. All rights of reproduction in any form reserved.

58

S. B. POWLES ET AL.

mize the growth and yield of crop plants while minimizing the growth and reproduction of weeds such that weed species, although not eradicated, can be maintained at densities below tolerable economic loss levels. Weed control is achieved through judicious management using good crop agronomy and a variety of weed control practices. These may include a range of cultural and physical techniques, for example, crop and pasture rotations, grazing animals, cultivation, and so on. However, in recent decades the predominant weed control method in many parts of the world is the use of effective and reliable chemical herbicides. During the past 50 years, the agrochemical industry has produced a range of increasingly sophisticated herbicides that have been enthusiastically adopted by growers worldwide (especially in industrialized nations), such that herbicides are the principal tool used to manage weeds within many cropping systems. Abundant food and fiber production is achieved due, in no small part, to the success of herbicides in controlling weedy plant species. Although herbicides are a major factor contributing to world crop production, their continuing success is threatened by the evolution of herbicide resistance in weedy plant species. Since the first report of a formerly susceptible weed population acquiring resistance (Ryan, 1970), there has been a rapid increase in the incidence of herbicide resistance worldwide, especially over the past decade. In this chapter, herbicide resistance is defined as the inherited ability of a weed population to survive a herbicide application that is normally lethal to the vast majority of individuals of that species. Testimony to the economic and scientific importance of herbicide resistance in weeds is the substantial literature on this subject (see reviews by LeBaron and Gressel, 1982; Caseley et al., 1991; Holt et al., 1993; Powles and Holtum, 1994). This contribution addresses what can be done to minimize and manage the threat of herbicide resistance to world agriculture. We briefly consider the current herbicide usage in major world crops and the consequent selection pressure imposed for herbicide resistance in weeds. We then move to address management strategies, considering both the grower perspective as well as the role of the international agrochemical industry. Finally, we emphasize that application of the principles and techniques of integrated weed management (IWM), embracing but not wholly dependent on herbicide technology, can provide sustainable strategies for weed management in cropping systems worldwide.

A. GLOBAL HERBICIDE USEAND CROPPRODUCTION During 1994, global sales of herbicides amounted to about $13 billion.* These cover a range of herbicide chemistries, but 12 groups of compounds account for over 80% of sales and 5 in particular, triazines, glyphosate salts, amides, sul*I995 sales have now been reported by Wood Mackenzie as $14.28 billion.

HERBICIDE RESISTANCE

59

Figure 1 Global herbicide sales in 1994 by herbicide chemistry (adapted from Wood Mackenzie, 1995).

fonylureas, and imidazolinones, make up half of the herbicide market (Fig. 1). Despite the array of individual herbicides marketed, and the large number of distinct chemistries that they represent, it is known that there is only a limited number of distinct modes of action of herbicides in commercial use. For example, the sulfonylurea and imidazolinone herbicides represent 10 and 7% of the global herbicide market, respectively. These two herbicide groups, along with the triazolopyrimidine and pyrimidinyl(thio)ethers, are all inhibitors of acetolactate synthase (ALS) and therefore share the same target site, an enzyme in the branched chain amino acid biosynthetic pathway. Likewise, the aryloxyphenoxypropanoate and cyclohexanedione herbicide classes share the same mode of action, inhibition of acetyl-coenzyme A carboxylase (ACCase), which interferes with fatty acid biosynthesis. Since many herbicide chemistries share modes of action, it is better to classify herbicides by mode of action rather than by chemical type alone. Classification of herbicides in this way allows grouping under 1 of 14 principal modes of action. The remaining herbicides can then be grouped as compounds with diverse modes of action. Several classification schemes for herbicides have been developed on a local scale, but significant effort is now being addressed to the introduction of a single mode of action classification that is accepted globally. The terminology in this chapter will reflect a classification scheme that has been proposed by Jutsum and Graham (1995a,b) as outlined in Table 1. These herbicides are used in all major crops in the world, and herbicide sales by crop sector are illustrated in Fig. 2. Current crop areas and production are summarized in Table 11.

60

S. B. POWLES ET AL. Table I Classification of Herbicides by Mode of Action

Group A B

Principal mode of action

Chemical families

Inhibitors of acetyl CoA carboxylase ( ACCase) Inhibitors of acetolacate synthase (ALS)

C

Inhibitors of photosynthesis at photosystem 11

D E

Inhibitors of tubulin formation Inhibitors of mitosis

F

Inhibitors of carotenoid biosynthesis

G

Inhibitors of protoporphyrinogen oxidase

H

Inhibitors of plastoquinone biosynthesis Disrupters of plant cell growth (hormone mimics) Inhibitors of cell wall synthesis Herbicides with diverse sites of action

I J

K

L M N 0

Disruptors of photosynthesis at photosystem I Inhibitors of EPSP-synthase Inhibitors of glutamine synthetase Uncouplers of energy transfer

Aryloxyphenoxypropanoates, cyclohexanediones Sulfonylureas, imidazolinones, triazolopyrimidines, pyrimidinyl(thio)ethers Triazines, triazinones, phenylureas. nitriles, benzothiadiazoles, acetamides. uracils, pyridazinones, phenyl pyridazines Dinitroanilines, pyridazines Carbamates, thiocarbamates, organophosphates Nicotinanilides, triazoles, pyridazinones, isoxazolidinones Diphenyl ethers, oxadiazoles, N-phenylphthalimides Triketones Benzoic acids. phenoxys, pyridine carboxylic acids Benzamides, dichlobenil Aminopropanoates. benzofurans, chloroacetamides, nitriles, phenylcarbamates, phthalamates, quinoline carboxylic acids Bipyridyls Glyphosate, sulfosate Glufosinate Organoarsenicals

Details of production and herbicide use for the eight major crops grown globally are given below. 1. Maize

The major maize growing area is the United States, which accounts for nearly 70% of the maize herbicide market, whereas Europe is second with a 17%share. A range of herbicides is available for preplant incorporation, preemergence, or postemergence treatments. There is significant usage of group C herbicides (dominated by atrazine) and group K products (alachlor and metolachlor), but sales of these products are generally declining, whereas group B products (nico-

HERBICIDE RESISTANCE

61

Oilseed rape (3 %

Rice (10 %)

Figure 2 Herbicide sales by crop sector (adapted from Wood Mackenzie, 1995).

sulfuron and flumetsulam) and other group K products (acetochlor and dimethenamid) are all increasing in market share. The maize herbicide sector is a key target for the agrochemical industry and six promising new molecules have been launched during this decade. However, these are all either group B or group K herbicides.

2. Cereals Global cereal production is dominated by wheat (63%) followed by barley (20%), sorghum (7%), oats (4%), and rye (3%). Western Europe, while accounting for 8% of the planted area, produces 17% of the yield and accounts for 43% of cereal herbicide usage. Numerous products are available to the grower, but these are mainly group A (including diclofop, fenoxaprop, clodinafop, and tralkoxydim), group B (including chlorsulfuron, metsulfuron, and triasulfuron), or group C (including isoproturon, chlorotoluron, and bromoxynil) herbicides. Ten new products, mainly group B herbicides, have been introduced during this decade.

3. Rice The rice herbicide market is dominated by Japan, but there is significant growth in countries such as South Korea. Herbicides are applied for the control of grasses, sedges, and broadleaf weeds, and mixtures account for much of the market. Key products include molinate, butachlor, and propanil, together with a number of increasingly important group B herbicides such as bensulfuron, pyrazosulfuron, cinosulfuron, and imazosulfuron.

S. B. POWLES ET AL.

62

Table I1

Current Crop Production Figures" Crop sector

Area grown (millions ha)

Yield (millions tonnes)

Maize Cereals Rice Soybeans Sugar beet Oilseed rape Cotton Vegetables. fruits, and nuts

131 395 I45 63 8 21 32

465 829 354 I38 282 27 18

NAh

I500

OAdapted from Wood Mackenzie (1995). bNA. not available.

4. Snybeans Nearly 90% of all soybean production is in the Americas, with the United States accounting for 64%of sales oi' soybean herbicides. The major compounds used are the group B herbicides, imazethJpyr, imazaquin, and chlorimuron, which account for nearly half the market. Other key herbicides used are trifluralin, pendimethalin, bentazone, metolachlcr, and metribuzin.

5. Sugar Beet Seventy-five percent of global sugar beet production and herbicide use is in Europe where the market is dominated by phenmedipham, metamitron, clopyralid, and chloridazon. Postemergence group A herbicides (fluazifop-p, quizalofop, sethoxydim, and cycloxydim) are continuing to play an important role in this market.

6. Oilseed Rape The major production areas for oilseed rape in order of area grown are India, China, Canada, and Western Europe. Despite occupying only 11% of the area cultivated, Western Europe accounts for over 60% of herbicides used. Key products include trifluralin, clopyralid, metazachlor, and propyzamide for broadleaf weed control, and group A herbicides (sethoxydim, quizalofop, haloxyfop, and fluazifop-p) for grass and weed control. Nonselective compounds such as glyphosate and paraquat are also used as preplant and preemergence applications,

HERBICIDE RESISTANCE

63

7. Cotton Thirty-two million hectares of cotton are planted globally yielding 18 million tons of lint. Cotton herbicides are used in all cotton producing countries, but over 50% of global sales are in the United States accounting for $275 million. Key products include group A herbicides (fluazifop-p and sethoxydim) for grass weed control and broad-spectrum selective compounds such as trifluralin, pendimethalin, fluometuron, nodurazon, and prometryn.

8. Vegetables, Fruit, and Nuts Global production of vegetables, fruit, and nuts amounts to 1.5 billion tons. The sector is very diverse and is typically satisfied by commodity products. Western Europe, East Asia, and North America account for nearly 90% of this $1.72 billion herbicide market. Fruit (primarily citrus, vines, and apples) accounts for over half the market, vegetables make up 40%, of which more than one-third is potatoes, and nuts represent 8% of the sector. Herbicide use in these markets is increasing mainly as a result of nonselective products such as paraquat, glyphosate, sulfosate, and glufosinate. These products have replaced laborintensive weeding and have increased grower awareness of the detrimental effect of weeds on crop yield. A wide range of other products is used, including atrazine, simazine, trifluralin, linuron, fluazifop-p, and norflurazon.

B. TRENDS IN HERBICIDE USE The total global herbicide market is around $13 billion per year and is predicted to grow to about $14 billion per year by the end of the century (Wood Mackenzie, 1995). The current market split as shown in Fig. 3 identifies the predominance of North America and Western Europe as the major markets. Further market growth is projected in North and Latin America, and in East Asia in the next 5-10 years. New products continue to be introduced, but the vast majority of novel active ingredients, although more selective, more effective, or environmentally safer, represent a small number of modes of action. In other words, few novel modes of action are being introduced. Many of the recent introductions include further group A or B herbicides. As will be discussed, resistance is now widespread and increasing to herbicides in groups A, B, and C.

S. B. POWLES ET AL.

64

Rest of World (4 YO) Western Europe (23 Y)

North America (42 Yo)

Eastern Europe (3%)

Latin America (9 %)

Figure 3 Herbicide sales by region (from Wood Mackenzie, 1995).

11. THE THREAT FROM HERBICIDE-

RESISTANT WEEDS Herbicide-resistant weeds threaten the continuing success of herbicide technology to contribute to world crop production. Over the past decade, there has been a rapid increase in the incidence of herbicide resistance worldwide. The evolution of herbicide resistance in populations of initially susceptible weedy plant species is a dramatic example of evolution in action (Maxwell and Mortimer, 1994; Jasieniuk and Morrison, 1994).

A. SELECTION PRESSURE AND THE EVOLUTION OF HERBICIDE RESISTANCE Classical Darwinian evolution was proposed to occur primarily through a series of gradual changes in response to a selection pressure. The material for these changes is the natural genetic variation inherent in populations. Such gradual evolution in response to small changes in the environment is usually the result of several or many genes acting in concert. However, the situation is quite different when a rapid and catastrophic change in the environment causes largescale mortality. For populations of annual weedy plants, the sudden application of a herbicide is an example of a catastrophic event causing upwards of 90% mortality. In such cases, survival can be endowed by the action of a single major gene (reviewed by Darmency, 1994a). Should resistant individuals be present in the population, although initially only a small fraction of the population, they produce seed and contribute a disproportionate number of progeny to the next

65

HERBICIDE RESISTANCE

generation (reviewed by Maxwell and Mortimer, 1994). Many factors contribute to the rate of appearance of herbicide resistance in a population. These include the initial frequency of herbicide-resistant individuals, the number of individuals treated, the mode of inheritance of the gene or genes endowing resistance, and the nature and extent of herbicide use. The interplay of these factors can be readily observed by simple modeling, and the results of such a modeling exercise will be used to illustrate key points. Most cases of field-selected herbicide resistance are due to the action of a single gene with a high degree of dominance (reviewed by Darmency, 1994a; Jasieniuk et al., 1996). The degree of dominance of resistance genes is particularly important for outcrossing weed species because fully recessive genes will tend to be diluted into heterozygous individuals by the larger number of susceptible alleles (see Fig. 4). In contrast, the level of dominance has little impact on the evolution of resistance for a species that is entirely selfing, because among the progeny of individuals carrying a resistance gene will always be some that are homozygous for that gene. The initial number of resistant individuals in a population will dramatically influence the appearance of resistance. The genes for resistance are assumed to occur as a result of the underlying mutation rate that may be in the order of 1 X 10-9 (Haughn and Somerville, 1987, 1990); however, there is no reason to

I

-

A

0

1

A

A

2 3 4 5 6 7 Herbicide applications

A

0

b

9

Figure 4 Predicted appearance of herbicide resistance following repeated selection with herbicide. Predictions for an outcrossing species with a dominant).( or recessive (0)mode of inheritance and for a selfing species with a dominant (0) or recessive (0) mode of inheritance. Model parameters are an unlimited population size, an initial frequency of resistance genes of 1 x lo-', 95% mortality of susceptible individuals due to herbicide application, 20% of seed remaining dormant, 20% mortality of seed in the soil, and 50% nonherbicidal mortality of seedlings.

66

S. B. POWLES ET AL.

assume that all genes have the same mutation rate. The number of resistant individuals present in an unselected population is influenced by the fitness of the resistant mutation (Gressel and Segal, 1982, 1990; Jasieniuk et al., 1996; Maxwell and Mortimer, 1994). In the most extreme case of all mutations being lethal, there will be no resistant individuals in the population. Studies of fitness of herbicide-resistant populations have found that, with the clear exception of target site-based triazine resistance, there is little or no fitness penalty associated with herbicide resistance (reviewed by Holt and Thill, 1994). In the case of a dominant allele with little or no fitness penalty, the frequency of resistant individuals will be 100-400 times the mutation rate (Jasieniuk et al., 1996). In the few studies that have examined the frequency of resistant individuals in previously unselected weed populations, the initial number of resistant individuals can be surprisingly high. Darmency and Gasquez (1990) found a frequency of 3 X lop3 triazine herbicide-resistant individuals in a population of Chenopodium album and J. M. Matthews and S. B. Powles (unpublished results) found an average frequency of 1 X 10-2 diclofop-methyl-resistant individuals in previously unselected farm populations of Loliurn rigidum. Clearly, there can be many more resistant individuals in a population than might be initially expected. Because herbicides in broad area agriculture are often used against huge numbers of individuals, even if resistant individuals exist at very low frequencies in the population, continued use of the same or similar herbicides will undoubtedly select for them. In addition to factors inherent in the plant population, the nature of herbicide use is also important. Highly efficacious herbicides impose a strong selection pressure for resistance. For example, a herbicide that controls 99% of a susceptible population will leave considerably fewer susceptible individuals to contribute to the next generation than a herbicide that gives 80% control. With all other factors being equal, this can make a difference of several to many additional applications before the appearance of noticeable resistance (see Fig. 5). Additionally, for species in which seed remains residual in the soil seedbank, the appearance of resistance will be delayed by the continued recruitment of susceptible individuals from the soil seedbank. Therefore, species that germinate readily from the seedbank, exhausting the seedbank rapidly, will respond more rapidly to herbicide selection than species in which the majority of seeds remain viable but dormant and emerge from the seedbank over many years. Taken together, these factors suggest that the evolution of herbicide resistance can vary widely between species and between herbicides. This is exactly what has been observed in the field. In some species, there is the potential for herbicide resistance to develop very rapidly. This has been illustrated dramatically in the case of L. rigidum, in which resistance to the group A and B herbicides can be observed following as few as three herbicide treatments (Tardif et al., 1993; Gill, 1995).

HERBICIDE RESISTANCE h

$?

v

67

100

ff 0

>

80

.E 60 9 .-ff v)

2

40

s

20

0 2. 0

0

e!

LL

0

4 6 8 1012 Herbicide applications Figure 5 Influence of herbicidal efficacy on predicted appearance of herbicide resistance. Herbicidal control was set at 95% (0)or 80%).( mortality of susceptible individuals. Other model parameters are an unlimited population size, an initial frequency of resistance genes of I X lo-’, an outcrossing species with a dominant mode of inheritance, 20% of seed remaining dormant, 20% mortality of seed in the soil, and 50% nonherbicidal mortality of seedlings. 0

2

There has been a number of recent reviews of herbicide resistance (see LeBaron and Gressel, 1982; Caseley et al., 1991; Moss and Rubin, 1993; Holt et a l . , 1993; Powles and Holtum, 1994). Therefore, we have confined discussion to a brief overview of resistance in weed species to the most important commercial herbicide groups.

B. RESISTANCE TO GROUP A -ACETYL-COENZYME A CAR~OXYLASE (ACCASE)HERJUCIDES Group A herbicides currently account for sales of $870 million per year and comprise two major classes of ACCase inhibitors-aryloxyphenoxypropanoates and cyclohexanediones. Key products include fluazifop-p, fenoxaprop-p, sethoxydim, tralkoxydim, and clethodim. Group A herbicides are only effective against grass species and are widely employed against grass weeds of broadleaf crops as well as in certain cereal crops in which there is selectivity. Since the first report of resistance to a group A herbicide (Heap and Knight, 1982), resistance has become widespread in L . rigidum in Australia and is an increasing problem in Lolium rnultijlorum, Avena furua, and Setaria viridis in North America (reviewed by Devine and Shimabukuro, 1994). There are also group A herbicide-resistant populations reported for a number of other important grass weed species

68

S. B. POWLES ET AL.

(Moss, 1990; Barrentine et al., 1992; Mansooji et al., 1992; Marshall et al., 1994; Stoltenberg and Wiederholt, 1995; Wiederholt and Stoltenberg, 1995). Most cases of resistance to the group A herbicides have occurred in broad area cropping, principally cereals, following 3- 10 treatments. Many cases of resistance to group A herbicides are due to a modification of the target site, ACCase (reviewed by Devine and Shimabukuro, 1994). There are many different patterns of ACCase target site cross-resistance, suggesting that there are several possible mutations of ACCase that endow resistance (Tardif and Powles, 1993; Tardif ef al., 1996). However, the specific mutations of ACCase that endow resistance are not yet known. A non-target site, enhanced metabolism mechanism of resistance to the group A herbicides is also evident. Enhanced metabolism endowing resistance has been observed for diclofop, fenoxaprop, fluazifop, or tralkoxydim in Alopecurus myosuroides (Mendenez et al., 1993; Hall et al., 1996a), L . rigidum (Preston et al., 1996a), and Digitaria sanguinalis (I. Hidayat and C. Preston, unpublished).

C. RESISTANCETO GROUPB -ACETOLACTATE SYNTHASE (ALS) HERBICIDES Group B herbicides have enjoyed spectacular success since their introduction during the 1980s. Group B herbicides include products from four chemical classes-sulfonylureas, imidazolinones, triazolopyrimidines, and pyrimidinyl(thio)ethers. Combined sales currently exceed $2 billion per year. A feature of the group B herbicides is that there are many herbicide products used in many different crops and targeted at a wide spectrum of weed species, including grasses, sedges, and many broadleaf weeds. Unfortunately, weed species have readily developed resistance to the group B herbicides and, given the wide range of herbicides used in many different crops, resistance is a major threat to their continued efficacy (reviewed by Saari et al., 1994). Resistance has become a serious problem in broadleaf weeds in North America and the grass weed L. rigidurn in Australia, but is not yet a significant problem elsewhere in the world (Saari et al., 1994). Resistance to the group B herbicides has most often appeared in weeds of cereal cropping; however, recent resistance in weeds of soybean in the United States (Schmitzer et al., 1993), and rice in California and Australia (Pappas-Fader et al., 1993; Hill et al., 1994) has been reported. All cases of resistance in broadleaf weeds documented, and many cases in grass weeds, are the result of a resistant ALS (reviewed by Saari et al., 1994). There is considerable diversity in the mutations within ALS shown to endow resistance (Guttieri et a l . , 1995; Bernasconi et al., 1995). In addition, a non-target site resistance mechanism is evident in some L. rigidum populations that are resistant to some group B herbicides due to enhanced herbicide metabolism mediated by cyto-

HERBICIDE RESISTANCE

69

chrome P450-dependent microsomal oxidases (Christopher et af., 1991, 1992; Cotterman and Saari, 1992).

D. RESISTANCE TO GROUP C - PHOTOSYSTEM I1 (PS II) HERBICIDES Group C herbicides also embrace a number of herbicide chemistries (triazines, triazinones, phenylureas, benzonitriles, and uracils). The triazine herbicides are the most important group and account for over $1.5 billion per year in sales, with the best known active ingredient being atrazine. Atrazine sales are being reduced, in part due to the development of resistance that has been confirmed in over 20 countries in biotypes of more than 58 species (reviewed by Moss and Rubin, 1993; Gronwald, 1994). Mostly, resistance has occurred in monoculture situations in which atrazine and/or simazine have been intensively used. Following eight or more applications, biotypes containing a mutation of the D1 protein in PS I1 have become established (reviewed by Gronwald, 1994). The mutation, a Ser264 to Gly change, dramatically reduces triazine binding to PS 11 (Pfister and Amtzen, 1979). In addition, it also decreases binding of the endogenous substrate plastoquinone to its binding site (Bowes et af., 1980). This results in a reduced rate of electron transport and increased susceptibility to photoinhibition (Hart and Stemler, 1990). As a result, triazine-resistant biotypes often show substantially reduced growth and reproductive output compared to susceptible biotypes (reviewed by Holt and Thill, 1994). Non-target site-based resistances to the group C herbicides, due to enhanced metabolism, are less common. A biotype of Abutilon threophrasti is resistant to atrazine due to enhanced metabolism mediated by glutathione-S-transferase(Anderson and Gronwald, 1991). Also, a number of biotypes of A . myosuroides (Kemp et al., 1990; Hall et al., 1996b) and L . rigidum (Bumet et al., 1993a,b) have resistance to triazine and/or substituted urea herbicides endowed by cytochrome P450-dependent enhanced herbicide metabolism.

E. RESISTANCE TO OTHER MAJORHERBICIDE GROUPS 1. Auxinic-Disrupting Herbicides (Group I) Despite the extensive and widespread use of the group I, auxin-disrupting herbicides ( 2 , 4 - ~and related compounds), resistance has appeared only infrequently (reviewed by Coupland, 1994). Resistance has occurred in relatively few broadleaf weed species from both cropping and pasture situations. Both en-

70

S. B. POWLES ET AL.

hanced metabolism and reduced binding of herbicide to auxin-binding proteins are documented mechanisms of resistance to these herbicides (Coupland et al., 1990; Webb and Hall, 1995).

2. Photosystem I-Disrupting Herbicides (Group L) Paraquat, and to a lesser extent diquat, have been in widespread use for nonselective weed control around the world for over three decades. Resistance to these group L herbicides has occurred in at least 16 weed species following either multiple applications of herbicide over a number of years or single annual applications for 12 or more years (reviewed by Preston, 1994). Given the widespread use of paraquat over long periods, the relative paucity of resistance cases indicates that resistance gene frequencies must be very low. Paraquat resistance has appeared principally in weeds of orchards and vineyards where total control of ground cover was desired. Additionally, paraquat resistance has developed in several weeds of alfalfa crops. The mechanisms of paraquat resistance are not as well characterized as those for PS 11-inhibiting herbicides, but for a number of biotypes, resistance is due to decreased herbicide translocation (Fuerst et al., 1985; Tanaka et al., 1986; Bishop et al., 1987; Preston et al., 1992; Purba et al., 1995).

E MULTIPLE HERBICIDE RESISTANCE An alarming occurrence in many L. rigidum and some A . myosuroides populations is the appearance of multiple herbicide resistance. Multiple resistance is defined as the expression of more than one resistance mechanism within individual plants (Hall et al., 1994), and can lead to simultaneous resistance to herbicides with different modes of action. Multiple resistance can develop sequentially from selection in time from several herbicides, but can also occur by repeated use of a single herbicide. Multiple resistance has evolved in some populations of A. myosuroides in Europe (Moss, 1990; Kemp et al., 1990; Mendenez et al., 1993) and numerous populations of L. rigidurn in Australia (Hall et al., 1994; Tardif et al., 1996). As outlined by Powles and Matthews (1992), a combination of huge numbers over vast areas, large genetic variation, cross-pollination, and strong selection pressure are the factors responsible for the appearance of multiple resistance in L. rigidum. Multiple resistance in L. rigidum is due to the simultaneous expression of resistant target site enzymes (both ACCase and ALS) as well as enhanced rates of herbicide metabolism which themselves endow resistance across dissimilar herbicide chemistries (Preston et al., 1996b). When cytochrome P450-dependent enhanced herbicide metabolism is one of the resistance mechanisms, the extent of resistance can be large because non-target site cross-resistance endowed by the enhanced activity of several

HERBICIDE RESISTANCE

71

isozymes can metabolize several, unrelated herbicide chemistries (Preston er a / ., 1996a). Multiple resistance is of particular practical concern because resistance can occur simultaneously to most or all herbicide options available to a grower. The widespread appearance of multiple resistant populations of L. rigidum in Australian agriculture has been the driving force for the local adoption of IWM strategies (see Section 111,D).

111. MANAGING HERBICIDE RESISTANCE

A. THENEEDFOR INTEGRATED WEEDMANAGEMENT (IWM) For many reasons, herbicides are very useful technological tools, as is evident from their widespread use (Figs. 1-3). An unfortunate result of the obvious benefits and continued success of herbicides has been, at least in developed nations, a focus on herbicide technology to the exclusion of other weed control methods (see Ennis, 1977; Thill et af., 1991; Zimdahl, 1991). Such major reliance on herbicides fails to recognize the power of the evolutionary process to lead to resistant plant populations. In agro-ecosystemsaround the world there is a continuing enrichment of herbicide-resistancegenes in weedy plant populations. What is perhaps surprising is that this overreliance on herbicides for weed control has occurred despite concerns raised about exclusive reliance on insecticides for pest insect control (see Perkins, 1982). For many years, it has been evident that insecticide resistance threatens the viability of insecticides for insect pest control. This realization resulted in the development of integrated pest management (IPM) strategies in which insecticides are only one, albeit the principal, method of insect control (Forrester, 1990; Denholm and Rowland, 1992). IWM is the weed equivalent of IPM and can be defined as the use of a range of control techniques, embracing physical, chemical, and biological methods in an integrated fashion without excessive reliance on any one single method (Powles and Matthews, 1992). As discussed by Swanton and Weise (1991), Thill er al. (1991), and Zimdahl (1991), weed science has been dominated by herbicide technology and, consequently, IWM strategies have received little attention. There is a pressing need to develop IWM strategies that include but are not solely focussed on herbicide technology.

B. THESIGNIFICANCE OF HERBICIDE MODEOF ACTION Despite the appearance of resistance, herbicides will remain an integral part of weed control systems for two reasons. First, herbicides are easy to use and are cost effective. Second, in most situations, resistance has occurred only to certain

72

S. B. POWLES ET AL.

herbicide groups and a common response has been to simply change to an alternate, effective herbicide. There is the potential to manage herbicide resistance by judicious herbicide rotations or mixtures not only after resistance has occurred but also as a component of a management strategy to minimize the likelihood of resistance occurring. Because most herbicide-resistant weed populations are resistant only to one herbicide mode of action, control can be achieved with an effective, alternative herbicide that has a different mode of action. Hence, herbicide rotation, after resistance has occurred, can be a viable strategy, providing cost-effective altemative herbicides are available. A wiser strategy would be to implement rotation or mixtures between different herbicide modes of action before resistance develops, so as to prolong the effective lives of both herbicides. Although pesticide rotations and/or mixtures to delay resistance are a vital component of insecticide (Roush, 1989) and fungicide (Cohen and Levy, 1990) management strategies, there has been limited attention paid to herbicide rotations or mixtures as a herbicide resistance minimization strategy. We believe that it is in the interests of both herbicide manufacturers and users to recommend herbicide mixtures or rotations as part of a resistance minimization strategy. Rotation or mixtures of herbicides with different modes of action should maintain control for a longer period than individual herbicides used alone (Fig. 6). For example, if the frequency of two unlinked resistance genes for each mode of action is 1 X 10-7, then the frequency of individuals with at least one copy of both genes will initially be about 4 X 10-14. Therefore, the chances of an individual weed having resistance simultaneously to both herbicides are low (Gressel and Segal, 1990). An exception to this occurs where resistance is endowed by a non-target site mechanism giving cross-resistance to other modes of action. Models predict that mixtures of two modes of action that are equally effective against the target weed species will delay resistance appearance in a population for longer than will a yearly rotation between the two components (Fig. 6). Use of an effective mixture will also maintain weed population numbers at a lower level for longer than a rotation; however, both strategies are better than use of each herbicide alone. The benefits of using a mixture, as opposed to a rotation of the two components, become even greater if the risk of developing resistance to one component is substantially lower than that of the other (see Wrubel and Gressel, 1994). Herbicides are often used in mixtures, but usually to broaden the spectrum of weeds controlled. However, resistance management requires that both components of the mixture control the same spectrum of weeds and have a similar biological persistence, yet have different target sites and are detoxified in a different manner (Wrubel and Gressel, 1994). Mixtures of two herbicides with both components at or near the label rate to control a target weed will obviously increase cost. Therefore, it may be difficult to persuade growers to use a more expensive mixture that provides no tangible &crease in weed control in any one

HERBICIDE RESISTANCE

0

2

4

73

6

8 10 12 14 16 Generations

Figure 6 Predicted appearance of herbicide resistance following repeated selection with herbicide rotations or mixtures. Predictions for resistance to herbicides A or B used alone (D), A).( and B (0)used in a rotation, or for the two herbicides used as a mixture (0). Model parameters are an unlimited population size, an outcrossing species, a dominant mode of inheritance for each resistance gene, a frequency of resistance genes for each mode of action of 1 X lo-', 95% control of susceptible individuals by each herbicide, no cross-resistance or negative cross-resistance, 20% of seed remaining dormant. 20% mortality of seed in the soil, and 50% nonherbicidal mortality of seedlings.

season, despite the additional long-term benefits in delaying the appearance of herbicide resistance. Mixtures are not immune to resistance and simultaneous resistance to both components of a mixture has been observed (Burnet et al., 1991; Wrubel and Gressel, 1994). Therefore, in an ideal situation, use of herbicide rotations or mixtures should be within IWM strategies. Crucial to the success of herbicide mixtures and rotations as resistance management strategies is grower recognition of the classification of herbicides into distinct modes of action. However, given the large number of herbicide products available, current labeling practices, and persuasive marketing strategies, the great majority of growers do not identify herbicides according to mode of action but rather according to retail product names. Frequently, from the promotion literature available it is easy for a grower to assume that a new product is a new, unique chemical rather than a new member of an already well-established mode of action. Until growers can easily identify and record the modes of action of herbicides it will be difficult to achieve adoption of herbicide mixtures and rotations as part of a herbicide resistance management strategy. To achieve this end, herbicide users must be provided with adequate information on herbicide modes of action. This could be achieved by introducing user-friendly symbols that identify the specific mode of action. In response to widespread herbicide resistance in Australia, a herbicide mode

74

S. B. POWLES ET AL.

of action labeling system has been developed in a collaborative initiative between the Australian Herbicide Resistance Action Committee and academia (R. Davis and S. B. Powles, unpublished results). It is now mandatory for all herbicide labels in Australia to carry a large symbol identifying the herbicide mode of action. Alphabetical symbols were chosen after lengthy debate within the industry in which numbers were rejected because a similar system is used to indicate the poison schedule of particular products, and a color-based system was rejected because some individuals are color-blind. The system is “user friendly” because the user only needs to recognize and record that a different alphabetical symbol indicates a different mode of action. The introduction of this system in Australia was accompanied by a wide-ranging extension program to educate growers and advisers. A numerical system has been introduced in Canada, but it is not mandatory, and other countries, such as the United States, are at various stages in developing systems. The agrochemical industry’s international Herbicide Resistance Action Committee (HRAC) (Fenton and Jutsum, 1991; Jutsum and Shaner, 1992; O’Keefe et al., 1993) is now progressing the development of a single, uniform, accurate set of guidelines to be introduced globally (Table I; Jutsum and Graham, 1995a,b). This is based on the Australian classification of herbicides by mode of action. To achieve mode of action classification on herbicide labels as an aid in resistance management requires the cooperation of all interested parties and has the potential to avoid situations in which growers (unwittingly)rotate herbicides within the same mode of action even when they are from different chemical groups (e.g., sulfonylureas-v-imidazolinones).

C. USEOF TRANSGENIC HERBICIDE-RESISTANT CROPS Until now, the commercialization of herbicides has been determined by whether or not a new herbicide discovery is lethal to particular crops and particular weed species. Herbicides are either selective in that they kill weeds without injuring certain crops or nonselective in that they kill most annual weed and crop species. A good selective herbicide can be used in particular crops but will injure other crop species. In the herbicide discovery process, crop selectivity is largely the result of the skill of the inventive scientist and/or serendipity. Crop selectivity is usually due to the ability of the crop to detoxify the herbicide at a faster rate than that which occurs in susceptible weed species (Shimabukuro, 1985). In most cases, this means that there are certain weed species also with the ability to degrade the herbicide and hence the herbicide is never fully effective on these species. This is why the great majority of selective herbicides control only a certain number of weed species rather than the full spectrum of weed species present in fields. In contrast, nonselective herbicides (e.g., glufosinate, glyphosate, and paraquat) are lethal to a wide spectrum of annual weed and crop plants

HERBICIDE RESISTANCE

75

and there is little or no metabolism of these herbicides in either crops or weeds. The ideal herbicide would control a wide spectrum of weed species, be harmless to crop species of choice, and have desirable environmental characteristics. For the reasons outlined previously, such an ideal herbicide is difficult to obtain by the conventional chemical synthesis herbicide discovery process. However, plant genetics and biotechnology offer an alternative method of attaining this goal. Recent biotechnological advances in the plant sciences provide a genetic means of endowing crop species with resistance to herbicides of choice (herbicide-resistant crops). These exciting developments mark a new era in agriculture because herbicide-resistant crops will be the first products of biotechnology to be grown on an economic scale. The idea to endow particular crops with resistance to specific herbicides is not new. For example, the herbicide chlorotoluron is lethal to some wheat varieties, whereas other varieties tolerate this herbicide (Ryan and Owen, 1983; Cabanne et al., 1985). Hence, plant breeders in Europe have sought to incorporate chlorotoluron tolerance into specific hybrids. In another example, a weedy Brussica species containing a triazine-resistant PsbA gene following exposure to triazine herbicides was hybridized with commercial rapeseed to produce a canola cultivar (OAC Triton) resistant to triazine herbicides (Beversdorf et al., 1980; Beversdorf and Hume, 1984). Unfortunately, the PsbA triazine resistance gene also carries a fitness penalty, thus this cultivar of canola yields significantly less (20-30%) than conventional hybrids (Beversdorf et af., 1988). Although the “Triton” canola does offer the grower the opportunity to control particularly difficult weeds with a triazine herbicide, the associated yield penalty has limited the commercial success of this variety. Even in 1986, only 3.2% of the canola grown in western Canada was sown to OAC Triton. The subsequent introduction of canola-selective herbicides has led to the commercial demise of this cultivar (Marshall, 1987). There is considerable interest in the development of transgenic herbicideresistant crops, as can be seen by the growing numbers of publications in this area and the many excellent reviews (Mazur and Falco, 1989; Bright, 1991; Rubin, 1991; Cole, 1994; Dekker and Duke, 1995; Townson, 1996; Duke, 1996). Here, we will consider some of the advantages and disadvantages inherent in this new technology, review the current status of herbicide-resistant crops that are either on the market or are soon to be commercialized, and discuss how these crops might be used as part of an IWM system, particularly with respect to managing herbicide-resistant weeds.

1. The Need for Herbicide-Resistant Crops The agrochemical industry is finding it increasingly difficult to discover and develop novel herbicides which will satisfy regulatory, grower, and consumer

76

S. B. POWLES ET AL.

requirements, which are themselves becoming more demanding. Agrochemical research is not only high risk and extremely expensive, but it is also a very competitive business; the net result is that in recent years there has been a reduction in the number of companies willing to invest in research and development, thus reducing further the probability of bringing novel herbicides to the market. Transgenic crop resistance to herbicides provides useful tools for the grower. Although in principle it is feasible to select or engineer plants for resistance to any herbicide, the reality is that there are relatively few forms of herbicide-resistant crops that are either in development or have been comrnercialized. For this reason, the remainder of this section will deal only with these specific cases (i.e., resistance to glyphosate, glufosinate, bromoxynil, and ALS inhibitors).

2. Glyphosate Resistance (Group M) Glyphosate is a postemergence, nonselective herbicide (Grossbard and Atkinson, 1985). Research at Monsanto and elsewhere over the past decade has led to the identification of glyphosate resistance genes and their subsequent successful expression in crop species where resistance to glyphosate is endowed (Comai et al., 1985; Waters 1991; Dyer, 1994a; Wells, 1995). Glyphosate kills plants by inhibiting the biosynthesis of aromatic amino acids (Amrhein et al., 1980; Steinrucken and Amrhein, 1980), specifically by inhibiting EPSP synthase, a key enzyme in the shikimate pathway. The degree to which plants metabolize and thus detoxify glyphosate in vivo is uncertain (Dyer, 1994a; Komossa et al., 1992); the general belief being that plants do not metabolize the herbicide in vivo to any great extent, although there are reports of plant cell cultures degrading glyphosate to aminomethylphosphonic acid (AMPA). The situation with soil microbes is much clearer where the herbicide is rapidly degraded to AMPA (Torstensson, 1985; Jacob et al., 1988; Pipke and Amrhein, 1988). Several approaches have been adopted to develop crops resistant to glyphosate. The major success has been the introduction of a gene from a soil bacterium (Agrobacteriurn, strain CP4) encoding a glyphosate-resistant EPSP synthase into plants (Padgette et al., 1996). In some crops, this gene does not confer a sufficient level of resistance to glyphosate. Therefore, another approach has been to supplement the EPSP synthase resistance mechanism with the introduction of a glyphosate-oxidizing gene (the GOX gene) obtained from a soil bacterium, Achrumabacter, which was found in a glyphosate waste stream treatment facility (Barry et al., 1992). Monsanto, in collaboration with several seed companies, is currently developing several glyphosate-resistantcrops, including soybeans, cotton, corn, and canola, for the American markets, whereas sugar beet and oilseed rape are being evaluated in Europe (Waters, 1991; Kishore et al., 1992; Tirnmerman, 1995;

HERBICIDE RESISTANCE

77

Wells, 1995). These crops are to be sold under the Roundup Ready trademark, and Roundup Ready soybean, which contains the CP4 EPSP synthase gene, will be commercialized in the United States and Argentina in 1996. Yield trials in 1992 and 1993 confirmed that glyphosate-resistant soybean has a high level of tolerance to twice the average use rate of glyphosate and that applications of 0.63 kg acid equivalent (a.e.) ha-' to 0.84 kg a.e. ha-l provide effective seasonlong, broad-spectrum weed control in soybeans. Transformed cotton plants, containing the resistant enzyme, also show high levels of resistance, with no yield penalties evident even when treated with 1.68 kg a.e. ha-1 of glyphosate. Two glyphosate-resistant canola lines, which contain both the CP4 EPSP synthase and the GOX genes for resistance, are being developed primarily for westem Canada, with first sales expected in 1996. The introduction of glyphosateresistant sugar beet in Europe is probably several years from commercialization, but initial reports are promising, with three applications of glyphosate (equivalent to 720 g a.e. ha-1 total) matching conventional herbicide programs (Dekker and Duke, 1995). By comparison, obtaining sufficient levels of resistance in corn is proving more difficult, but lead lines with both resistant EPSP synthase and the GOX gene are being evaluated.

3. Glufosinate Resistance (Group N) The herbicides glufosinate (L-phosphinothricin) and bialaphos (phosphinothricyl alanylalanine) are inhibitors of glutamine synthetase, the only plant enzyme that can detoxify ammonia released by nitrate reduction, photorespiration, and amino acid accumulation (Devine et al., 1993). Inhibition of this enzyme results in an accumulation of ammonia which leads to death of the plant (Tachibana et ul., 1986). Both products are used as nonselective herbicides and are active following postemergent applications on a wide range of grass and broadleaved weeds. A gene endowing glufosinate resistance (the bur gene) was isolated from Streptomyces hygroscopicus (Thompson et al.,1987) and found to encode for a polypeptide with acetyl transferase activity, which converts phosphinothricin into a nontoxic acetylated product (Kumada et ul., 1988). Another group independently isolated and characterized a similar resistance gene from S. viridochromogenes, which also codes for a phosphinothricin acetyl transferase, or PAT enzyme (Strauch et d., 1988). AgrEvo, in collaboration with many seed companies, is developing a wide range of crops that are resistant to glufosinate by virtue of the bar or PAT genes. At least 20 different crop species have now been transformed with the bur gene (Mullner et al., 1993; Dekker and Duke, 1995; Vasil, 1996), including many broadleaved crops, such as tobacco, lupins, tomato, potato, sugar beet, rape, and soybean (De Block et al., 1987; Donn et al., 1990a), together with cereals such

78

S. B. POWLES ET AL.

as maize (Donn et al., 19990b; Morocz et al., 1990), wheat (Vasil et al., 1993), and rice (Datta et al., 1992). Once introduced, the resistance genes have been shown to confer high levels of resistance to glufosinate and bialaphos, irrespective of species, and to have no effect on crop performance (Botterman and Leemans, 1989). Although most crops appear to be amenable to this technology, research has been focused on developing glufosinate resistance in four major crops-canola, maize, soybean, and sugar beet (Rasche et al., 1995). Of these, canola is the most advanced, with the first varieties being registered in Canada in 1995. The variety, known as Innovator, has been sold on a relatively small area (ca. 16,300 ha) and, although it could not be exported in 1995, it has been well received by growers. Application of glufosinate at 300-600 g ha-' consistently gave good weed control and proved safe to the crop. Similar trials with winter oilseed rape in Europe also appear to be promising with excellent crop safety and good, broad-spectrum weed control at rates of 300-600 g ha-' (Rasche et al., 1995).

4. Bromoxynil Resistance (Group C) Bromoxynil is a PS I1 inhibitor that inhibits electron transport. A bromoxynilresistant cotton has been developed by Calgene in conjunction with Rh6nePoulenc (Freyssinet and DeRose, 1994) and there has also been some interest in transforming other species with resistance to this herbicide (Khan ef al., 1994; Dear et al., 1995). The bxn gene derived from the soil bacterium Klebsiella ozaenae encodes a nitralase protein that degrades bromoxynil (McBride et al., 1986). Transgenic cotton containing the bxn gene can efficiently metabolize bromoxynil to 3,5-dibromo-4-hydroxybenzoicacid and other products (Stalker et al., 1995). Cotton plants containing the bxn gene resist bromoxynil, and lint yield, fiber strength, and fiber length of BXN cotton are not reduced relative to commercial varieties (Stalker et al., 1995). Conditional approval was granted in May 1995 for the sale of bromoxynil for use on BXN cotton in the United States.

5. ALS-Inhibitor Resistance (Group B) Maize with resistance to the imidazolinone herbicides is currently marketed by ICI Seeds (Zeneca) and Pioneer Hybrid International in the United States. These lines have been selected either from embryogenic maize callus cultures (Anderson and Georgeson, 1989; Shaner et al., 1996) or by pollen mutagenesis (Shaner ef al., 1996). ALS inhibitor-resistant cultivars of canola have been generated by selection in cell and microspore culture (Swanson et al., 1988, 1989) or by Agrobacterium-mediated gene transfer (Moloney et al., 1989). Mass selection of mutagenized seed has been used to generate ALS inhibitor-resistant soybean (Sebastian et al., 1989). Resistance in all of these herbicide-resistant crops is due

HERBICIDE RESISTANCE

79

to the possession of a herbicide-resistant ALS (Swanson er al., 1988; Moloney et al., 1989; Newhouse et al., 1991a,b). Canola expressing a herbicide-resistant ALS is highly resistant to sulfonylurea and triazolopyrimidine herbicides and has similar maturity, seed yield, and oil content as standard cultivars (Blackshaw et al., 1994). The imidazolinone-resistant maize lines display varying levels of resistance to ALS inhibitors (Siehl et al., 1996) and also have no yield penalties (Shaner, 1994).

6. Advantages and Disadvantages of Transgenic Herbicide-Resistant Crops The pros and cons of herbicide-resistant crops have been, and will continue to be, debated vigorously within the scientific and political communities (e.g., Fraley et al., 1987; Gressel, 1992; Hill, 1995; Miflin, 1995). A list of the perceived benefits that may arise through the use of such crops are as follows: selective control of “difficult weeds” (Bright, 1991), increased use of “environmentally benign” herbicides (Dekker and Duke, 1995), reduced herbicide rates (Waters, 1991), additional weed control options available to the grower (Netzer, 1984), greater timing flexibility for growers (Wells, 1995), reduced soil erosion (Waters, 1991), control of herbicide-resistant weeds (Gressel, 1992; Shaner, 1994), selective control of parasitic weeds and use in third world crops (Gressel, 1993), use in minor crops and forestry (Marshall, 1995), economic advantages to growers and consumers (Bright, 1991), improved returns for herbicide discovery in industry (Marshall, 1995). In contrast, the perceived risks associated with herbicide-resistant crops can be summarized as follows: greater use and reliance on a few herbicides (Burnside, 1992), unfavorable environmental impact (Hill, 1995), gene transgression to wild or cultivated plants (Darmency, 1994b), pleiotropic effects of transgenes (Dyer, 1994b), development of volunteer weed problems (Williamson, 1994), development of herbicide-resistant weeds (Goldberg et al., 1990), public and consumer acceptability of transgenic plants and their products, reduced biodiversity (Fox, 1994), development of monopolistic chemical/seed companies (Deo, 1991).

80

S. B. POWLES ET AL.

Notwithstanding the validity or otherwise of some of the above-mentioned arguments, it is evident that herbicide-resistant crops are now a commercial reality. These issues have been reviewed in detail by Dyer ef al. (1993), Dekker and Duke (1995), and Duke (1996); therefore, the remainder of this section will consider benefits and risks with regard to weed control only. The introduction of transgenic herbicide-resistant crops will offer growers increased flexibility in managing their weed problems. This will be particularly important for the management of herbicide-resistant weeds, where additional modes of action can be introduced to supplement modes of action that have been compromised due to resistance. As described under Section 11, herbicide resistance has developed due to overreliance on herbicides from a few modes of action. If sensibly managed, as proposed under Section III,B, transgenic herbicide-resistant crops will provide growers with an additional tool with which to manage herbicide-resistant weeds. One issue of obvious importance is whether the use of herbicide-resistant crops will lead to the evolution of new cases of herbicide-resistant weeds. Glyphosate and glufosinate-resistanttransgenic crops will allow the introduction of new modes of action into specific cropping practices because these herbicides have never been used in a selective manner before. The same is true to a lesser extent for bromoxynil; however, the group B herbicides are already used extensively in many cropping situations (Section I). In the context of herbicide-resistant crop use, resistant weeds might arise in two ways. The first being through herbicide selection pressure as described under Section 11, the second through gene flow from resistant crops to closely related weed species. The first point will be illustrated by two contrasting examples: glyphosate and group B herbicides. From the experience to date it is clear that any allele(s) for glyphosate resistance must be quite rare because glyphosate has been used worldwide for 20 years with only one suspected case of resistance being reported (L. rigidurn). This fact, coupled with the view that glyphosate usage will not increase dramatically through the introduction of resistant crops (Waters, 1991), has led some to predict that the likelihood of weed resistance developing is extremely low (Kishore and Shah, 1988; Waters, 1991). Although the estimates of the initial frequency of resistance to glyphosate are extremely low (much less than I in 10-6) and the probability of plants possessing the enzyme systems to degrade the herbicide is also low, it must be recognized that the evolutionary selection pressures as described under Section I1 will favor the proliferation of individuals that possess any means of surviving glyphosate application. Hence, if the glyphosate use pattern was to change dramatically through the introduction of glyphosate-resistantcrops, such that this herbicide is used in much the same way that the group A, B , and C herbicides have been used, then there must be the possibility that resistance will evolve. Therefore, the introduction and use of herbicide-resistant crops should be within IWM systems.

HERBICIDE RESISTANCE

81

In contrast to glyphosate, there are already many cases of weeds that have evolved resistance to the group B herbicides (reviewed in Saari et al., 1994). Increased use of these herbicides will only exacerbate the problem. Clearly, group B-resistant cultivars must be rotated with crops in which non-Group B herbicides, and other weed control techniques, can be used. The second way that weeds might acquire resistance through use of herbicideresistant crops is via gene migration from the transgenic crop to a related weed species. This subject has been extensively debated and reviewed (Gressel, 1993; Raybould and Gray, 1993; Dale and Irwin, 1995; Hill, 1995; Wilkinson et al., 1995). As Dyer (1994b) concludes, “the likelihood of cross-hybridisation depends entirely on the crop in question, its breeding system and the incidence of wild relatives with which hybridisation can occur.” Thus, cross-pollinated crops with close weedy relatives, such as sugar beet, Lolium, and some Brussicu species, may present a relatively high risk, whereas there would be minimal risk with self-pollinating crops such as wheat and barley. This is obviously an area of some concern, yet should not be a reason for rejection of herbicide-resistant crops, and only highlights the importance of using herbicide-resistant crops wisely. It is vitally important that herbicide-resistant crops are not seen as a panacea for control of herbicide-resistant weeds, but rather as the introduction of excellent new mode of action herbicides. These herbicides and the transgenic herbicide-resistant crops must themselves be used as part of an IWM strategy.

D. IWM STRATEGIES The appearance of herbicide-resistant weeds is testimony to the powerful evolutionary forces that act on weed populations to overcome the initially high mortality from an efficient herbicide. We should be ever mindful that exclusive reliance on any single, highly efficient control method, chemical or nonchemical, can fail as a result of evolutionary forces finding a way to circumvent the control method. Exclusive herbicide reliance can result in redundancy due to resistance. This is likely, ultimately, to be the fate of the group A and B herbicides (ACCase and ALS inhibitors). Given these facts, what should be the response of herbicide producers, advisers, and users? As new herbicide discovery is the ruison detre of international research and development-based corporations’-they will continue to search for new modes of action, as well as work to exploit biotechnological tools that endow crop species with resistance to desirable herbicides. Both of these avenues will and should be vigorously pursued. However, herbicide discovery is becoming increasingly more difficult such that the introduction of new, unique mode-of-action herbicides, with all of the characteristics necessary for regulatory and commercial success, will be rare events. Given the paucity of new products, and notwithstanding market realities, industry has become increasingly

S. B. POWLES ET AL.

more interested in preserving the longevity of existing modes of action (Roush and Powles, 1996). The establishment and activities of HRAC are clearly a major step in this direction (Fenton and Jutsum, 1991; Jutsum and Shaner, 1992; O’Keefe et al., 1993; Jutsum and Graham, 1995a,b). Minimizing resistance should also be the goal of the herbicide user who would like to be able to use a particular mode of action for a lengthy period without failure due to resistance. Therefore, all sectors have the same goal of preserving herbicide efficacy. It should not be discounted that for both herbicide manufacturers and users there are powerful and quite rational forces to exploit fully a herbicide chemistry in the short term without regard for resistance development. Dependent on patent protection, competitor products, and a range of other factors, it may be in the interest of the herbicide inventor to maximize sales of a new product in the short term (Roush and Powles, 1996). Similarly, economic factors often constrain growers to herbicide use strategies that lead to resistance. In free-market societies, whether or not some form of resistance management should occur is a choice made by individuals on a case-by-case basis. It should be recognized that there has been little evidence that either herbicide producers or users have consciously restrained herbicide use patterns in order to minimize or delay the onset of resistance. Of course, changes are instituted when resistance has occurred, but mostly to an alternative, effective herbicide that is then used in the same manner as that for the previous herbicide. This strategy works when there are alternative options available. A dramatic example in which resistance is forcing the adoption of IWM strategies exists for L. rigidum in Australia. L . rigidum is a major weed of cropping throughout southern Australia and thousands of populations have evolved herbicide resistance (Powles and Matthews, 1992; Matthews, 1994). Multiple resistance is also evident in L. rigidum and extends across many herbicide chemistries. In the worst cases, there are virtually no selective herbicides that remain effective. Under these circumstances, farmers have been forced to adopt IWM strategies because selective herbicide options are not effective against multiple resistant L. rigidum. It is now clear from farmer experience and longterm field experiments that IWM strategies, with limited herbicide inputs, can be very effective in controlling herbicide-resistant L. rigidum (J. M. Matthews, R. S. Llewellyn, T. G. Reeves and S. B. Powles, unpublished results). The IWM strategies employed to manage herbicide-resistant L. rigidum include various combinations of factors including pasture and crop rotations, variation in seeding date, use of nonselective herbicides, high crop seeding rates, vigorous crop growth, and capture of weed seed in the harvest operation (Powles and Matthews, 1996). Growers have confirmed that IWM strategies can successfully manage herbicide-resistant L. rigidum while maintaining farm profitability and sustainability. It should be noted that despite resistance, herbicides remain an integral component of IWM strategies for L. rigidurn control in cropping systems

HERBICIDE RESISTANCE

83

in Australia. Thus, even when faced with multiple resistance, herbicides have remained pivotal to the success of an IWM strategy. This Australian experience with IWM of herbicide resistance has lessons elsewhere in the world. To date, however, farmers only adopt such IWM strategies after the development of herbicide resistance. It is much more difficult to achieve adoption of IWM practices (however desirable) in order to minimize the potential for resistance to occur. There is clearly a shared responsibility for the agrochemical companies, distributors, and advisory personnel to educate and inform growers of the benefits associated with adopting IWM practices.

IV. CONCLUSIONS This chapter is concerned with the threat to the productivity of world agriculture imposed by the evolution of herbicide-resistant weed populations. Implicit in this chapter is that herbicides should and will continue to be a major tool for weed control. We believe that modem herbicides are a cost-effective, efficient, and environmentally benign means for obtaining weed control. Proponents rightly identify the benefits of herbicides in achieving weed control as well as positive environmental influences in substituting for soil cultivation in weed management. Reliance on herbicides for weed control is expected to continue because there is no attractive superior technology available (Duke et al., 1993). However, for sustainable weed management to be achieved, changes to current herbicide use patterns are required. In many agricultural systems, herbicide use does not reflect biological and evolutionary realities. The clear benefits and continued success of herbicides has resulted in a focus on herbicide technology to the exclusion of other control methods. Marketing strategies, price, ease of use, and efficacy of herbicides has resulted in overreliance on herbicides as a control method with more holistic IWM approaches to weed management discarded. Such reliance on a single control method fails to acknowledge the power of the evolutionary process and results in herbicide-resistant weed populations. As outlined under Section 11, herbicide resistance genes to many of the major world herbicides have been enriched in weed populations. In some cropping regions, resistance is now widespread to herbicide groups A, B, and C in several prominent weeds. Of even greater concern is the appearance of multiple resistance. Multiple-resistant populations of weeds, such as L. rigidum and A . myosuroides, are current indicators of potential worst-case weed problems of the future. Resistance will continue to increase if present herbicide use patterns are not altered. We cannot expect that resistance can be overcome by the regular introduction of new mode-of-action herbicides effective on existing resistant weeds.

S. B. POWLES ET AL.

84

It must be emphasized that herbicide resistance can be delayed or prevented from occurring and can be managed when it does occur. To achieve this, herbicides should be used within IWM strategies in which herbicides are only one of the control methods employed. Within IWM strategies, weed populations are not exposed to constant selection pressures. Relatively simple measures, such as correct herbicide mode-of-action rotation or mixtures, are effective components of an IWM strategy. To assist the employment of rotations and mixtures, the international introduction of uniform labeling for herbicide mode of action (Jutsum and Graham, 1995a,b) should occur. Equally, the introduction of transgenic herbicide-resistant crops, in particular glyphosate- and glufosinate-resistant crops, will have an impact on weed management. In essence, the use of applications of glyphosate and glufosinate in crops represents the introduction of two new modes of action. Weed biotypes resistant to existing selective herbicides will be controlled by these herbicides. This and other benefits will ensure that these transgenic crops will be readily adopted by growers. However, it is equally important that these new tools should not be relied on to the exclusion of other weed control methods. They too should be used within IWM strategies. The key message that can be taken from this chapter is the realization that herbicides are a precious resource of great importance to agricultural productivity. They need to be used sparingly rather than wastefully if they are to retain their utility for more than a few years. It is hoped that this chapter will increase knowledge and lead to action to help prevent widespread herbicide resistance in world agriculture.

REFERENCES Amrhein, N., Deus, B., Gehrke, P., and Steinrucken, H. C. (1980). The site of the inhibition of the shikimate pathway by glyphosate. Plum Physiol. 66, 830-834. Anderson, M. P., and Gronwald, J. (1991). Atrazine resistance in a velvetleaf (Aburilon rhreophrusri) biotype due to enhanced glutathione-S-transferase activity. Plunr Physiol. 96, 104- 109. Anderson, P. C., and Georgeson, M. (1989). Herbicide-tolerant mutants of corn. Genome 31,994999.

Bamntine, W. L., Snipes, C. E., and Smeda, J. J. (1992). Herbicide resistance confirmed in johnsongrass biotypes. Mississippi Agric. For. Exp. Srn. Res. Rep. 17, 1-5. Barry, G., Kishore, G., Padgette, S., Taylor, M., Kolacz, K., Weldon, M., Re, D., Eichholtz, D., Fincher, K., and Hallas, L. (1992). Inhibitors of amino acid biosynthesis: Strategies for imparting glyphosate tolerance to crop plants. In “Biosynthesis and Molecular Recognition of Amino Acids in Plants”(B. K. Singh, H.E. Flores, and J. C. Shannon, Eds.), pp. 139-145. American Society of Plant Physiology, Rockville, MD. Bernasconi, P., Woodworth, A. R., Rooen, B. A,, Subramanian, M. V., and Siehl, D. L. (1995). A naturally occurring point mutation confers broad range tolerance to herbicides that target acetolactate synthase. J. Biol. Chem. 270, 17381-17385. Beversdorf, W. D., and Hume, D. J. (1984). OAC Triton spring rapeseed. Can. J. Plunr Sci. 64, 1007-1009.

HERBICIDE RESISTANCE

85

Beversdorf, W. D., Hume, D. J., and Donnelly-Vanderloo, M. J. (1988). Agronomic performance of triazine-resistant and susceptible reciprocal spring canola hybrids. Crop Sci. 28, 932-934. Beversdorf, W. D., Weiss-Lerman, J., Erickson, L. R., and Souza Machado, V. (1980). Transfer of cytoplasmically-inherited triazine resistance for birds-rape to cultivated oilseed rape (Brassica campesrris and B . napus). Can. J . Genet. Cyrol. 22, 167-172. Bishop, T., Powles, S. B., and Comic, G . (1987). Mechanism of paraquat resistance in Hordeurn glaucum. 11. Paraquat uptake and translocation. Aust. J . Plant Physiol. 14, 539-547. Blackshaw, R. E., Kanashiro, D., Moloney, M. M., and Crosby, W. L. (1994). Growth, yield, and quality of canola expressing resistance to acetolactate synthase inhibiting herbicides. Can. J . Plant Sci. 74, 745-751. Botterman, J., and Leemans, J. (1989). Discovery, transfer to crops, expression and biological significance of a bialaphos resistance gene. In “Prospects for Amino Acid Biosynthesis Inhibitors in Crop Protection and Pharmaceutical Chemistry” (L. C. Copping, J. Dalziel, and A. D. Dodge, Eds.), pp. 61-66. British Crop Protection Council, Surrey, UK. Bowes, J., Crofts, A. R., and Arntzen, C. J. (1980). Redox reactions on the reducing side of Photosystem I1 in chloroplasts with altered herbicide binding properties. Arch. Biochem. Biophys. 200, 303-308. Bright, S. W. I. (1991). Opportunities for introducing herbicide tolerant crops. In “Herbicide Resistance in Weeds and Crops” (J. C. Caseley, G . W. Cussans, and R. K. Atkin, Eds.), pp. 365374. Butterworth-Heinemann, Oxford, UK. Burnet, M. W. M.. Hildebrand, 0. B., Holtum, J. A. M., and Powles, S. B. (1991). Amitrole, triazine, substituted urea, and metribuzin resistance in a biotype of rigid ryegrass (Loliurn rigidurn). Weed Sci. 39, 317-323. Burnet, M. W. M., Loveys, B. R., Holtum, J. A. M., and Powles, S. B. (1993a). A mechanism of chlorotoluron resistance in Lolium rigidum. Planta 190, 182- 189. Burnet, M. W. M., Loveys, B. R., Holtum. J. A. M., and Powles, S. B. (1993b). Increased detoxification is a mechanism of simazine resistance in Lolium rigidum. Pestic. Biochem. Physiol. 46, 207-218. Burnside, 0. C. (1992). Rationale for developing herbicide-resistant crops. Weed Technol. 6 , 621625. Cabanne, F., Galliardon, P., and Scalla. R. (1985). Phytotoxicity and metabolism of chlorotoluron in two wheat varieties. Pestic. Biochem. Physiol. 23, 212-220. Caseley, J. C., Cussans, G. W., and Atkin, R. K. (Eds.) (1991). “Herbicide Resistancein Weeds and Crops.” Butterworth-Heinemann, Oxford, UK. Christopher, J. T., Powles, S. B., Liljegren, D. R., and Holtum, J. A. M. (1991). Cross-resistance to herbicides in annual ryegrass (Loliumrigidum). 11. Chlorsulfuron resistance involves a wheatlike detoxification system. Plant Physiol. 95, 1036- 1043. Christopher, J. T., Powles. S . B., and Holtum, J. A. M. (1992). Resistance to acetolactate synthase inhibiting herbicides in annual ryegrass (Lolium rigidum) involves at least two mechanisms. Planr Physiol. 100, 1909-1913. Cohen, Y., and Levy, Y. (1990). Joint action of fungicides in mixtures: Theory and practice. Phyroparasirica 18, 159- 169. Cole, D. J. (1994). Molecular mechanisms to confer herbicide resistance. In “Molecular Biology of Crop Protection” (G, Marshall and D. Walters, Eds.), pp. 146-176. Chapman & Hall, London. Comai, L., Facciotti, D., Hiatt, W. R.. Thompson, G . , Rose, R. E., and Stalker, D. M. (1985). Expression in plants of a mutant aroA gene from Salmonella ryphimurium confers tolerance to glyphosate. Nature 317, 741-744. Cotterman, J. C., and Saari, L. L. (1992). Rapid metabolic inactivation is the basis for crossresistance to chlorsulfuron in diclofop-methyl-resistant rigid ryegrass (Loliurnrigidurn) biotype SR4l84. Pesric. Biochem. Phvsiol. 43, 273-280.

86

S. B. POWLES ET AL.

Coupland, D. (1994). Resistance to the auxin analog herbicides. In “Herbicide Resistance in Plants: Biology and Biochemistry” (S. B. Powles and J. A. M. Holtum, Eds.), pp. 171-214. Lewis, Boca Raton, FL. Coupland, D., Lutman, P. J. W., and Heath, C. (1990). Uptake, translocation and metabolism of mecoprop in a sensitive and resistant biotype of Stellaria media. Pestic. Biochem. Physiol. 36, 61-67. Dale, P. J., and Irwin, J. A. (1995). The transfer of herbicide resistance genes into weedy relatives. In “Herbicide-Resistant Crops and Pastures in Australian Farming Systems” ( G . D. McLean and G. Evans, Eds.), pp. 201-206. Bureau of Resource Sciences, Canberra, Australia. Darmency, H. (1994a). Genetics of herbicide resistance in weeds and crops. In “Herbicide Resistance in Plants: Biology and Biochemistry” (S. B. Powles and 1. A. M. Holtum, Eds.), pp. 263-297. Lewis, Boca Raton, FL. Darmency, H. (1994b). The impact of hybrids between genetically modified crop plants and their related species: Introgressson and weediness. Mol. Ecol. 3,37-40. Darmency, H., and Gasquez, J. (1990). Appearance and spread of triazine-resistance in common lambsquarters (Chenopodium album). Weed Technol. 4, 173- 177. Datta, S. K., Datta, K.,Soltanifar, N., Donn, G., and Potrykus, I. (1992). Herbicide-resistant Indica rice plants from IRRl breeding line IR72 after PEG-mediated transformation of protoplasts. Plant Mol. Biol. 20, 619-629. Dear, B. S., Higgins, T. J., Sandral, G. A., and Bellotti, W. D. (1995). The integration of herbicidetolerant pasture legumes into Australian farming systems. In “Herbicide-Resistant Crops and Pastures in Australian Fanning Systems.” (G.D. McLean and G. Evans, Eds.), pp. 101-125. Bureau of Resource Sciences, Canberra, Australia. De Block, M., Bottermann, J., Vandewiele, M., Dockx, T., Thoen, C., Gossele, V., Movva, N. R., Thompson, C., Van Montagu, M., and Leemans, J. (1987). Engineering herbicide resistance in plants by expression of a detoxifying enzyme. EMBO J. 6, 2513-2518. Dekker, J., and Duke, S. 0. (1995). Herbicide-resistant field crops. Adv. Agron. 54, 69-116. Denholm, I., and Rowland, M. W. (1992). Tactics for managing pesticide resistance in arthropods: Theory and practice. Annu. Rev. Enromol. 37, 91-1 12. Deo, S. D. (1991). Implications of biotechnologies for third world agriculture: Lessons of the past and prospects. In “Biotechnologies in Perspective: Socioeconomic Implications for Developing Countries” (A. Sasson and V. Costatini, Eds.), pp. 19-25. UNESCO, Paris. Devine, M. D., and Shimabukuro, R. H. (1994). Resistance to acetyl coenzyme A carboxylase inhibiting herbicides. In “Herbicide Resistance in Plants: Biology and Biochemistry” (S. B. Powles and J. A. M. Holtum, Eds.), pp. 141-169. Lewis, Boca Raton, FL. Devine, M., Duke, S. 0..and Fedtke, C. (1993). “Physiology of Herbicide Action.” Prentice Hall, Englewood Cliffs, NJ. Donn, G . , Dirks, R., Eckes, P., and Uijtewaal, B. (1990a). Transfer and expression of modified phosphinothricin-acetyltransferasegene from Srreptomyces viridochromogenes in tomato, melon, carrot and strawbeny. In “Abstract VII International Congress of Plant Tissue and Cell Culture,” p. 176. Amsterdam. Donn, G., Nilges, M., and Morocz, S. (1990b). Stable transformation of maize with a chimaeric modified phosphinothricin-acetyltransferasegene from Streptornyces viridochrornogenes. In “Abstract VII International Congress of Plant Tissue and Cell Culture,” p. 53. Amsterdam. Duke, S. 0. (Ed.) (1996). “Herbicide Resistant Crops: Agricultural, Environmental, Economic, Regulatory, and Technical Aspects,” in press. Lewis, Chelsea, MI. Duke, S. 0..Menn, J. J., and Plimmer, J. R. (1993). Challenges of pest control with enhanced toxicological and environmental safety. An overview. In “Pest Control with Enhanced Environmental Safety,” (S. 0. Duke, J. J. Menn, and J. R. Plimmer, Eds.), pp. 188-204. American Chemical Society, Washington, DC.

HERBICIDE RESISTANCE

87

Dyer, W. E. (1994a). Resistance to glyphosate. In “Herbicide Resistance in Plants: Biology and Biochemistry” (S. B. Powles and I. A. M. Holtum, Eds.), pp. 229-241. Lewis, Boca Raton, FL . Dyer, W. E. ( 1994b). Herbicide-resistant crops: A weed scientist’s perspective. Phytoprotection ), 7 1-77, 75(S~ppl. Dyer, W. E., Hess, F. D., Holt, I. S., and Duke, S. 0. (1993). Potential benefits and risks of herbicide-resistant crops produced by biotechnology. Hortic. Rev. 15, 367-408. Ennis, W. B. (1977). Integration of weed control technologies. In “Integrated Control of Weeds” (J. D. Fryer and S. Matsunaka, Eds.), pp. 229-243. University of Tokyo Press, Tokyo. Fenton, R., and Jutsum, A. R. (1991). Herbicide resistant weeds: Developing our defences. Shell Agric. 11, 22-25. Forrester, N. W. ( I 990). Designing, implementing and servicing an insecticide resistance management strategy. Pestic. Sci. 28, 167-179. Fox, J. L. (1994). Do transgenic crops pose ecological risks? BiolTechnology 12, 127-128. Fraley, R., Kishore, G., Gasser, C., Padgette, S., Horsch, R., Rogers, S., Della-Cioppa, G., and Shah, D. ( 1987). Genetically-engineered herbicide tolerance-Technical and commercial considerations. In “British Crop Protection Conference-Weeds,” Vol. 2, pp. 463-471, British Crop Protection Council. Surrey, UK. Freyssinet, G . , and DeRose, R. T. (1994). Development of genetically modified crops resistant to 5, 3-7. herbicides and pests. Agro-Food-Indust.-Hi-tech Fuerst, E. P., Nakatani, H. Y., Dodge, A. D., Penner, D., and Amtzen, C. J. (1985). Paraquat resistance in Conyza. Plant Physiol. 77, 984-989. Gill, G. (1995). Development of herbicide resistance in annual ryegrass populations (Lolium rigidum Gaud.) in the cropping belt of Westem Australia. Ausr. J . Exp. Agric. 35, 67-72. Goldburg, R., Rissler, J., Shand, H., and Hassebrook, C. (1990). “Biotechnology’s Bitter Harvest: Herbicide Tolerant Crops and the Threat to Sustainable Agriculture.” Environmental Defense Fund, City of New York, NY. Gressel, J. (1988). Multiple resistances to wheat selective herbicides: New challenges to molecular biology. Ogord Survey Plant Mol. Cell Biol. 5 , 195-203. Gressel, J. (1992). The need for herbicide-resistant crops. In “Resistance ’91: Achievements and Developments in Combating Pest Resistance” (1. Denholm, A. Devonshire, and D. Hollomon, Eds.), pp. 283-294. Elsevier, London. Gressel, J. (1993). Advances in achieving the needs for biotechnologically-derived herbicide resistant crops. Plant Breed. Rev. 11, 155-198. Gressel, J., and Segal, L. A. (1982). Interrelating factors controlling the rate of appearance of resistance: The outlook for the future? In “Herbicide Resistance in Plants” (H. M. LeBaron and J. Gressel, Eds.), pp. 325-347. Wiley, New York. Gressel, J., and Segal, L. A. (1990). Modelling the effectiveness of herbicide rotations and mixtures as strategies to delay or preclude resistance. Weed Technol. 4, 186-198. Gronwald, J. W. ( 1994). Resistance to Photosystem I1 inhibiting herbicides. In “Herbicide Resistance in Plants: Biology and Biochemistry” ( S . B. Powles and I. A. M. Holtum, Eds.), pp. 27-60. Lewis, Boca Raton, FL. Grossbard, E., and Atkinson, D. (1985). “The Herbicide Glyphosate.” Buttenvorths, London. Guttieri, M. J., Eberlein, C. V., and Thill, D. C. (1995). Diverse mutations in the acetolactate synthase gene confer chlorsulfuron resistance in kochia (Kochia scoparia) biotypes. Weed Sci. 43, 175-178. Hall, L. M., Holtum, J. A. M.,and Powles, S. B. (1994). Mechanisms responsible for cross resistance and multiple resistance. In “Herbicide Resistance in Plants: Biology and Biochemistry” (S. B. Powles and J. A. M. Holtum, Eds.), pp. 243-261. Lewis, Boca Raton, FL. Hall, L. M., Moss, S. R., and Powles, S. B. (1996a). Mechanisms of resistance to aryloxyphenoxy-

88

S. B. POWLES ET AL.

propionate herbicides in two resistant biotypes of Alopecurus myosuroides (black-grass). Herbicide metabolism as a cross-resistance mechanism. Pestic. Biochem. Physiol., in press. Hall, L. M., Moss, S. R., and Powles, S. B. (1996b).Mechanism ofresistance tochlorotoluron in two biotypes of the grass weed Alopecurus myosuroides. Pesric. Biochem. Physiol. 53, 180-192. Hart, J. J., and Stemler, A. (1990). High light-induced reduction and low light-enhanced recovery of photon yield in triazine-resistant Brassica napus L. Plant Physiol. 94, 1301-1307. Haughn, C., and Somerville, C. R. (1987).Selection for herbicide resistance at the whole plant level. In “Biotechnology in Agricultural Chemistry” (H. M. LeBaron, R. 0. Mumma, R. C. Honeycutt, and J. H. Duesing, Eds.), pp. 98-107. American Chemical Society, Washington, DC. Haughn, C., and Somerville, C. R. (1990). A mutation causing imidazolinone resistance maps to the Csrl locus of Arubidopsis thaliana. Plant Physiol. 92, 1081-1085. Heap, I., and Knight, R. (1982). A population of ryegrass tolerant to the herbicide diclofopmethyl. J. Aust. Inst. Agric. Sci. 48, 156-157. Hill, J. E. ( I 995). Herbicide-tolerant crops-Environmentalists concerns and regulatory responses. In “Proceedings of the Brighton Crop Protection Conference-Weeds,” Vol. 3, pp. 1027- 1034. British Crop Protection Council, Surrey, UK. Hill, J. E., Smith, R. J., Jr., and Bayer, D. E. (1994). Rice weed control: Current technology and emerging issues in temperate rice. Ausr. J. Exp. Agric. 34, 1021-1029. Holt, J. S . , and Thill, D. C. (1994). Growth and productivity of resistant plants. In “Herbicide Resistance in Plants: Biology and Biochemistry” (S. B. Powles and J. A. M. Holtum, Eds.), pp. 299-316. Lewis, Boca Raton, FL. Holt, J. S., Holtum, J. A. M., and Powles, S. B. (1993). Mechanisms and agronomic aspects of herbicide resistance. Annu. Rev. Plant Physiol. Plant Mol. Biol. 44, 203-209. Jacob, G. S . , Garbow, J. R., Hallas, L. E., Kimack, N. M., Kishore, G . M., and Schaefer, J. (1988). Metabolism of glyphosate in Pseudomoms sp. strain LBr. Appl. Environ. Microbiol. 54,29532958. Jasieniuk, M., and Morrison, B. D. (1994). Population genetics and the evolution of herbicide resistant weeds. Phyzoprotection 75(Suppl.), 25-35. Jasieniuk, M., BrW-Babel, A. L., and Momson, I. N. (1996). The evolution and genetics of herbicide resistance in agricultural weeds. Weed Sci. 44, 176-193. Jutsum, A. R., and Graham, J. C. (1995a). Managing weed resistance: Activities of the Herbicide Resistance Action Committee (HRAC). Pesr. Outlook 6 , 30-33. Jutsum, A. R., and Graham, 3. C. (1995b). Managing weed resistance: The role of the agrochemical industry. In “British Crop Protection Conference-Weeds,” Vol. 2, pp. 557-566. British Crop Protection Council. Surrey, UK. Jutsum, A. R., and Shaner, D. (1992). Herbicide resistance: The stance of the agrochemical industry. In “Proceedings of the First International Weed Control Congress,” (R. G. Richardson, Ed.), Vol. 2, pp. 244-246. The Weed Science Society of Victoria, Melbourne, Australia. Kemp, M. S., Moss, S. R., and Thomas, T. H. (1990). Herbicide resistance in Alopecurus myosuroides. In “Managing Resistance to Agrochemicals: From Fundamental Research to Practical Strategies” (M. B. Green, H. M. LeBaron, and W. K. Moberg, Eds.), pp. 376-393. American Chemical Society, Washington, DC. Khan, M. R. I . , Tabe, L. M., Heath, L. C., Spencer, D., and Higgins, T. J. (1994).Agrobacteriummediated transformation of subterranean clover (Trifolium subterraneurn L.).P lanr Physiol. 105, 81-88. Kishore, G . M., and Shah, D. M. (1988). Amino acid biosynthesis inhibitors as herbicides. Annu. Rev. Biochem. 57, 627-663. Kishore, G. M., Padgette, S. R., and Fraley, R. T. (1992). History of herbicide-tolerant crops. methods of development and current state of the art-Emphasis on glyphosate tolerance. Weed Technol. 6, 626-634.

HERBICIDE RESISTANCE

89

Komossa, D., Gennity, D. I., and Sandermann, H. (1992). Plant metabolism of herbicides with C-P bonds: Glyphosate. Pesric. Biochem. Physiol. 43, 85-94. Kumada, Y., Anzai, H., Takano, E., Murakanii, T., Hara, 0.. Itoh, R., Imai, S., Satoh, A., and Nagaoka, K. (1988). The bialaphos resistance gene (bar) plays a role in both self-defence and bialaphos in Streptomyces hygroscopicus. J. Antibiotics 16, 1839- 1845. LeBaron, H. M., and Gressel, J. (Eds.) (1982). “Herbicide Resistance in Plants.” Academic Press, New York. Mansooji, A. M., Holtum, J. A. M., Boutsalis, P., Matthews, J. M., and Powles, S. B. (1992). Resistance to aryloxyphenoxypropionate herbicides in two wild oat species (Avena fatua and Avena sterilis ssp. ludoviciana). Weed Sci. 40,599-605. Marshall, G. (1987). Implications of herbicide-tolerant cultivars and herbicide-resistant weeds for weed control management. In “Proceedings of the British Crop Protection ConferenceWeeds,” Vol. 2, pp. 489-498. British Crop Protection Council, Surrey, UK. Marshall, G. ( I 995). Herbicide-tolerant crops-The new green revolution. In “Proceedings of the Brighton Crop Protection Conference-Weeds,’’ Vol. 3, pp. 1049-1056. British Crop Protection Council, Surrey, UK. Marshall, G., Kirkwood, R. C., and Leach, G. E. (1994). Comparative studies on graminicideresistant and susceptible biotypes of Eleusine indicu. Weed Res. 34, 177- 185. Matthews, J. M. (1994). Management of herbicide resistant weed populations. In “Herbicide Resistance in Plants: Biology and Biochemistry” (S. 9. Powles and J. A. M. Holtum, Eds.), pp. 317335. Lewis, Bofa Raton, FL. Maxwell, B. D., and Mortimer, A. M. (1994). Selection for herbicide resistance. In “Herbicide Resistance in Plants: Biology and Biochemistry” (S. 9. Powles and J. A. M. Holtum, Eds.), pp. 1-25. Lewis, Boca Raton, FL. Mazur, B. J., and Falco, S. C. (1989). The development of herbicide resistant crops. Annu. Rev. Plant Physiol. PIanr Mol. Biol. 40,441-470. McBride, K. E . , Kenny, J. W., and Stalker, D. M. (1986). Metabolism of the herbicide bromoxynil by Klebsiella pneumoniae subsp. ozaenae. Appl. Environ. Mirrobiol. 52, 325-330. Mendenez. J., De Prado, R., Jomn, J., and Taberner, A. (1993). Penetration, translocation and metabolism of diclofop-methyl in chlorotoluron-resistant and susceptible biotypes of Alopecurus myosuroides. In “Proceedings Brighton Crop Protection Conference-Weeds,’’ Vol. 1, pp. 2 13220. British Crop Protection Council, Surrey. UK. Miflin, B. J. (1995). Herbicide tolerant crops-A perspective. In “Proceedings of the Brighton Crop Protection Conference-Weeds,’’ Vol. 3, pp. 783-790. British Crop Protection Council, Surrey, UK. Moloney, M. M., Walker, J. M., and Sharma, K. K. (1989). High efficiency transformation of Brassica napus using Agrobacrerium vectors. Plant Cell Rep. 8, 238-242. Morocz, S., Donn, G., Nemeth, J., and Dudits, D. (1990). An improved system to obtain fertile regenerants via maize protoplasts isolated from a highly embryogenic suspension culture. Theor. Appl. Genet. 80, 721-726. Moss, S. R. (1990). Herbicide cross-resistance in slender foxtail (Alopecurus myosuroides). Weed sci. 38, 492-496. Moss, S. R . , and Rubin, B. (1993). Herbicide-resistant weeds: A worldwide perspective. J . Agric. Sci, Cambr. 120, 141-148. Mullner, H., Eckes, P., and Donn, G. (1993). Engineering crop resistance to the naturally occurring glutamine synthetase inhibitor phosphinothricin. In “Pest Control with Enhanced Environmental Safety” (S. 0. Duke, J. J. Menn, and J. R. Plimmer, Eds.), pp. 38-47. American Chemical Society, Washington, DC. Netzer, W. J. (1984). Engineering herbicide tolerance: When is it worthwhile? BiolTechnology 2, 939-944.

90

‘

S. B. POWLES ET AL.

Newhouse, K. E., Singh, B., Shaner, D., and Stidham, M. (1991a). Mutations to corn (Zeamays L.) conferring resistance to imidazolinone herbicides. Theor. Appl. Genet. 83, 65-70. Newhouse, K. E., Wang, T., and Anderson, P. C. (1991b). Imidazolinone-tolerant crops. In “The Imidazolinone Herbicides” (D. L. Shaner and S. L. O’Connor, Eds.), pp. 139-150. CRC Press, Boca Raton, FL. O’Keefe, M. G., Graham, J. C., and Jutsum, A. R. (1993). The Herbicide Resistance Action Committee (HRAC). Pest. Outlook 4, 15-19. Padgette, S., Re, D. B., Barry, G. F., Eichholtz, D. E., Delanney, X.,Fuchs, R. I . , Kishore, G. M., and Fraley, R. T. (1996). New weed control opportunities: Development of soybeans with a Roundup Ready gene. In “Herbicide-Resistant Crops-Agricultural, Environmental, Economic, Regulatory and Technical Aspects” (S. 0. Duke, Ed.), pp. 53-84. Lewis, Chelsea, MI. Pappas-Fader, T., Cook, J. F., Butler, T., Lana, P. J., and Hare, J. (1993). Resistance of Californian arrowhead and smallflower umbrella plant to sulfonylurea herbicides. Proc. West. SOC. Weed Sci. 76. Perkins, J. H. (1982). “Insects, Experts and the Insecticide Crisis.” Plenum, New York. Pfister, K., and Amtzen, C. J. (1979). The mode of action of Photosystem 11-specific inhibitors in herbicide-resistant weed biotypes. Z. Naturforsch. 54f, 996- 1009. Pipke, R., and Amrhein, N. (1988). Degradation of the phosphonate herbicide glyphosate by Arthrobacter atrocyaneus ATCC 13752. Appl. Environ. Microbiol. 54, 1298-1296. Powles, S. B., and Holtum, J. A. M. (Eds.) (1994). “Herbicide Resistance in Plants: Biology and Biochemistry.” Lewis, Boca Raton, FL. Powles, S. B., and Matthews, J. M. (1992). Multiple herbicide resistance in annual ryegrass Lolium rigidum: A driving force for the adoption of integrated weed management. In “Resistance ’91: Achievements and Developments in Combating Pesticide Resistance” (I. Denholm, A. L. Devonshire, and D. W. Holloman, Eds.), pp. 75-88. Elsevier, London. Preston, C. (1994). Resistance to Photosystem I disrupting herbicides. In “Herbicide Resistance in Plants: Biology and Biochemistry” (S. 9 . Powles and J. A. M. Holtum, Eds.), pp. 61-82. Lewis, Boca Raton, FL. Preston, C., Holtum, J. A. M., and Powles, S. 9 . (1992). On the mechanism of resistance to paraquat in Hordeurn glaucum and H. leporinum. Delayed inhibition of photosynthetic 0, evolution after paraquat application. Plant Physiol. 100, 630-636. Preston, C., Tardif, F. J., Christopher, J. T., and Powles, S. 9 . (1996a). Multiple resistance to dissimilar herbicide chemistries in a biotype of Lolium rigidum due to enhanced activity of several herbicide degrading enzymes. Pestic. Biochem. Physiol., 54, 123- 134. Preston, C., Tardif, F. J., and Powles, S. 9. (1996b). Multiple mechanismsendow multiple herbicide resistance in Lolium rigidum. In “Molecular Genetics and Ecology of Pesticide Resistance” (T. M. Brown, Ed.), in press. American Chemical Society, Washington, DC. Purba, E., Preston, C., and Powles, S. B. (1995). The mechanism of resistance to paraquat is strongly temperature dependent in resistant Hordeurn leporinum Link and H. glaucum Steud. Planra 196,464-468. Rasche, E., Cremer, J., Donn, G., and Zink, J. (1995). The development of glufosinate ammonium tolerant crops into the market. In “Proceedings of the Brighton Crop Protection ConferenceWeeds,” Vol. 3, pp. 791-800. British Crop Protection Council, Surrey, UK. Raybould, A. F., and Gray, A. J. (1993). Genetically modified crops and hybridization with wild relatives: A UK perspective. J. Appl. Ecol. 30, 199-219. Roush, R. T. (1989). Designing resistance management programs: How can you choose? Pestic. Sci. 26,423-441. Roush, R. T.,and Powles, S. 9. (1996). Pesticide resistance: Why be concerned. I n “Pesticide Resistance: Prevention and Management” ( G . Bourdbt and M. Suckling, Eds.), pp. 3-16. New Zealand Plant Protection Society, Rotorua, New Zealand.

HERBICIDE RESISTANCE

91

Rubin, B. (1991). Herbicide resistance in weeds and crops: Progress and prospects. In “Herbicide Resistance in Weeds and Crops” (I. C. Caseley, G. W. Cussans, and R. K. Atkin, Eds.), pp. 387-414. Butterworth-Heinemann, Oxford, UK. Ryan, G. F. (1970). Resistance of common groundsel to sirnazine and atrazine. Weed Sci. 18,614616. Ryan, P. J . , and Owen, W. J. (1983). Metabolism of chlortoluron in tolerant and sensitive cereal varieties. Asp. Appl. B i d . 3, 63-72. Saari, L. L., Cotterman, J. C., and Thill, D. C. (1994). Resistance to acetolactate synthase inhibiting herbicides. In “Herbicide Resistance in Plants: Biology and Biochemistry” (S. B. Powles and J. A. M. Holturn, Eds.), pp. 83-169. Lewis. Boca Raton, FL. Schmitzer. P. R.. Eilers, R. J.. and Cseke. C. (1993). Lack of cross-resistance of imazaquin-resistant Xanthium strumarium acetolactate synthase to flumetsulam and chlorimuron. PIanr Physiol. 103, 281-283. Sebastian, S. A., Fader, G. M., Ulrich, J. F., Forney, D. R., and Chaleff, R. S. (1989). Semidominant soybean mutation for resistance to sulfonylurea herbicides. Crop Sci. 29, 14031408. Shaner, D. (1994). Herbicide-resistant crops in resistant weed management: An industrial perspective. Phytopatho1og.v 75(Suppl.), 79-84. Shaner, D. L., Bascomb, N. F., and Smith, W. (1996). Imidazolinone-resistant crops: Selection. characterization and management. In “Herbicide Resistant Crops: Agricultural, Environmental, Economic, Regulatory, and Technical Aspects” (S. 0. Duke, Ed.), pp. 143-158. Lewis, Chelsea. MI. Shimabukuro, R. H. (1985). Detoxification of herbicides. In “Weed Physiology” (S. 0. Duke, Ed,), Vol. 11, pp. 215-240. CRC Press, Boca Raton, FL. Siehl, D. L., Bengtson, A. S.. Brockman, J. P., Butler, J. H., Kraatz, W., Larnoreaux, R. J., and Subramanian, M. V. ( 1996).Patterns of cross-tolerance to herbicides inhibiting acetohydroxyacid synthase in commercial corn varieties designed for tolerance to imidazolinones. Crop Sci. 36, 274-278. Stalker, D. M., Kiser. J. A,, Baldwin, G., Coulombe, B.. and Houck, C. M. (1995). Cotton weed control using the BXN’” system. In “Herbicide-Resistant Crops and Pastures in Australian Farming Systems” (G. D. McLean and G . Evans, Eds.), pp. 73-87. Bureau of Resource Sciences, Canberra, Australia. Steibrucken, H. C., and Amrhein, N. (1980). The herbicide glyphosate is a potent inhibitor of 5-enolpyruvylshikimic acid-3-phosphate synthase. Biochem. Biophys. Res. Commun. 94, 12071212. Strauch, E., Arnold, W., Alijah. R., Wohlleben, W., Puhler, A,, Eckes, P., Donn, G., Uhlmann, E., Hein, F., and Wengenmayer, F. (1988). Chemical synthesis and expression in plant cells and plants of phosphinothricin resistance gene with plant preferred codons. Eur. Pat. Appl. EP275957. Stoltenberg, D. E., and Wiederholt, R. J. (1995). Giant foxtail (Setariafaberi) resistant to aryloxyphenoxypropionate and cyclohexanedione herbicides. Weed Sci. 43, 527-535. Swanson, E. B., Coumans, M. P., Brown, G . L., Patel, J. D., and Beversdorf, B. D. (1988). The characterisation of herbicide tolerant plants in Brassica napus L. after in vitro selection of microspores and protoplasts. Plant Cell Rep. 7 , 83-87. Swanson, E. B., Herrgesell, M. J., Arnoldo, M., Sippell, D. W., and Wong, R. S. C. (1989). Microspore mutagenesis and selection: Canola plants with field tolerance to the imidazolinones. Theor. Appl. Genet. 78, 525-530. Swanton, C. J., and Weise, S. F. (1991). Integrated weed management: The rationale and approach. Weed Technol. 5 , 657-663. Tachibana, K., Watanabe. T., Sekizawa, Y., and Takematsu, T. (1986). Inhibition of glutamine

92

S. B. POWLES ET AL.

synthetase and quantitative changes of free amino acids in shoots of bialaphos-treated Japanese barnyard millet. J. Pestic. Sci. 11, 27-31. Tanaka, Y.,Chisaka, H., and Saka, H. (1986). Movement of paraquat in resistant and susceptible biotypes of Erigeron philadelphicus and E. canadensis. Physiol. Plant 66, 605-608. Tardif, F. J . , and Powles, S. B. (1993). Target site-based resistance to herbicides inhibiting acetylcoA carboxylase. In “Proceedings Brighton Crop Protection Conference-Weeds,” Vol. 2, pp. 533-540. British Crop Protection Council, Surrey, UK. Tardif, F. J., Holtum, J. A. M., and Powles, S.B. (1993). Occurrence of a herbicide-resistant acetylcoenzyme A carboxylase mutant in annual ryegrass (Lolium rigidurn) selected by sethoxydim. Planra 190, 176-181. Tardif, F. J., Preston, C., and Powles, S. B. (1996). Mechanisms of herbicide multiple resistance in Lolium rigidurn. In “Proceedings of International Symposium on Weed and Crop Resistance to Herbicides” (R.De Prado, Ed.), in press. Kluwer Academic, Dordrecht, The Netherlands. Thill, D. C., Lish, 1. M., Callihan, R. H., and Bechinski, E. J. (1991). Integrated weed management-A component of integrated pest management: A critical review. Weed Technol. 5,648656. Thompson, C. J., Movva, N. R., Tizard, R., Crameri, R., Davies, J. E., Lauwereys, M., and Botteman, J. (1987). Characterization of the herbicide-resistance gene bar from Streptomyces hygroscopicus. Eur. Mol. Bio. Org. J . 6, 2513-2518. Timmerman, B. R. L. (1995). Sugar beets tolerant to non-selective herbicides-A seed company’s perspective. In “Proceedings of the Brighton Crop Protection Conference-Weeds,” Vol. 3, pp. 801-809. British Crop Protection Council, Surrey, UK. Torstensson, L. (1985). Behaviour of glyphosate in soils and its degradation. In “The Herbicide Glyphosate” (E. Grossbard and D. Atkinson, Eds.), pp. 137-150. Butterworths, London. Townson, J. K. (1996). Herbicide resistance. In “Molecular Mechanisms of Drug Resistance” (J. Hayes and R. Wolf, Eds.), in press. Academic Publishers, Switzerland. Vasil, V. (1996). Phosphinothricin-resistant crops. In “Herbicide-Resistant Crops: Agricultural, Environmental, Regulatory, and Technical Aspects” (S. 0. Duke, Ed.), pp. 85-92. Lewis, Chelsea, MI. Vasil, V., Castillo, A. M., Fromm, M. E., and Vasil, I. K. (1993). Herbicide resistant fertile transgenic wheat plants obtained by microprojectile bombardment of regenerable embryonic callus. BiolTechnology 10, 667-674. Waters, S. (1991). Glyphosate-tolerant crops for the future: Development, risks, and benefits. In “Proceedings of the Brighton Crop Protection Conference-Weeds,” Vol. I , pp. 165-170. British Crop Protection Council, Surrey, UK. Webb, S. R., and Hall, J. C. (1995). Auxinic herbicide-resistant and -susceptible wild mustard (Sinapsis arvensis) biotypes: Effect of auxinic herbicides on seedling growth and auxin-binding activity. Pestic. Biochern. Physiol. 52, 137- 148. Wells, B. H. (1995). Development of glyphosate tolerant crops into the market. In “Proceedings of the Brighton Crop Protection Conference-Weeds,” Vol. 3, pp. 787-790. British Crop Protection Council, Surrey, UK. Wiederholt, R. J., and Stoltenberg, D. E. (1995). Cross-resistance of a large crabgrass (Digitaria sanguinalis) accessions to aryloxyphenoxypropionate and cyclohexanedione herbicides. Weed Technol. 9, 518-524. Wilkinson, M. J . , Timmons, A. M., Charters, Y., Dubbels, S., Robertson, A,, Wilson, N., Scott, S., O’Brien, E., and Lawson, H. M. (1995). Problems of risk assessment with genetically modified oilseed rape. In “Proceedings of Brighton Crop Protection Conference-Weeds,” Vol. 3, pp. 1035-1044. British Crop Protection Council, Surrey, UK. Williamson, M. (1994). Community response to transgenic plant release: Predictions from British experience of invasive plants and feral crop plants. Mol. Ecol. 3, 75-79.

HERBICIDE RESISTANCE

93

Wood Mackenzie (1995). Agrochernical Service: Crop Pesticide Sectors, May 1995. Wrubel, R . P., and Gressel, J. (1994). Are herbicide mixtures useful for delaying the rapid evolution of resistance? A case study. Weed Technol. 8,635-648. Zimdahl, R . L. (1991). Weed science: A plea for thought. USDA Symposium Preprint, pp. I 34.

This Page Intentionally Left Blank

PHYSICAL NONEQUILIBRIUM MODELING APPROACHES TO SOLUTE TRANSPORT IN SOILS Liwang Ma and H. M. Selim Department of Agronomy, Louisiana State University, Agricultural Center, Baton Rouge, Louisiana 70803

I. Introduction 11. Mobile-Immobile Two-Region Models 111. Two-Flow Domain Models IV. Capillary Bundle Models V. Multiple-Flow Domain Models VI. Coupled Physical and Chemical Nonequilibrium Models A. Freundlich Equilibrium Model B. Kinetic Two-Site Model C. Equilibrium Ion-Exchange Model D. Second-Order Mobile-Immobile Model E. Modified Second-Order Two-Site Model F. Modified Mobile-Immobile Model MI. Field Applications VIII. Summary and Conclusion Appendix: Nomenclature References

I. INTRODUCTION Soil is a porous medium with extensive chemical as well as physical heterogeneities. Chemical heterogeneity is a result of complex soil composition, whereas physical heterogeneity is often attributed to nonuniform aggregate size distribution which in turn determines a nonuniform pore size distribution. The connectiveness and size of soil pores determine ultimately the flow characteristics of water and solute in soils. Intuitively, well-connected large pores facilitate solute transport, whereas disconnected small pores retard solute transport. Be95 Mrma in ,4gmnm.y, Volume 58 Copyright 0 1997 by Academic Press, Inc. All nghts of reproduction in any form reserved

96

L. MA AND H. M. SELIM

cause the geometry of a soil system is ill-defined, it is not possible to give a comprehensive description of the pores and their connectiveness in porous soil media. A common way is to group the soil pores by some arbitrary criteria assuming that the pores in one category are connected to each other and the porosity of each category is a temporal and spatial invariant. Solute exchange between soil pores of different categories can take place in three ways: (a) convection according to the pressure gradient, (b) diffusion according to concentration gradient, and (c) mixing due to turbulence at high velocities. The necessity of accounting for physical heterogeneity during solute transport has been extensively documented in the literature for more than three decades. One of the earliest studies suggesting the importance of physical nonequilibrium is that of Biggar and Nielsen (1963). In their Mg2+ - Ca2+ exchange experiment, they found that Mg2+ breakthrough curves (BTCs) cannot be described by either an ion-exchange or a first-order adsorption model. The lack of agreement between theoretical and experimental results was explained by the differences in solute transport among pores of various sizes. Elrick et al. (1966) found that there was a delay in atrazine adsorption that could be caused by intraaggregate adsorption and dead-end pore volumes. Kay and Elrick (1967) attributed observed tailing of C1 BTCs to diffusion of C1 into stagnant pores where the flow rate is lower than the average flow rate. In an atrazine study, Green et al. (1968) explained the early breakthrough (preferential flow) by the higher than average flow velocity in large soil pores and observed tailing by the slow equilibrium of atrazine in small pores with incoming solution. In a column study with picloram, Davidson and Chang (1972) found that incomplete mixing of solution among various pore sizes resulted in early arrival of picloram in the effluent for large soil aggregate size at high flow velocities (Fig. 1). The importance of flow heterogeneity in solute transport was also inferred from the study of Davidson and McDougal (1 973). Numerous studies have been designed to characterize physical nonequilibrium of water and solute transport in soils. Direct evidence of physical nonequilibrium transport is often based on quantifying the contribution of macropores to the total water flow. In fact, only a small fraction (as low as 0.32%) of the total soil volume conducts most (73-98%) of saturated water flux (Watson and Luxmoore, 1986, Dunn and Philips 1991a, Wilson and Luxmoore, 1988, Beven and Germann, 1982). In studies with grid lysimeters, the outflow from the bottom of the soil columns was not evenly distributed among sampling cells. Some cells showed zero flow rate, whereas others had a flow rate higher than the application rate (Andreini and Steenhuis, 1990; Quisenbeny et al., 1994; Edwards et al., 1992; Shipitalo et al., 1990). BTCs from individual cells differed in residence time and peak concentration (Fig. 2), which resulted in preferential flow and tailing of the overall BTCs. Evidence of physical nonequilibrium of water flow in soils was also obtained from internal measurement of soil water pressure using

97

APPROACHES TO SOLUTE TRANSPORT IN SOILS

-

NORGE LOAM

'1

vo * 5.5 c m 82.0lWtl

h

p

1.55q/cd

VO

* 0.56 cmlhr

VO

= O.Sbcm/hr

-

Figure 1 Experimental and calculated picloram concentration distributions from Norge Loam (1.55 g/cm2 bulk density) for two average pore-water velocities and two aggregate size ranges. Solid lines are predicted using the CDE (Eq. ( I ) ) with the local equilibrium assumption [from Davidson and Chang (1972) with permission of the publisher].

small tensiometer cups (Booltink and Bouma, 1991). Among the small tensiometers randomly installed at the same soil depth, some detected water shortly after infiltration was initiated, whereas others showed changes in water content after infiltration was terminated (Fig. 3). More direct support for nonuniform flow

nAC

".-..#

0.40

0

0.35 0.30

CELL 25 - BROMIDE CEU50-BROMlDE CELL25-BLUEDYE CELL 50 - BLUE DYE

0.25 0.20 0.15 0.10 0.05 0.00 0.2

0.4

0.6

0.8

1.0

1.2

4

W R E VOLUMES

Figure 2 Bromide and blue dye breakthrough curves for cells 25 and 50 of an undisturbed soil column packed with Rhinebeck variant fine sandy loam under conventional tillage. Cell 25 has an average flow rate of 5.2 cm/day and cell 50 has an average flow rate of 7.2 cm/day [from Andreini and Steenhuis (1990) with permission).

98

L. MA AND H. M. SELIM Pressure head (kPa) 00

-4.0

-8.0

-12.0

-18.0

-20 0

0.0

- b

65 mm depth .w:.

--

- xA'.*.*.*.*.: - *. ., .- - -

A

Rgure 3 Tensiometric measurements of soil-water pressure heads for an undisturbed soil column at depths 25 (top), 65 (middle), and 125 mm (bottom) below the soil surface. Tl-T7 are tensiometers randomly distributed at a depth of 25 mm. T8-TI4 are tensiometers randomly distributed at a depth of 65 mm. T15-T21 are tensiometers randomly distributed at a depth of 125 mm [from Booltink and Bouma (1991) with permission].

distribution is the high peak concentration of BTCs (see Fig. 4) measured at lower soil depth as observed by Yasuda er al. (1994). Field and laboratory observations that support physical nonequilibrium include

APPROACHES TO SOLUTE TRANSPORT IN SOILS

99

Pressure head (kPa) I

00

-4 0

-8 0

-12 0

-160

-...* -200

--./.

I

,.-.1

-24 0

1

.

!;

,

1

1

l

l

l

1

1

,

,

1

,

,

-

Figure 3 (continued)

the bimodal peaks of nonreactive solutes when steady soil water flow was dominant (Hornberger et al., 1990; Hamlen and Kachanoski, 1992; Ma and Selim, 1994a; Ward et al., 1994). Hornberger et af. (1990) observed multiple peak BTCs of Br from subsurface drainage at a hillslope under constant water flow velocities (Fig. 5). The multiple peaks disappeared at high flow velocity when contribution of large pores was dominant. In a study with saturated soil columns, Ma and Selim (1994a) observed distinct double-peak BTCs for several soils when tritium was applied as an impulse for only a few minutes (Fig. 6). However, contrary to previous findings (Hornberger et af., 1990), Ma and Selim (1994a) found that the double peaks became more obvious with increasing flow velocity. Double-peak BTCs were reported in unsaturated and undisturbed soil columns by Ward et al. (1994) when C1 was applied for a period of 32 s (Fig. 7). Hamlen and Kachanoski (1992) also observed bimodal C1 BTCs in the B horizon of a sandy soil but not in the A horizon or A + B horizon. Such findings illustrate the significance of the connectiveness of macropores across soil layers (or boundary interface) on observed BTCs. Skopp er al. (1981) proposed a two-flow domain model by assuming the presence of distinct dual porosities in the soil in order to explain solute bimodal behavior in soils. Ma and Selim (1995) successfully applied this two-flow domain model to explain the observed bimodal tritium BTCs. Using time series analysis, Hornberger et al. (1990) found that soil porosity may be divided into two parallel mixing volumes. For a water

Plot 2

Plot 1 0.20

0.6

(OJmrI Deplh Obs. Cal.

0.4

0.3 h

p

CDE

CDE

h

0.5

I

I

\

0.5 m

- 1I

0

--

10.8m A

0.2

0

v

c 0.1 0

x o

e

1

2

3

4

5

6

"0

1

2

3

4

5

6

7

6

7

0.20 NECDE

"0

1

2

3

4

5

Time (hr) figure 4 Observed BTCs from ceramic soil water samplers at different soil depths compared with model results using the CDE (Eq. (1)) and the mobile-immobile model [from Yasuda et al. (1994)with permission].

APPROACHES TO SOLUTE TRANSPORT IN SOILS

101

application rate of 10 cm/h, they obtained one mixing volume with a mixing time of 64 min and the other with a mixing time of 10 min. The characteristics and mathematical models of physical nonequilibrium have been reviewed by several authors (McCoy et al., 1994; Brusseau, 1994; Sardin et al., 1991; Brusseau and Rao, 1990; Jury and Fliihler, 1992; Bouma, 1991; Selim, 1992). The reviews of McCoy et al. (1994) and Bouma (1991) emphasized the characterization of macropore flow and measurement of macroporosity. Brusseau ( 1994) reviewed several field-scale experiments and factors controlling reactive solute transport in porous media. Sardin et al. (1991) provided a thorough discussion on the mobile-immobile two-region model using time-moment analyses and transfer function approaches. Brusseau and Rao (1990) presented a brief review on the mobile-immobile model as well as its coupling to chemical nonequilibrium models. Selim ( 1992) reviewed the application of mobile-immobile models for inorganic solute transport. The review of Jury and Fliihler (1992) covered the mobile-immobile model approach and preferential flow in a more broad context. McCoy et al. (1994) classified frequently used models into “Richards”- and “non-Richards”-type models. The Richards-type model is simply based on the Richards flow equation, Darcy’s equations, and the assumption of a uniform flow domain. The non-Richards-type model makes no assumptions regarding flow domains. The latter, including probability density function approaches and time series analysis, has been reviewed by McCoy et al. (1994). This chapter deals primarily with Richards-type models. Emphasis is given here for physical nonequilibrium behavior in an attempt to provide the reader with a comprehensive understanding of existing models. For each model discussed, experimental justifications, model assumptions, model development, and parameter estimation are provided.

11. MOBILE-IMMOBILE TWO-REGION MODELS In Richards models, soils are commonly treated as uniform porous media and solute transport under steady-state flow and uniform water content is assumed to obey the classical convective-dispersive equation (CDE): 0-ac = D O -a2c - ~ 0ac- - p-as at ax2 ax at ’ where C represents solute concentration in soil solution (pg/ml), S is the amount of solute sorbed or retained (pg/g soil), 0 is the soil-water content (cm3 cm-9, D is the hydrodynamic dispersion coefficient (cm2 h-I), u is the average pore water velocity (cm/h), p is the soil bulk density (g cm-3), x is the spatial coordinate (cm) parallel to the flow direction, and t is time (h).

102

L. MA AND H. M. SELIM a

0

0

2

4

e

8

I0

12

14

nmo from tracer application, hour8

Time from tracer rppllcatlon, hour8

Figure 5 Bromide breakthrough from a hillslope drainage study as affected by water application rates of (a) 2.5, (b) 5, and (c) 10 cm/h. Solid lines are fitted with a time-series model [from Homberger er al. (1990) with permission].

APPROACHES TO SOLUTE TRANSPORT IN SOILS

103

C

BE

Time from tracer application, hours

Blgure 5 (continued)

The CDE given in Eq. ( I ) can only provide unimodal solute distribution for solute transport and cannot explain the bimodal behavior of BTCs (Ma and Selim, 1995). The inadequacy of Eq. ( I ) lies in the assumption of uniform pore size distribution and uniform flow field. However, due to the complexity of pore size distribution or flow field, extensive simplifications are needed for the purpose of development of mathematical approaches. One of the simplest assumptions is to divide the soil pores into two categories. The first category consists of large soil pores in which convection takes place and contributes to water flow in soils. The second category is composed of the small pores in which no water flow is assumed. Water in the small pores is thus stagnant or immobile. Solute exchange between the two water phases is commonly assumed to be a first-order diffusion process according to a concentration gradient. One of the first applications of the mobile-immobile concept can be traced back to Deans (1963) who extended a one-parameter mixing cell model to a three-parameter mixing cell model in which a fraction of water was assumed stagnant and behaved as a capacitance. He found that the three-parameter model was capable of explaining the asymmetry and tailing of solute transport. The three-parameter mixing model can be formulated as

L. MA AND H. M. SELIM

104 0.03

.... ..

C.ON ((2mm)

v = 1.07 cm h-l

.

0.02

A

Short Pulae 6°C

Exp. Data

0.01

0.00

10

0.03 CKll ((2mm)

30

20

- Short Pulse 6°C

B v

0 0

-

0.02

2.23 cm h-'

Exp. Data

'3. 0.01

0.00

. - -

15

10

0.04

CKil ((2mm)

.* 0.03

s

0*02

.

v

.. . .

0

- Small Pulae BTC *

-

C 5.21 cm h-l

Exp. Data

..**.

0.01 tm

-

3.97 h

0.00 2

4

6

8

0

Time (h)

Figure 6 Breakthrough curves from small tritium pulses in Cecil soil at different pore water velocities of (A) 1.07, (B) 2.23, and (C) 5.21 cm/h. t, is the mean residence time [from Ma and Selim (1994b) with permission].

where C, and Ci, are solute concentrations in the mobile and immobile water phases (pg/ml), respectively. The three parameters are as follows: a,the first-

APPROACHES TO SOLUTE TRANSPORT IN SOILS

-

10s

- J,=l - . -1.41 - - . -- - . J cm day-' : e =0.29 m3m.3 :

7

0

4

0

12

16

20

24

Time [days]

Figure 7 Chloride BTCs at a depth of 1. I m for three different fluxes and water content during one-dimension transport in an undisturbed soil column as measured by time domain reflectometry. C1 was applied for only 32 s [from Ward el a / . (1994) with permission].

order mass transfer coefficient (h-I); F, the mobile phase fraction ( F = 8,/0); and u , the mean pore water velocity (cm/h). 8, is the mobile water content (cm3 cm-3). The above model was later extended by Coats and Smith (1964) to include the hydrodynamic dispersion coefficient D (cm2 h-1). The model of Coats and Smith has been used extensively in the literature since then and can be expressed as

(1 - F ) & at

= a(C,

- Cim).

(5)

106

L. MA AND H. M. SELIM

This concept was applied to aggregated media by Passioura (1971) who viewed the mobile phase as interaggregate pores and the immobile phase as intraaggregate pores. However, this approach was not fully tested in soil science until 1976 when van Genuchten and Wierenga (1976) published their paper on mass transfer studies in sorbing porous media. Figure 8 shows a diagram of their simplified soil system that consists of the following five regions; 1. Air spaces for unsaturated soil systems. 2. Mobile water located inside the large (interaggregate) pores. Water flow is assumed to occur in this region only. Solute transfer in this region occurs by both convection and longitudinal diffusion. 3. Immobile water located inside the small (intraaggregate) pores and at the contact points of aggregates. Solute exchange between mobile and immobile waters is by diffusion only. 4. A dynamic soil region located sufficiently close to the mobile water phase. Solute adsorbed by this portion of soil is in equilibrium with that in the mobile water phase. 5 . A stagnant soil region that is in direct contact with the immobile water phase. Solute adsorption in this region is limited by solute diffusion from the mobile water phase into the immobile water phase. By introducing a parameterfas the fraction of retention or sorption sites that is in direct contact with the mobile water phase (fraction of the dynamic soil region), the CDE can be modified as

where D, is the hydrodynamic dispersion coefficient in the mobile phase (cm2 h-1) and Y , is the flow velocity in the mobile phase (cm h-1). S, and Si, are the amounts of solute adsorbed from the mobile and immobile water phases (pg/g soil), respectively. This model was expanded to three-dimensional flow by Goltz and Roberts (1986a, 1988). Analytical solutions to Eqs. (6) and (7) under different initial and boundary conditions can be found in van Genuchten and Wierenga ( 1976) and De Smedt and Wierenga (1979). Toride et al. (1993) provided a set of analytical solutions for the two-region model with first-order degradation and zero-order production for semi-infinite soil systems. The mobile-immobile model was also analyzed with the method of moments by Sardin et al. (1991) and Valocchi (1990). The mobile-immobile concept was capable of explaining the early breakthrough and extensive tailing of tritium BTCs in porous media (van Genuchten

APPROACHES TO SOLUTE TRANSPORT IN SOILS

I07

B DYN

2 Figure 8 A schematic diagram of an unsaturated aggregated porous medium. (A) Actual model; ( B ) simplified model. The shading patterns in A and B represent the same region [from van Genuchten and Wierenga (1976) with permission].

and Wierenga, 1977). Yasuda et a / . (1994) found that the mobile-immobile approach best described Br BTCs at different soil depths (Fig. 4). van Genuchten et al. (1977) also found that the mobile-immobile approach provided a better description of 2,4,5-T BTC in soils when the local (chemical) equilibrium assumption (LEA) was used (see Fig. 9). Li and Ghodrati (1994, 1995) compared the mobile-immobile model with the classical CDE and a stochastic model in soils with earthworm holes or root channels and concluded that, although the mobile-immobile model provided the best fit to their experimental data, the tworegion model does not describe the complexity of solute behavior in porous systems. Under steady-state water flow and linear adsorption, the two-region model can be expressed in the same dimensionless form as the two-site chemical nonequilibrium model (Toride et al., 1993; Nkedi-Kizza et al., 1984; van Genuchten and Wagenet, 1989). The mathematical equivalency between physically and chemically based nonequilibrium transport models increases the difficulty in identifying transport mechanisms prevailing during solute transport. Unless model parameters can be independently obtained (other than by use of curve fitting), it is not possible to derive solute transport mechanisms in soils based on observed experimental phenomena. One major assumption of the mobile-immobile model is the uniform solute distribution in each water phase. Solute transfer between the two water phases is

L. MA AND H. M. SELIM

108 1.OL

-

-

Exp. 1 4 2,4,5 T

SINGLE VALUE

E .0-

0

2

0

SINGLE VALUED NON SINGLE VALUED

-

-

-6I - .

z W

0' 0

q=S.ll

cddsy

A-

0

r!j . '2-

W

u

0. 0

.

1

3

2

4

5

6

7

figure 9 Observed and calculated 2.4.5-TBTC in a Glendale clay loam soil column. Solid line is calculated from the mobile-immobile model and dashed lines are based on the CDE (Eq. ( I ) ) with a single valued adsorption constant or nonsingle valued adsorption constant [from van Genuchten er al. (1977) with permission].

assumed to follow an empirical first-order diffusion model. An alternative to this approach is the sheet film diffusion models. Gottschlich (1963) viewed the immobile water phase as a layer of uniform film over the soil particles. Solute inside the film is not uniform and is governed by a diffusion equation such as

where the boundary conditions for the immobile phase are Ci,

=

C,

y

=

d

where d is the water film thickness (cm), y is the normal distance to the convective flow measured from the pore wall (cm), and Do is the molecular diffusion coefficient (cm2 h-1). A similar approach was used by Skopp and Warrick (1974) in their mobile-immobile approach (Fig. 10). They assumed that only convection takes place in the mobile zone. The continuity equation was written as

APPROACHES TO SOLUTE TRANSPORT IN SOILS

109

Figure 10 Idealized How regime in porous media. ( A ) Model of flow in soils: ( B ) simplitied model. analogous to A . The shading patterns in A and B correspond [from Skopp and Warrick ( 1974) with penni\sion]

where D,,,, is the diffusion coefficient in the immobile water phase, d is the thickness of the immobile water. and H is the thickness of mobile water. Coordinates .r and y represent distances parallel and normal to the convective flow, respectively. The dift'usion equation for the immobile phase is the same as that in Eq. (8). However, the boundary conditions at the mobile-immobile interface are different from that of Gottschlich (1963) and can be written as

c,,,,= PC,,

Y =

where p is a partition coefficient. Boundary conditions given by Eq. ( I I ) were also used by Pignatello et d.(1993) in an atrazine study. Similarly, Rao ef d. ( 1980a) assumed a spherical aggregate geometry. Water inside the spherical aggregates was viewed as immobile water and solute distribution inside the sphere was not uniform. Solute diffusion into the aggregates was governed by

110

L. MA AND H. M. SELIM

Fick’s second law, and the diffusion model can be expressed in spherical coordinates as

where D,(= D , T ~ )is the effective molecular diffusion coefficient, T~ is the tortuosity factor (
The boundary and initial conditions for the immobile phase are C,(t)

t

z0

Cim(r,t ) = Cy,

0

Ir Ia,

Ci,(a, t )

=

(14) t = 0,

where Cp, is solute concentration in a sphere at t = 0 and a is sphere radius. Gaston and Locke (1995) found that both sheet and spherical diffusion models offered similar predictions of fluometuron transport in a silty clay soil. For nonreactive solute transport in spherical aggregates at t > 0.3a2/De, Passioura (1971) approximated the concentration difference [C, - Cim]by

c,

- ci,

=

va2 aC, 15D, ax

’

Passioura (197 1) further derived an effective solute dispersion coefficient D equivalent to that in Eq. (1) for spherical aggregates as D = D,F

F)aW + (1 -15D, ’

with a limiting condition of

where L is solute transport length. As evidenced by Eq. (16), when physical nonequilibrium exists, the overall effective dispersion coefficient D increases (Skopp and Gardner, 1992; Li and Ghodrati, 1995). As a result, the fitted overall D is usually large for models in which flow field is not well presented (Oostrom et al., 1992; De Smedt et al., 1986; De Smedt and Wierenga, 1984).

APPROACHES TO SOLUTE TRANSPORT IN SOILS

111

De Smedt et al. (1986) obtained large dispersivity values under unsaturated water flow when Eq. (1) was applied. The increase in dispersivity at unsaturated water content is due to flow heterogeneity and diffusion between mobile and immobile water phases. However, the dispersivity decreased 10-fold when using a two-region mobile-immobile model. Ma and Selim (1995) also demonstrated that the dispersivity obtained from a one-flow domain model (Eq. (1)) is one order of magnitude larger than that from a two-flow domain model. Koch and Fliihler ( 1 993) found that fitted D is much higher in porous glass beads than that in solid beads. Equation (16) was extended to include a reactive solute with a retardation factor R by van Genuchten and Dalton (1986) and Parker and Valocchi (1986). D

=

D,F

+ (1

- F)a2v2Ri?,

15DP2

’

where R i m is the retardation factor of the immobile phase, and R is an overall retardation factor (OR = Bi,Ri, + O,R,). A similar effective dispersion coefficient was obtained for rectangular aggregates with a half width of a,: D

=

D,F

+ (1

- F)afu2R;,

3 D 32

For solid cylindrical diffusion aggregates with a radius of a,, D = D,F

+ (1

- F)u:u~R;, 8D32

f

and for hollow cylindrical aggregates with a macropore radius of up and a soil cylinder radius of b (cylinder of soil surrounding the macropore), D = D,F

1](1 - F)uwR;,,, + [21n(b/ap) -4D,R2

An overall D can also be derived from the empirical first-order mass transfer equation (Eq. 7) in which uniform solute distribution in the immobile water phase is assumed. De Smedt and Wierenga (1984) derived an effective dispersion coefficient D for nonreactive solutes for long columns as

and for reactive solutes (van Genuchten and Dalton 1986) as D = D,F

F)~v~R;, + e(i -ciR2 ,

112

L. MA AND H. M. SELLM

Comparing Eqs. (16) and (22), an equivalent first-order transfer coefficient (a)for spherical aggregates is obtained: a=

15De(1 - F)O a*

This equation can also be obtained through moment analysis (Valocchi, 1990) and has been used widely in solute transport (Selim et al., 1987; Selim and Amacher, 1988). Similar a values were obtained for rectangular aggregates: a=

3D,(1 - F)O , a:

for solid cylindrical aggregates: a=

sn,(i - F ) e a,'

f

and for hollow cylindrical aggregates: a=

4 ~ , ( i- F ) e b*[21n(blaP) - 11 '

A more general form of Eqs. (24)-(27) is a =

n(l - F)OD, , a2

where n is a geometry factor and a, is an average effective diffusion length. Further details can be found in van Genuchten and Dalton (1986) and van Genuchten (1985). Equation (28) has been widely used to estimate a in modeling solute transport in porous media (Selim and Amacher, 1988; Goltz and Roberts, 1988). Another way of estimating a from aggregate geometry is that of Rao et al. (1980a,b) who found that a is time dependent. They derived a time-averaging a for spherical aggregates: a =

[ DeOim ---J a*.

a* can be estimated from the following equations:

T* > 0.1 ,

(31)

APPROACHES TO SOLUTE TRANSPORT IN SOILS

113

where b

=

($1

0.14472 In -

and

T T* = D 2 a*

(33)

The q , depends on F and was listed in Table 1 of Rao et af. (1980a) and T is the time over which a average is taken. As suggested by Rao et al. (1980b), the mean residence time T (=L/u,) provided a good estimation of a.The a values for cubic aggregates were also derived by Rao et af. (1982) using an equivalent spherical radius such as a = 0.62031, where 1 is the length of the side of a cubic. Such an approach was utilized by Selim and Ma (1995) to describe atrazine transport in an aggregated clay soil. However, estimated a values based on the aggregate geometry do not always provide adequate prediction of BTCs. Moreover, a is equally affected by experimental conditions. Additional adjustment of (Y to individual experimental conditions is usually made by fitting the mobile-immobile model to experimental data. Fitted (Y values increase with flow velocity (De Smedt and Wierenga, 1984; De Smedt et af., 1986; van Genuchten and Wierenga, 1977; Kookana et al., 1993; Li and Ghodrati, 1994). This velocity dependence is perhaps reasonable because turbulent mixing may occur at high flow velocities. In a laboratory study with glass beads under both saturated and unsaturated conditions, De Smedt and Wierenga (1984) found that a (day-') is linearly related to the mobile phase velocity ,Y (cm/day) as a = 0 . 0 4 2 ~+ ~2.2. They also obtained a = 0.02 v, in a later study with sand under unsaturated water conditions (De Smedt er al., 1986). Under unsaturated water conditions, convective water transfer between flow domains significantly contributes to solute transfer and results in large a values (Gerke and van Genuchten, 1993; Jarvis et af., 1991a). Steenhuis et al. (1990) assumed that solute transfer between flow domains is mainly through convection transfer of water. Thus, when convective mass transfer exists, the empirical first-order diffusion transfer equation (Eq. (7)) may not be sufficient. In addition to flow velocity, pore connectiveness also affects a values. Greater pore continuity results in smaller a's (Skopp and Gardner, 1992). A major difficulty in applying the two-region model is the definition of the mobile and immobile water phases. A common way of differentiating the two phases is through curve fitting (Li and Ghodrati, 1994; van Genuchten and Wierenga, 1977). De Smedt and Wierenga (1984) found 0, = 0.8530 is unique for unsaturated glass beads with diameters in the neighborhood of 100 Fm. However, the definition of macropores and micropores is vague and somewhat

114

L. MA AND H. M. SELIM

arbitrary (Beven and Germann, 1982; Smettem and Kirkby, 1990). An experimental method to estimate the two porosities is to measure soil water content at some arbitrary water tension (+). Smettem and Kirkby (1990) used water content at a $ of 14 cm as the matching point between the interaggregate (macro-) and the intraaggregate (micro-) porosity by examining the - 8 soil moisture characteristic curve (Fig. 1 I). J m i s et al. (1991b) estimated macroporosity from specific yield under a water tension of 100 cm. Other water tensions used to differentiate macropores from micropores are 3 cm (Luxmoore, 1981), 10 cm (Wilson et al., 1992), 20 cm (Selim er al., 1987), and 80 cm (Nkedi-Kizza et al., 1982). A list of water tensions used by different authors was provided by Chen and Wagenet (1992). The equivalent diameters at these water tensions range from 10 to 10000 pm based on capillary flow. Other methods used to estimate visible macropores from fauna activity include dye tracing (Trojan and Linden, 1992; Booltink and Bouma, 1991), dental plaster casting (Wang et al., 1994), resin impregnation (Singh et al., 1991), macropore tracing on clear plastic sheets (Ela et al., 1992; Logsdon et al., 1990), X-ray computed tomography or CT-scan (Warner et al., 1989), and in sifu photography of soil profile (Edwards et al., 1988). The macroporosity estimated using the previous methods ignores the continuity of macropores as well as their distribution with soil depth and may not be suitable for transport modeling (Chen et al., 1993; Munyankusi et al., 1994; McCoy et al., 1994). Another experimental measurement of 8, is based on the following mass balance equation:

+

ec = e,cm+ eimcim.

(34)

When a is small enough to assume Ci, = 0 and C , = Co (input concentration)at certain infiltration times, an approximate equation is obtained:

Applications of this method can be found in Clothier et al. (1992) and Jaynes et al. (1995). By assuming that the tracer concentration in the mobile water phase (C,) equals the input concentration (CJ, Jaynes et al. (1995) derived the following formula from Eq. (3) and (34):

where a and Oim can be estimated by plotting ln(1 - C/C,) versus application (infiltration) time. However, the assumption of C, = C , associated with this method is questionable and may not be correct as long as a # 0. Slightly different from the approach of Jaynes et al. (1995) and Clothier et al. (1992),

APPROACHES TO SOLUTE TRANSPORT IN SOILS

1.0

-

0.1

-

115

-wm(m)

I

0

I

I

I

I

I

0.1

0.2

0.3

0.4

0.5

l i 0.6

e (m3m73 Experimental data from TemFigure 11 Experimental wetting soil moisture characteristic. (0) pe pressure cell; ( 0 )experimental data from constant pressure absorption; solid and dashed lines are bimodal model predictions using match points of 14 cm water tension [from Smettem and Kirkby (1990) with permission].

Goltz and Roberts (1988) estimated the fraction of mobile water as the ratio of velocity calculated from hydraulic conductivity to the velocity measured from tracer experiments. Chen et af. (1993) estimated functional macropores by monitoring water content in the soil during water infiltration. They assumed that drainage from a saturated soil took place in three stages: initial drainage from macropores only, middle drainage from both macropore and micropores, and final drainage from micropores only. The demarcation lines between the three stages were based on some calculated characteristic time. The initial drainage was used to estimate macroporosity and macrohydraulic conductivity. Another alternative way to study macropore continuity is to measure air permeability at certain water tension (Roseberg and McCoy, 1990).Nevertheless, the determination of 8, is arbitrary. In fact, the concepts of macropores and micropores are relative and change with experimental conditions such as flow velocity (van Genuchten and Wierenga,

116

L. MA AND H. M. SELIM

1977; Nkedi-Kizza et al., 1983; Watson and Luxmoore, 1986; Li and Ghodrati, 1994). Nkedi-Kizza efal. ( 1 983) found a decrease in F with flow velocity. On the other hand, van Genuchten and Wierenga (1977) found an increase in F with flow velocity. Li and Ghodrati (1994) observed no consistent relationship between F and flow velocity in a NO3 study. This inconsistency may be explained by differences in soil type, flow velocity range, and applicability of the mobileimmobile approach.

111. TWO-FLOW DOMAIN MODELS A well-known approach for physical nonequilibrium in soils is the two-flow domain model in which water was assumed to flow through two different-sized pores with distinct velocities. The porosity of each flow domain was treated as constant in time and space (Chen and Wagenet, 1992; Jarvis e f al., 1991a). One of the earliest studies of this concept is that of Skopp et al. (1981). The twodomain model divides soil water into two regions based on their flow velocities. Both water regions have a nonzero flow rate (Fig. 12). Without loss of generality, we denoted the fast flow region as A and the slow flow region as B. The soil system was characterized by velocity (vA, uB), water content (eA,eB),solute concentration (CA, CB), and dispersion coefficient (DA,DB) (Skopp et al., 1981). The two domains are related by an interaction term r such that

r = a(cA- cB).

(37)

Figure 12 Idealized geometry of a two-flow domain model [from Skopp et a / . (1981) with permission].

APPROACHES TO SOLUTE TRANSPORT IN SOILS

117

This model reduces to a capillary bundle model when r or (Y is negligible (Rao et a / . , 1976; Steenhuis et a / . , 1990; Skopp et al., 1981), and it approaches the classical one-domain CDE as r increases. The two-flow domain model can also be reduced to a mobile-immobile model when w, equals zero. The convectivedispersion equation in the two domains can be written as (Skopp et a f . 1981)

where 01 is now a first-order transfer coefficient between the two water flow domains (h- I ) . Solute concentration at the inlet was assumed to be the same for both fast and slow flow domains. This model was capable of describing observed double-peak BTCs (Ma and Selim, 1995). Figure 13 shows fitted tritium BTCs for an input pulse of one pore volume with the two-flow domain model in comparison to the one-flow domain model. Figure 14 is their corresponding small-pulse BTCs (about 0.03 pore vol) predicted with parameters from the large-pulse BTCs. Thus, the two-flow domain model successfully explained the double-peak behavior of tritium in the soil and the differences in BTC shape for small and large input pulses. Ma and Selim (1995) further showed that the BTC shape was very much affected by relative flow velocity (y = w,/w,) (Fig. 15) and the relative size of the two flow domains (F = 0 A / 0 ) (Fig. 16). Skopp et a / . (1981) further compared two different ways to obtain the firstorder solute transfer coefficient CI for the two-flow domain model. One was based on Eq. (28), and the other was derived from Duguid and Lee (1977) for steadystate flow:

where g is the acceleration due to gravity (cm h-z), a is the aggregate size (crn), 5 is the interaggregate pore size (cm), and K , is the conductivity of the soil matrix (cm h-1). This equation assumed that mass transfer is a function of pressure gradient between the two flow domains. Applying these two methods to the data of Anderson and Bouma ( 1977), Skopp e t a / . (1981) found that 01 estimated from the Duguid and Lee’s model (Eq. (40))was two orders of magnitude smaller than that from Eq. (28). The two-flow domain model concept has been used in the recent literature for unsaturated flow conditions when large macropores are dominant. Othmer et al. (1991) and Smettem and Kirkby (1990) obtained an improved description of

I18

L. MA AND H. M. SELIM 0.8

0.0

Mahan I

- Large Pulae

a

Mahon II

- Large Pulse

b

0.6 - -

s

v

-

3.82 cm hr-l

0

Exp. Data -- one domoln two domoln

0.4--

~

0.2-

o.o-+r 1

0

0.8

Mahan 111

2

- Large Pulae

C v = 5.28 cm hr-l

,-

E X ~ Dota .

- - - one domain - two domain

0.4

0.2

1

2

3

Pore Volume (V/Vo) Figure 13 Large-pulse BTCs from three Mahan soil columns with pulse input of one pore volume. Solid and dashed curves are fitted BTCs with the two- and one-flow domain models [from Ma and Selim (1995) with permission].

soil-moisture characteristic curves using the dual-porosity concept (Fig. 1 1). Gerke and van Genuchten (1993) modeled solute transport under unsaturated flow conditions using the two-flow domain concept. The Richards flow equation was used to simulate water flow and the CDE was applied to solute transport in both flow domains. Water transfer between the two domains was assumed to be

APPROACHES TO SOLUTE TRANSPORT IN SOILS 0.04

Mahan I - Small Pulse Predictlons

I19

a

v = 2.02 cm hr-'

0.03

Exp. Data

.--.one domain

- two domain

1

2

0.03

v

3

= 3.82 cm hr-'

8

Exp. Data one domaln - two domain

3 0 0.04-

0.03

i

2

Mahan Ill - Small Pulse Predictions

..

v

4

3

C

5.28 cm hr-'

Exp. Data

.... one domain

- two domain

0

1

2 3 Pore Volume (V/Vo)

4

Figure 14 Small-pulse BTCs from three Mahan soil columns with pulse input of 0.03 pore volume. Solid and dashed curves are predicted BTCs with the two- and one-flow domain models using model parameters derived from Fig. 17 [from Ma and Selim (1995) with permission].

first-order linear diffusion and solute transfer was controlled by both diffusion and convection mechanisms. The two-flow domain models of Jarvis et ul. (1991a) and Chen and Wagenet ( 1992) differed from that of Gerke and van Genuchten (1993) mainly in their approaches to macropore flow. Jarvis er ul. (1991a) evaluated macropore water

L. MA AND H. M. SELIM

120 0.15.-

a

L = 15cm pv = 0.1 8 = 0.47 crn ern-' F = 0.48 a = 0.03 hr-' A = 0.54crn -1 v = 5.21 crn hr

0.100

0

2

- y y -y '-

..-.. 0.05

--

..... y

0.007-

0.8 --

1

2

b

-

\

't. 0.40.2-CLOT.

0

. . .\ . I,.'

...,

'..

/:\ :; ! .:.:.I

.:

'. ....

. .,' I.,/:/.

, ,I

1

2

10.0

4

- y = 2.0 4.0 3.0

v.. . .

.:;I

=

L = 15crn pv = 1.0 8 = 0.47 cm crn-l F = 0.48 a = 0.03 hr-' A = 0.54crn -l v = 5.21 crn hr

;

0 0.6.0

2.0

= 3.0 = 4.0

3

.. .. /

=

\,

.(.(.7 = 10.0

*. .

...... :T,'.,.!......

..

\.,

....

3

.+..-. 4

i

Pore Volume (V/Vo)

Figure 15 Simulated BTCs for different relative flow velocities (V,/V,) and for an input pulse of 0.1 (a) and I .O (b) pore volume with the two-flow domain model [from Ma and Selim (1995) with permission].

flow using the unit hydraulic gradient assumption and solute transport was by convection only. Chen and Wagenet (1992) treated macropore flow as tube flow and applied the Hagen-Poiseuille equation for laminar flow and the ChezyManning equation for turbulent flow. Logsdon (1995) experimentally proved the existence of turbulent flow in macropores under ponded conditions. Although water and solute transfer between flow domains are commonly treated as a firstorder linear function of pressure or concentration gradients, nonlinear approaches can be found in the literature (see Zimmerman et af.,1993; Dykhuizen, 1990). Water transfer was also described by Philip's infiltration (sorptivity) equation (Chen and Wagenet, 1992; Ahuja et al., 1993). Similar to the mobile-immobile approach, it is not possible to have general and definitive criteria for differentiating the two-flow domains. Another difficulty in applying the two-flow domain model is the determination of water flow veloc-

APPROACHES TO SOLUTE TRANSPORT IN SOILS

12 1

ity in each flow domain, even under a saturated soil system. One way to determine the flow velocities is to use the Hagen-Poiseuille equation for laminar flow and the Chezy-Manning equation for turbulent flow (Lindstrom and Boersma, 1971; Chen and Wagenet, 1992; Logsdon, 1995). Another way may be to estimate the hydraulic conductivity (Steenhuis et al., 1990; Jarvis et al., 1991a; Scotter and Ross, 1994). Ma and Selim (1995) obtained “best-fit’’ velocities for the two domains (u, and u,) based on an average pore flux u (0 u = 0AuA 0 , ~ ~ In ) .addition, the dispersion coefficients for each flow domain must also be provided (DA and D s ) . Because the flow domains and their flow velocities cannot be experimentally differentiated, it is not an easy task to measure D for each flow domain. Lindstrom and Boersma (1971) estimated D as an exponential function of pore radius. Rao et al. (1976) and Ma and Selim (1995) assumed D to be a linear

+

a 0.15-

L = 15crn pv = 0.1 0 = 0.47 crn crn-l y = 2.76 a = 0.03 hr-’ A = 0 54 cm v = 5:21 crn hr-l

( .

,

,:....\.. ..

0

t,

. ’,

;

:

b .

L=15crn pv = 1.0 0 = 0.47 crn crn-l

. . ..

.

h = 0.54 crn

v = 5.21 crn hr-l

0

0

3

- F = 0.2 0.4

0.2 0.0

0

1

2

3

4

Pore Volume (V/Vo)

Figure 16 Simulated BTCs for different values of the fast flow domain fraction (8,/8) and for an input pulse of 0.1 (a) and 1 .O (b) pore volume with the two-Row domain model [from Ma and Selim (1995) with permission].

122

L. MA AND H. M. SELLM

IV. CAPILLARY BUNDLE MODELS Another category of the physical nonequilibrium models is based on the capillary bundle concept. In this type of model, solute transport is assumed to pass through capillary channels without transverse solute exchange between channels. The CDE was applied to each flow domain independently. One of the earliest capillary bundle models can be found in the chemical engineering literature (’hmer, 1958). For nonreactive solutes, the transport equation of each channel can be expressed as (Lindstrom and Boersma, 1971)

where Dj = Do(l -

where 5 is a diffusion coefficient parameter, rj is the average jth pore radius, 6 is the hydrostatic pressure gradient, and I.L is the coefficient of viscosity. For a small average pore size diameter rj,

Dj = D,C;rj” (43) The average distribution of solute along a soil column can be expressed as N

C

Aj

i= 1

where Aj is the weight coefficient of the jth partition fraction. This model is capable of simulating irregular BTCs including multiple peaks, early breakthrough, and tailing (Fig. 17). However, when the pore size distribution was based on soil moisture characteristic curves, Rao et al. (1976) found that the capillary bundle model offered little success in describing tritium transport data from two Hawaiian soils (Fig. 18). Rather than the exponential Eq. (53) of Lindstrom and Boersma (1971), Rao et al. (1976) used the following dispersion coefficient Dj for each channel: Dj = (Do

+ Avj)7*.

(45 )

APPROACHES TO SOLUTE TRANSPORT IN SOILS

123

Figure 17 Distribution of chemical in water-saturated, porous media with three pore sizes present in each medium. The model parameters are: (1V) A,’s = 0.25,0.5,0.25; v,’s = 2.0,4.0,6.0 cm/ day: andD,’s = 0.5.0.7. I.0cm2/day: (V)A,’s = 0.60,0.20,0.20; v,’s = 2.0,4.0.6.0cm/day;and D,’s = 0.5,0.7, I.0cm2/day: (VI)A,’s = 0.20,0.20,0.60; vj’s = 2.0,4.0, 6.0cm/day;andDJ’s = 0.5, 0.7, 1.0 cmz/day [from Lindstrom and Boersma (1971) with permission]. 0

I

1

-

MOLOKAI

Trit lated W a t e r

0 ,C-8

Model

0

$3

0

v I v, Figure 18 Comparison of measured (data points) and predicted (solid lines) BTCs for displacement of a tritium water pulse through a Molokai subsoil column with a capillary bundle model (C-B model) and the classical CDE (Avg. velocity) [from Rao CI nl. (1976) with permission].

124

L. MA AND H. M. SELIM

They found that an average pore velocity improved the description of tritium transport in soils versus the capillary bundle model (Fig. 18). The failure of the capillary bundle model may be attributed to a lack of knowledge of the complexity of the pore geometry, inaccurate estimation of model parameters, and lack of interexchange among channels. It is well accepted that pore geometries do not resemble a fully parallel and continuous system of channels that are connected along the flow direction. In fact, the pores in each category are not totally connected. Solute transport paths consist of pores of different sizes. Therefore, it is important to include interexchange between flow regions even though there may not be concentration and pressure gradients.

V. MULTIPLE-FLOW DOMAIN MODELS Because soils consist of a continuous pore size distribution, it is more conceptually sound to view the pore geometry as multiple-flow domains. The simplest multidomain model is the three-domain model proposed by Luxmoore (198 1) in which the soil pores were arbitrarily classified into macro- (>1000 Fm), meso(10- 1000 pm), and micropores (<10 Fm). These pore size ranges correspond to water tensions of <3, 3-300, and >300 cm. However, Wilson et al. (1992) found that water tension of 10 cm is a better demarcation between macro- and mesopores. These definitions ignored pore continuity that cannot be measured under static water conditions. Because the contribution of pores to water and solute flows highly depend on experimental conditions (such as antecedent water content, flow velocity, and solute application method), it is necessary to define an effective pore volume relating to flow conditions. One of the efforts in that direction is that of Steenhuis et al. (1990), who differentiated flow paths from measured hydraulic conductivity at different water contents using a piecewise linear approximation method. One of the successful applications of the three pore-region model is the hydraulic property study of Wilson et al. (1992). They classified the soil pores according to water tensions of <10 (macro-), 10-250 (meso-), and >250 cm (micropores). The soil moisture characteristic curve for the macropores was described by the Fermi function and the van Genuchten (1980) equation model was used for the meso- and microregions. They found that the system described by such equations provided continuity between pore regions and good fit to the measured soil moisture characteristic curve B(h)-h (Fig. 19). However, the application of such a model to solute transport represents a formidable task. The criteria for the demarcation between flow domains are not well understood. Moreover, these demarcation criteria should be conceptually justifiable and experimentally feasible. Another major difficulty is that this approach results in

APPROACHES TO SOLUTE TRANSPORT IN SOILS

125

A Horizon

.

0 0.05 m

0.6

t

I I

I

. Q

I

8

0.5

T 0

-

0.4

E v 0.3

e

c

s8 5

0.2

e

I

I

E i

8 I

R-2

R-1

0.1

R.3 I

I

I I

I I

I

I

0

loo

I

10'

,11111,

I

10'

I

:,

,111,.

I

1

I

,1111,

10'

I

10'

,

1 I

,y

106

Pressure Head (-cm) figure 19 Water retention for the A horizon for five soil cores of the Montevallo soil and the water content-pressure head [O(h)] model (Eqs. ( 5 5 ) , (56), and (57)) for the three regions (R-1,R-2, and R-3) [from Wilson et al. (1992) with permission].

several unknown parameters. Currently, such parameters can only be obtained through curve fitting and thus may not have a physical meaning. Steenhuis et af. (1 990) developed a numerical model for soils with multipleflow domains. Solute and water transport in each flow domain was calculated independently based on Darcy's law for a small time and distance interval and then followed by a mixing process. The mixing process of fluid and solute took place by allowing a fraction of the fluid to be taken out of the flow domain and placed in a common pool and then retrieving a fraction from the common pool. Water content and solute concentrations were updated accordingly from mass balance calculations. Another recent multiregion approach is that proposed by Hutson and Wagenet (1995). The approach is characterized by its flexibility to accommodate various experimental conditions such as unsaturated water flow. The model can be used as a single- or a multiple-flow domain model. Each flow domain was characterized by a saturated hydraulic conductivity and a fractional cross-section area. Flow velocity of each domain can be steady state or transient. Mass transfer between any two flow domains may be convective and/or diffusive depending on water potential. This model has not been tested with experimental results, however. In the two- and multiflow domain approaches, two or more distinct flow velocities must be identified. Each flow domain is represented by an average of

126

L. MA AND H. M. SELIM

pore velocity distribution. This results in an arbitrary discontinuity across flow domains. Skopp and Gardner (1992) considered analyzing solute transport by assuming continuous flow distribution in porous media. To achieve this, a characteristic length (W,) perpendicular to the flow direction was used to describe a unit cell that embodies the full range of velocities in the porous media. Using the method of moment, they were able to predict the velocity dependence of the dispersion coefficient and to explain the increase in D in heterogenous media. However, application of this model to solute transport studies is not feasible at this time.

VI. COUPLED PHYSICAL AND CHEMICAL, NONEQUILIBRIUM MODELS A. FREUNDLICH EQUILIBRIUM MODEL In modeling reactive solutes, physical nonequilibrium models are commonly coupled with chemical retention r-ieuhanisms. The mobile-immobile model described by Eqs. (6) and (7) is the simplest phjsical nonequilibrium model and has been extensively used as a coupled solute transport model. Parameters associated with the mobile-immobile concept are often c stimated from tracer studies, although the behavior of a tracer may not depict that for reactive solutes when physical nonequilibrium is dominant (Brusseau, 1993). Chemical nonequilibrium is described by a series of kinetic reactions. Each reaction is featured by several rate coefficients or equilibrium constants. However, because of the uncertainty about the chemical characteristics of the two soil regions, it is difficult to justify the rate coefficients associated with each region. A commonly used approach is to assume that the same rate coefficient represents both dynamic and stagnant soil regions (Selim and Amacher, 1988; van Genuchten and Wierenga, 1976; Selim et al., 1987; Gaston and Selim, 1990; Selim and Ma, 1995). This assumption is reasonable if the two regions only differ in aggregate size (or pore size). The simplest chemical reaction model is that of linear adsorption isotherms. van Genuchten and Wierenga (1976) divided the soil into two parts. One is in direct contact with the mobile water phase (dynamic soil region), and the other is reacting with solutes in the immobile water phases (stagnant soil region) only. Both regions were assumed to be in equilibrium with solutes in their corresponding water phases:

APPROACHES TO SOLUTE TRANSPORT IN SOILS

127

where K, and N are Freundlich parameters. Incorporation into the CD Eq. (6) requires the derivatives of Sm and Sim,

For the linear case (N = l), this approach was capable of describing 2,4,5-T transport through a Glendale clay loam. The retention mechanisms associated with the mobile and immobile phases were extended to the nonlinear or Freundlich models by Rao et al. ( 1 979).

B. KINETIC TWO-SITEMODEL The two-site (equilibrium-kinetic) model of Selim et al. (1976) can be easily extended to the mobile-immobile model. In the coupled chemical/physical nonequilibrium model, S , represents equilibrium-type sites, and S, represents kinetic-type sites. The retention reactions associated with the equilibrium sites for the mobile and immobile phases, when expressed in a simple linear form, are

where

128

L. MA AND H. M. SELIM

and (&), and (K&, are the equilibrium constants associated with S , and (K& and (K,Jirn are associated with S,. In addition, kl-k4 are the rate coefficients associated with S,. Moreover, ( S , ) , and (S,), are the amounts adsorbed from the mobile water phase on the equilibrium and the kinetic sites, respectively. Similarly, (Sl)imand (S2)imare the corresponding adsorption amounts from the immobile water phase, andf, andfi, are the soil fractions assigned to S , in the mobile and immobile regions, respectively. For linear reactions, an overall equilibrium constant K may be defined as the sum of K , and Kim. This model has been applied to transport data from soil columns (Brusseau et al., 1989; Brusseau, 1991) as well as field data (Brusseau, 1992). It was later extended to account for first-order degradation in each soil region (Brusseau et al., 1992).

C. EQUILIBRIUM ION-EXCHANGE MODEL The mobile-immobile approach was also extended to account for solute retention and transport when ion-exchange reactions are the governing mechanisms. The simplest cases are for two competing (homovalent) ions. A commonly used model assumes that the equilibrium reaction between solution and the exchanger phases for any two cations is not influenced by other species present in the solution. The necessary inputs are the cation-exchange capacity and exchange selectivity coefficients for each ion pair (i and j). Specifically, the ion selectivity coefficient Kli for the dynamic region is considered the same as that for the less accessible sites, i.e., ( KII- )

=

(Kij)i,

=

Kij.

(52)

Such an assumption was used by van Eijkeren and Loch (1984) and Selim et al. (1987) for ion-exchange reactions. Here, the selectivity coefficient Kij may be written as (Sposito, 1981)

where Kij is the selectivity coefficient of ions i over j . In addition, ciand cj are the relative ion concentrations (dimensionless) such that ci = Ci/CTand cj = Cj/C, where Ci and Cj (mmol(+)/ml) are the concentrations of ions i and j in the soil solution, and CT is the total concentration (mmol(+)/ml). Also, si and sj are the amounts retained on the soil matrix surfaces (dimensionless) and are expressed as equivalent fractions where si = Si/STand sj = Sj/S,. Here, Si and Sj are the amounts adsorbed (mmol( +)/g soil) and ST is the cation-exchange (or adsorption) capacity of the soil (mmol(+)/g soil). The parameters ei and ej are the

APPROACHES TO SOLUTE TRANSPORT IN SOILS

129

valency for ions i and j , respectively. For a binary system, the mobile-immobile model for ion 1 can be rewritten under conditions of variable total concentration (CT) as (Selim et al. 1987)

and the transfer equation for the immobile phase is

where P = urn -

(

2Dm

acrnT

crnT

at

and the terms R, and Rimare the retardation factors for the mobile and immobile regions and are expressed as R,

=

1

fpST +() OrnCrnT

2

and the adsorption isotherms are

Similar equations can be obtained for ion 2 in a binary system. This exchange approach was successfully used for binary (Ca-Mg) and ternary systems (NaMg-Ca) (Selim et al., 1987; Gaston and Selim, 1990). Examples of these predictions are shown in Figs. 20 and 21. This classical approach has been modified by incorporating selectivity coefficients that vary with the fractional coverage on the exchanger phase (Mansell et al., 1988). Such a model approach more accurately represents the cation-exchange reactions and has yielded an improved description of Ca, Mg, and Na breakthroughs in the effluent solution from soil columns (see Fig. 22). Selim et al. (1992) developed a nonequilibrium

130

L. MA AND H. M. SELIM

e I?

=: '1z

:

E

Y

z

0

0

6

-

1

0

5

10

15

20

1-2 rnm (aggregates)

5

10

15

25

TWO

20

30

- REGION

25

30

v/v, figure 20 Calcium and Mg BTCs for 1 or 2 nun aggregated Abist soil. Predictions obtained using the classical ion-exchange model are shown by the smooth curves [from Selim et nl. (1987) with permission].

1.0-

0.8 -"0

0.6.-

\ 0

0.4-

0.2 -. 0.o-r 5

10

15

20

25

, 0

Pore Volume (V/V,) figure 21 Calcium and Mg BTCs for aggregated Sharkey clay soil. Predictions obtained using the classical ion-exchange model (sokd curves) and for different a values using the mobile-immobile ion-exchange model (dashed curves) [from Gaston and Selim (1989) with permission].

APPROACHES TO SOLUTE TRANSPORT IN SOILS

131

0.5 CONSTANT SELECTIVITIES 0.4

0.3 0.2 0.1

f

0 0

3

0.5

1

2

3

4

5

I

VARIABLE SELECTIVITIES

0

1

2

3

4

6

5

6

VIV, Figure 22 Sodium and Mg BTCs from soil columns of Yolo soil. Solid and dashed curves are predictions using the classical and two-region ion-exchange models, respectively. Exchange selectivity coefficients were considered constant (top) or variable (bottom) [from Mansell el a / . (1988) with permission].

ion-exchange model. However, to our knowledge this kinetic ion exchange has not been extended to the multiple-domain model, the mobile-immobile model, or other physical nonequilibrium approaches.

D. SECOND-ORDER MOBILE-IMMOBILE MODEL Selim and Amacher (1988) were the first to introduce chemical nonequilibrium retention reactions along with the physical nonequilibrium approach. Specifically, they introduced a second-order two-site kinetic model into the mobileimmobile model. An important feature of the second-order approach is that an adsorption maximum (or capacity) is assumed. This maximum represents the total number of adsorption sites per unit mass or volume of the soil matrix. It is also considered an intrinsic property of an individual soil and is thus assumed constant. An irreversible reaction and a reversible reaction were assumed to occur independently in each soil region. The irreversible sites may also be

132

L. MA AND H. M. SELIM

viewed as a sink or a source. The reversible sites are constrained by the adsorpand (Sim),,,. Specifically, a tion maxima S,,,, which was divided into (S,),,, dimensionless temfdenotes the ratio or fraction of dynamic or active sites to the maximum or total adsorption sites Lf = (S,),,,~S,,,,,]. For irreversible retention, the governing mechanism of retention for each region was assumed to follow a first-order type reaction in the mobile and immobile regions:

where ks is an irreversible rate coefficient (h-1) common to both regions. In contrast, for the reversible kinetic reaction, the reaction rates are governed by second-order kinetics and written as (Selim and Amacher, 1988)

and

where k, and k2 are forward and backward rate coefficients (h-l), respectively. and +im represent the vacant or unfilled sites (pg/g soil) within the Here, and dynamic and the stagnant regions, respectively. In addition, the terms +im can be expressed as

+,

+,

+m +im

(66)

= (Srn)max - S m = f s r n a x - Srn

= (Sim)max

- Sim

=

(1

-f)Smax

-

Sim,

(67)

where S,,,, (S,),,,, and (S,,),,, are the total amount of sites in the soil matrix, total sites in the dynamic region, and the total sites in the less accessible region (pg/g soil), respectively. An important feature of the second-order retention approach is that similar reaction rate coefficients (k, and k2) associated with the dynamic and stagnant regions were chosen; that is, the retention mechanism is equally valid for both regions. A similar assumption was made by van Genuchten and Wierenga ( 1976) for equilibrium linear and Freundlich-type reactions and by Selim et al. (1987) and Mansell et al. (1988) for selectivity coefficients for homovalent ion-exchange reactions as discussed previously. As t a, i.e., when both the dynamic (or active) sites and the sites in the stagnant region

APPROACHES TO SOLUTE TRANSPORT IN SOILS

133

achieve local equilibrium, the following expressions are obtained. For the active sites associated with the mobile region.

or

Here, (K&, and (Kk)im represent equilibrium constants for the retention reactions associated with the mobile and immobile regions, respectively. These equilibrium constants are analogous to the selectivity coefficients associated with ionexchange reactions (see Selim et al., 1987). An example of the capability of the second-order mobile-immobile model for BTC predictions is given in Fig. 23. Here, the predictions of Cr(V1) BTCs were based on parameter estimates (S,,,,, k , , k2, etc.) obtained from kinetic batch results. Estimates for the parameter F (= em/@) were based on soil-moisture retention relations andfwas assumed to be

I

OLIVIER-SOMIM 1.0--

.

0

1

2

3

4

5

v 1 vo

6

7

8

Figure 23 Effluent concentation distributions for Cr(V1) in Windsor soil. Curves A, B , C, and D are predictions using the second-order mobile-immobile model with batch rate coefficients for C,, of 25, 5 . 2, and 1 pg/ml. respectively [from Selim and Amacher (1988) with permission].

L. MA AND H. M. SELIM

134

equal to F. Estimates for D, and a were obtained from 9 s . (18) and (24), respectively, in which uniform size aggregates were assumed (see Selim and Amacher, 1988). Predicted BTCs were obtained using different sets of batch rate coefficients due to their strong dependence on input (initial) concentrations (C,’s). The closest predictions to experimental Cr(V1) measurements were obtained from batch rate coefficients at low C, values (C, I 10 pg/ml).

E. MODIFIED SECOND-ORDER TWO-SITEMODEL Selim and Amacher (1988) proposed a second-order two-site kinetic model in which two reversible (S, and S,) and one irreversible (Sir)type sites were considered. The amount of solutes adsorbed on the reversible sites was constrained by the adsorption maximum (Smax).A parameterf, was introduced such thatf, = (S,)max/Smax. Vacant sites available for sites 1 and 2 are denoted as and +*, respectively, and may be described by

41= (Sllrnax +2 = (Sdrnax

-

- S,

SI = f s S m a x - $1 =

(1 - fs)Smax -

S2.

(70)

In its present formulation, this model was not coupled with physical nonequilibrium models. This is due to the excessive number of model parameters. In addition, methods for estimating or for directly making measurements of several of the model parameters are not available. Ma and Selim (1994b) proposed a modified second-order formulation that is based on the assumption of two reversible reactions (S, and S,) and one irreversible reaction (Sir).Unlike that of Selim and Amacher (1988), the modified model assumed that the vacant sites are equally accessible to S , and S,. That is, S , and S, can compete for the unoccupied adsorption sites regardless whether they are of type-1 or type-2 sites. Thus, the partitioning parameter& is not required. In addition, an irreversible site Sirwas assumed to be consecutive to S,. The total adsorption sites (Smax)can be defined as Smax

+

=

4 + S, + &,

(71)

where is the total vacant sites in the soil (pg/g soil). The governing kinetic reaction rates present in the soil solution phase ( C ) and that in the reversible and irreversible phases (Si,S,, and Sir)may be written as

135

APPROACHES T O SOLUTE TRANSPORT IN SOILS

where k , and k, are forward rate coefficients (ml/kg/h), and k, and k4 are the corresponding backward rate coefficients (h- I ) associated with the second-order model. Assuming S, is in equilibrium, the adsorption term ( d s l a t ) for a modified second-order equilibrium/kinetic model can be expressed as

where K (= k , / k , ) is an equilibrium parameter associated with S , (ml/pg). Ma and Selim (1994b) found that this model not only described atrazine adsorptiondesorption kinetics at several initial concentrations and soi1:water ratios but also successfully predicted atrazin BTCs from soil columns under different experimental conditions (Ma and Selim, 1994~).Selim and Ma (1995) successfully applied this modified model to the mobile-immobile physical nonequilibrium approach as discussed in the following section.

F. MODIFIED MOBILE-IMMOBILE MODEL One difficulty in implementing the coupled chemical-physical nonequilibrium So far, models is how to estimate the fraction of the dynamic soil region there is no experimental technique for quantifyingfdue to the vague definition of mobile-immobile water phases. One way to obtainfis through curve fitting (van Genuchten and Wierenga, 1977). The other commonly adopted approach is to assume the value off to be the same as the volumetric fraction of the mobile water phase ( F ) (Nkedi-Kizza et al., 1983, 1984; Selim et al., 1987; Selim and Amacher, 1988; Gaston and Selim, 1990). However, there is no evidence that this assumption is valid (Selim and Ma, 1995). Due to the difficulty in estimating this conceptual parameterf, Selim and Ma (1995) proposed a modified two-region model in which nofis incorporated into the model. Figure 24 shows a schematic diagram of the modified mobile-immobile model in which soil aggregates are surrounded by both mobile and immobile water phases where the dynamic and stagnant soil regions in the soil are regarded as a continuum and connected to one another. Solute retention may occur concurrently in the dynamic and stagnant regions until equilibrium conditions are attained or all vacant sites for a soil aggregate become occupied (filled). The rates of retention reactions in the mobile and immobile phases are considered a function of the total vacant sites in the soil. Specifically, no distinctions between the fraction of sites associated with the dynamic region from that of the stagnant region are made, that is, the amount retained from the mobile phase, for exam-

u).

136

L. MA AND H. M. SELIM

cm

+

Cim

.... .... ... ... ... ... .... ..... ... .... .... .... . .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .... .... .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. .. i. .i i. 1. F J .... ................. . . . . .... .... . . . . . . . . . . . . . . . . . : : : : IMMOBILE WATER Figure 24 A schematic of the modified two-region model [from Selim and Ma (1995) with permission].

ple, affects the total amount of vacant sites available for retention in the immobile water phase and vice versa. Based on these concepts, the CDE for the modified approach is

with

Selim and Ma (1995) coupled this mobile-immobile model with a modified second-order model of Ma and Selim (1 994b). Applying the modified secondorder model from the previous section to the mobile-immobile concept, the maximum adsorption sites S,,, is described as

Sm,,

=

+ + (Sl), +

(s2)m

+ ( s , ) i m + (s2)im.

(78)

The reaction rates associated with S , , S,, and Sirin the mobile and immobile regions are

asim -

at

KBi,b,

%+

a+

KBi,Ci, at + k,Bi,Ci,+

-

k4(S2)im. (86)

This model is different from that of Brusseau et al. (1989) in two respects. One is the second-order assumption, and the other is the rate coefficients. Here, the ~~

f values

RMS

f = 1 f =0 f =F no f

0.0524 0.1512 0.2680 0.0498

.....

0.8

. _

- ._

0

0

3 0

5

10

15

3

Pore Volume (V/V,) Figure 25 Measured (solid circles) and predicted atrazine BTC with 2-4 mm aggregated Sharkey soil, column length of 15 cm, velocity of 1.08 cmih, and initial concentration of 10.15 &g/ml. Atrazine predictions were based on mobile-immobile approaches and the model parameters were from batch experiments. f is the fraction of mobile water [from Selim and Ma (1995) with permission].

L. MA AND H. M. SELLM

138

f volues

-.

0

5

....

10

f = 1

15

RMS 0.1706

20

5

Pore Volume (V/V,)

Ffgure 26 Measured (solid circles) and predicted atrazine BTC with 4-6 mm aggregated Sharkey soil, column length of 10 cm, velocity of 1.76 cm/h, and initial concentration of 11.33 pgiml. Atrazine predictions were based on mobile-immobile approaches and the model parameters were from batch experiments. f is the fraction of mobile water [from Selim and Ma (1995) with permission].

equilibrium constants are assumed to be different for the S, and S, sites and the rate coefficients associated with S, and S, are identical for both mobile-immobile regions. Using parameters independently derived from batch experiments, Selim and Ma (1995) found that this modified two-region model provided better overall predictions of atrazine BTCs in a Sharkey clay soil for different aggregate sizes, column lengths, flow velocities, and flow interruptions (Figs. 25 and 26). However, as shown in Figs. 25 and 26, the original mobile-immobile model did not provide adequate predictions of atrazine BTCs with either f = 0 or f = F . Thus, the introduction off might not be necessarily due to considerable uncertainty in f (Ma and Selim, 1996). Other than the mobile-immobile two-region model, coupling of chemical nonequilibrium with other physical nonequilibrium approaches has not been found in the literature.

VII. FIELD APPLICATIONS In field experiments, physical nonequilibrium behavior is commonly observed. Thus, models accounting for such a phenomenon are necessary (McCoy et al., 1994). Generally, field applications of physical nonequilibrium models are less successful than controlled laboratory experiments. Such an incapability can be attributed to spatial variability of field condition as well as ill-defined initial and boundary conditions. Two types of model applications are frequently found in the literature. One is deterministic and the other is stochastic. Stochastic

APPROACHES TO SOLUTE TKANSPORT IN SOILS

139

models that account for physical nonequilibrium have been proposed by Utermann et al. (1990) and Grochulska and Kladivko (1994). Here, the water flux was divided into preferential and matrix flow. Each flow domain was described with a different probability density function. Deterministic models that deal with physical nonequilibrium include CALF (Nicholls et al., 1982), LEACHA (Hutson and Wagenet, 1993), TETrans (Corwin ef al., 1991), and RZWQM (Ahuja et al., 1993). Due to their ease in accommodating chemical nonequilibrium retention reactions during solute transport, deterministic-type models have been more widely tested than stochastic models. The RZWQM model of Ahuja er al. (1993) accounts for macro-, meso-, and micropores with water flow in macro- and mesopores only. Half of the macroporosity was assumed to be continuous and the remaining was assumed discontinuous. In the surface horizon, the shape of macropores was considered cylindrical with vertical arrangements. Lateral flow was assumed to increase with soil depth. Macropore flow was assumed to obey Poiseuille’s law and was allowed to flow into dead-ended macropores. Water and solute may enter the soil matrix by radial or lateral infiltration. Adsorbed solute from macropores was uniformly mixed and equilibrated with the soil and water in the mesopores. Mass transfer between meso- and micropores was considered to be diffusion controlled. Water infiltration rate during rainfall was calculated based on a simplified Green-Ampt approach. Water redistribution in the soil matrix between rainfall events was governed by the Richards equation. Solute was adsorbed in the three regions simultaneously based on either equilibrium-type isotherms or first-order kinetic reactions. The RZWQM model was recently modified to incorporate the effects of soil compaction or organic lining on lateral infiltration between different pore regions. To improve predictions of tracer solutes such as Br, the model was further refined by assuming a buffer soil layer of 0.5 mm around the macropores. As stated by Ahuja er al. (1995), such a model must first be initialized by providing the appropriate initial and boundary conditions, defining soil properties, and specifying the experimental treatment. The methods used to estimate a number of model parameters remain at best controversial. The models of Nicholls et al. (1982) and Hutson and Wagenet (1993) are less rigorous than that of Ahuja ef al. (1993), but show considerable promise in modeling pesticide behavior under field conditions. These models regard the water in the soil pore space in the traditional sense according to mobile and immobile water with mass flow or percolation restricted to the mobile phase only. Water infiltration was assumed to be instantaneous at each time interval. The direction and amount of flow were calculated based on the soil-water capacity. Water redistribution was based on water content gradient. Solute was transferred along with water and may be partitioned into the soil matrix at each time interval. The TETrans model of Corwin et a / . ( 1991) adopted a similar approach in which flow velocity was indirectly controlled by irrigation and precipitation distribu-

L. MA AND H. M. SELLM

tion. However, Corwin et af. (1991) did not explicitly define solute concentrations in the mobile and immobile water phases; rather, they defined a mobility coefficient as the fraction of resident soil water subject to displacement. Such a mobility coefficient is temporally and spatially variable. Steenhuis et af. (1994) applied a two-flow domain (preferential flow and matrix flow) model with limited success to field experiments. Preferential flow through the macropores was considered instantaneous and treated as plug flow, while matrix flow obeys the CDE. A mixing layer was discriminated at the soil surface where water and solute are fully mixed before partitioning into the matrix and preferential flows in the lower soil layers. In addition, different distribution coefficients for reactive solutes were used during the adsorption and desorption processes. Other specific examples of applying the physical nonequilibrium concept in describing field data are those of Goltz and Roberts (1986b, 1988), Parrish et al. (1992), and Brusseau (1992). Parrish et af. (1992) found that the mobile-immobile approach better described the transport of aldicarb, metolachlor, and bromide in a field study than the traditional CDE approach. Brusseau (1 992) applied the coupled physical (mobile-immobile) and chemical (two-site adsorption) model to four field experimental data sets reported in the literature. Using independently derived parameter estimates, he was able to explain the behavior of tetrachloroethene, tetrahydrofuran, tetrachloromethane, diethylether, trichloroethene, and Sr under large-scaled field conditions. Goltz and Roberts (1986b) also applied the mobile-immobile model with instantaneous solute adsorption to an experiment from Borden, Ontario, and observed a better description of organic solute transport in an unconfined sand aquifer. This data set was later analyzed by Goltz and Roberts (1988) using the moment method and the three-dimensional mobile-immobile approach.

VIII. SUMMARY AND CONCLUSION In this chapter, we described several physical nonequilibrium approaches that are commonly found in the soil science literature. Because the soil is a complex system and is ill-defined, various degrees of simplifications are necessary depending on the experimental conditions. A number of options can be chosen for the purpose of simplification such as: transient flow versus steady flow, singleflow domain versus multiple-flow domains, laminar flow versus turbulent flow, restricted solute exchange versus solute exchange between flow domains, convective mass exchange versus diffusive mass exchange, deterministic versus stochastic models, laboratory versus field-scale models, among others (McCoy et al., 1994). On the other hand, criteria for selecting the appropriate model often depends on the availability and the certainty of the necessary parameters (Ma and Selim 1996).

APPROACHES TO SOLUTE TRANSPORT IN SOILS

141

Currently, physical nonequilibrium modeling approaches may be best regarded as conceptual rather than mechanistic or deterministic in nature. Tremendous difficulties are often encountered in the translation of ideas and conceptual approaches into workable mathematical tools. The problem often lies in the vague definition of the physical nonequilibrium concept, which is directly due to our inability to fully describe the water flow geometry and the associated pore size arrangements. There are no standard criteria such that the soil pores can be grouped experimentally. The spatial connectiveness of pores and temporal variation of porosities have seldom been studied in soil science. As pointed out by Beven and Germann (1982), classification of macropores, for example, remains rather arbitrary and may not necessarily relate to flow characteristics. The complexity of the soil system exceeds that manifested by the mobileimmobile or the multiflow domain concept. In fact, water can be intercepted in discontinuous macropores (internal catchment), which may be viewed as mobile phase in the traditional mobile-immobile concept (Booltink and Bouma, I99 1). As a result, the effective macropores may be less than the actual volume of macropores (Dunn and Philips, 1991b). Another associated concept is that of the dynamic and stagnant soil regions. Because one cannot distinguish the mobile from the immobile phase with absolute certainty, the associated soil regions cannot be equally quantified with total confidence. In addition to the difficulty of differentiating the flow domains, major soil parameters, such as the dispersion coefficient (D)and flow velocity (u), for each flow domain are often difficult to measure parameters. Moreover, there are an exceedingly large number of conceptual parameters. As a result, physical nonequilibrium models in their present forms are, in general, difficult to fully evaluate and apply to field data sets. Research is needed in the area of identifying the appropriate initial and boundary conditions imposed in physical nonequilibriiim models, especially those associated with the soil surface, channels, and macropore walls and their connectiveness across strata or horizons of the soil profile. Experimental methods are also needed to quantify the porosity and characteristics of each flow domain and associated mass transfer between flow domains under various flow and geometrical conditions. Physical nonequilibrium concepts of water and solute transport are one of the research areas in the forefront of soil physics and hydrology disciplines. Although no single model has been universally accepted and/or validated, the physical nonequilibrium concepts have advanced our understanding of the behavior of water and solute in natural soil environments. These concepts also reveal the limitation of traditional approaches in proposing valuable guidelines for future research. Despite their shortcomings and difficulties for application purposes, these models have explained, to some extent, physical nonequilibrium phenomena, such as preferential flow, multiple-peak breakthrough, extensive tailing, and irregular solute distribution in soils. Several instructive and empirical concepts andlor parameters have also been introduced in the soil physics and

142

L. MA AND H. M. SELIM

hydrology disciplines. Examples include the mobile-immobile waters, dynamic-stagnant soil regions, mass transfer coefficient, macroporosity, mesoporosity, and microposity. Future advances in experimental methods, mathematical solutions, and conceptual innovations are likely to make physical nonequilibrium approaches more widely used tools in the interpretation of the behavior of solute transport under field conditions.

APPENDIX: NOMENCLATURE First-order mass transfer coefficient (h-1) Mass transfer parameter Velocity ratio (ItAhB) Mass transfer between fast and slow flow domains (pg cm-3 h-1) Hydrostatic pressure gradient Interaggregate pore size (cm) Soil-water content (cm3 cm-3) Water content of the fast flow domain (cm3 cm-3) Water content of the slow flow domain (cm3 cm-3) Water content in the mobile phase (cm3 cm-3) Water content in the immobile phase (cm3 cm-3) Dispersivity (cm) Coefficient of viscosity Average pore water velocity (cm h-1) Average pore water velocity in the fast flow domain (cm h-1) Average pore water velocity in the slow flow domain (cm h-1) The valency for ion i The valency for ion j Pore water velocity in the mobile phase (cm h-1) Diffusion coefficient parameter Soil bulk density ( g cm-3) Tortuosity factor (
APPROACHES TO SOLUTE TRANSPORT IN SOILS

Ci

Ci

d

D

f fim

fs

F

Half width of rectangular aggregates (cm) Macropore radius of hollow cylindrical aggregates (cm) Soil cylinder radius surrounding macropores (cm) Solute concentration in soil solution (pg/ml) Solute concentration in the fast flow domain (pg/ml) Solute concentration in the slow flow domain (pglml) Relative ion concentrations of i ion, ci = Ci/CT Concentrations of ion i in the soil solution (mmol(+)/ml) Solute concentration in the immobile water phases (pg/ml) Relative ion concentrations of j ion, ci = C,/CT Concentrations of ion j in the soil solution (mmol(+)/ml) Solute concentration in the mobile water phases (pg/ml) Input solute concentration (pg/ml) Solute concentration in a sphere at t = 0 (pg/ml) Total ion concentration (mmol(+)/ml) Water film thickness of the film diffusion model or thickness of the immobile water (cm) Apparent hydrodynamic dispersion coefficient (cm2 h-1) Molecular diffusion coefficient (cm2 h-1) Dispersion coefficient of the fast flow domain (cm2 h-1) Dispersion coefficient of the slow flow domain (cm2 h-1) Effective molecular diffusion coefficient (= D06, cm2 h-l) Diffusion coefficient in the immobile water phase (cm2 h-l) Hydrodynamic dispersion coefficient in the mobile phase (cm2 h-1) Fraction of retention or sorption sites in direct contact with the mobile water phase or fraction of the dynamic soil region Soil fraction assigned to S1 in the mobile region Soil fraction assigned to S, in the immobile region Soil fraction assigned to S, Mobile water fraction (= Om/@) or fast flow water fraction (QA/@)

g H

143

Acceleration due to gravity (cm h-2) Thickness of mobile water (cm) Forward rate coefficient (h-l) Backward rate coefficient (h -1) Forward rate coefficient (h-l) Backward rate coefficient (h-1) Irreversible rate coefficient (h-l) Conductivity of the soil matrix (cm h-1) Freundlich adsorption coefficient Ion selectivity coefficient of i over j

L. MA AND H. M. SELIM

Side length of a cube (cm) Solute transport length (cm) A geometry factor Freundlich parameter A partition coefficient A parameter used to calculate a Radial coordinate in a sphere (cm) Overall retardation factor (OR = Bi,Rim + O,R,) Retardation factor of the mobile phase Retardation factor of the immobile phase Average jth pore radius Amount of solute sorbed or retained (&g soil) Relative amount of ion i retained on the soil matrix surfaces, si = si/sT

Amount of ion i adsorbed on the soil matrix (mmol(+)/g soil) Amount of solute adsorbed from the immobile water phases (Pg/g soil) Relative amount of ion j retained on the soil matrix surfaces, si = Sj/ST

Amount of ion j adsorbed on the soil matrix (mmol(+)/gsoil) Amount of solute adsorbed from the mobile water phases (Pg/g soil) Total amount of the sites in the soiI matrix (pg/g soiI) Cation-exchange (or adsorption) capacity of the soil (mmol(+)/g soil) Total sites in the less accessible region (pg/g soil) Total sites in the dynamic region (pg/g soil) Time (h) Mean residence time (h) The spatial coordinate in the flow direction (cm) Distance measured from the pore wall, perpendicular to the flow direction (cm)

REFERENCES Ahuja, L. R . , DeCoursey, D. G., Barnes, B. B . , and Rojas, K . W. (1993). Characteristics of macropore transport studied with the ARS root zone water quality model. Trans.ASAE 36,369380. Ahuja, L. R . , Johnsen, K . E., and Heathman, G. C. (1995). Macropore transport of a surface-applied bromide tracer: Model evaluation and refinement. Soil Sci. SOC. Am. J . 59, 1234-1241. Anderson, I. L., and Bouma, J. (1977). Water movement through pedal soils: I . Saturated flow. Soil Sci. Soc. Am. J . 41, 413-418.

APPROACHES TO SOLUTE TRANSPORT IN SOILS

145

Andreini, M. S.. and Steenhuis, T. S. (1990). Preferential paths of flow under conventional and conservation tillage. Geoderma 46, 85-102. Beven, K.. and Germann. P. (1982). Macropores and water flow in soils. Water Resour. Res. 18, 131 1- 1325. Biggar, J. W., and Nielsen, D. R. (1963). Miscible displacement: V. Exchange processes. Soil Sri. Soc. Am. Proc. 27, 623-627. Booltink, H. W. (3.. and Bouma, J. (1991). Physical and morphological characterization of bypass flow in a well-structured clay soil. Soil Sci. Soc. Am. J . 55, 1249-1254. Bouma, J. (1991). Influence of soil macroporosity on environmental quality. Adv. Agron. 46, 1-37. Bmsseau, M. L. (1994). Transport of reactive contaminants in heterogeneous porous media. Rev. Geophvs. 32, 285-313. Brusseau, M. L. (1993). The influence of solute size, pore water velocity, and intraparticle porosity on solute dispersion and transport in soil. Water Resour. Res. 29, 1071-1080. Brusseau, M. L. ( 1992). Transport of rate-limited sorbing solutes in heterogeneous porous media: Application of a one-dimensional multifactor nonideality model to field data. Water Resour. Res. 28, 485-2497. Brusseau, M. L. (1991). Transport of organic chemicals by gas advection in structured or heterogeneous porous media: Development of a model and application to column experiments. Water Resour. Res. 27, 3 189-3 199. Brusseau, M. L., and Rao, P. S. C. (1990). Modeling solute transport in structured soils: A review. Geoderma 46, 169- 192. Brusseau, M. L.. Jessup, R. E., and Rao, P. S. C. (1989). Modeling the transport of solutes influenced by multiprocess nonequilibrium. Water Resour. Res. 25, 1971- 1988. Brusseau, M. L., Jessup. R. E., and Rao, P. S. C. (1992). Modeling solute transport influenced by rnultiprocess nonequilibrium and transformation reactions. Water Resour. Res. 28, 175-182. Chen, C.. and Wagenet, R. J. (1992). Simulation of water and chemicals in macropore soils. I. Representation of the equivalent macropore influence and its effect on soil water flow. J. Hydro/. 130, 105-126. Chen, C., Thomas, D. M., Green, R. E., and Wagenet. R. J. (1993). Two-domain estimation of hydraulic properties in macropore soils. Soil Sci. Soc. Am. J . 57, 680-686. Clothier, B. E., Kirkham, M. B., and Mclean, J. E. (1992). In situ measurement of the effective transport volume for solute moving through soil. Soil Sci. Soc. Am. J . 56, 733-736. Coats, K. H., and Smith, B. D. (1964). Dead-end pore volume and dispersion in porous media. Soc. Pet. Eng. J. 4, 73-84. Corwin, D. L., Waggoner, B. L., and Rhoades, J. D. (1991). A functional model of solute transport that accounts for bypass. J. Environ. Qual. 20, 647-658. Davidson, J. M., and Chang, R. K. (1972). Transport of picloram in relation to soil physical conditions and pore-water velocity. Soil Sci. Sor. Am. Pror. 36, 257-261. Davidson, J. M., and McDougal, J. R. (1973). Experimental and predicted movement of three herbicides in a water-saturated soil. J. Environ. Qual. 2, 428-433. Deans, H. H. (1963). A mathematical model for dispersion in the direction of flow in porous media. Soc. Pet. Eng. J . 3, 49-52. De Smedt, F.. and Wierenga, P. I. (1979). Mass transfer in porous media with immobile water. J. Hvdrol. 41, 59-67. De Smedt. F.. and Wierenga, P. J. (1984). Solute transfer through columns of glass beads. Water Resour. Res. 20, 225-232. De Smedt, F., Wauters, F.. and Sevilla, J. (1986). Study of tracer movement through unsaturated sand. Geoderma 38, 223-236. Duguid, J. 0.. and Lee, P. C. Y. (1977). Flow in fractured porous media. Water Resour. Res. 13, 558-566.

L. MA AND H. M. SELIM Dunn, G. H., and Philips, R. E. (1991a). Macroporosity of a well-drained soil under no-till and conventional tillage. Soil Sci. SOC. Am. J. 55, 817-823. Dunn, G. H., and Philips, R. E. (1991b). Equivalent diameter of simulated macropore systems during saturated flow. Soil Sci. SOC. Am. J . 55, 1244-1248. Dykhuizen, R. C. (1990). A new coupling term for dual-porosity models. Water Resour. Res. 26, 35 1-356. Edwards, W. M., Norton, L. D., and Redmond, C. E. (1988). Characterizing macropores that affect infiltration into notilled soil. Soil Sci. SOC.Am. J. 52, 483-487. Edwards, W. M., Shipitalo, M. J., Dick, W. A,, and Owens, L. B. (1992). Rainfall intensity affects transport of water and chemicals through macropores in no-till soil. Soil Sci. SOC.Am. J . 56, 52-58. Ela, S. D., Gupta, S. C., and Rawls, W. J. (1992). Macropore and surface seal interactions affecting water infiltration into soil. Soil Sci. SOC. Am. J. 56, 714-721. Elrick, D. E., Erh, K. T., and Krupp, H. K. (1966). Application of miscibledisplacementtechniques to soils. Water Resour. Res. 2, 717-721. Gaston, L. A,, and Locke, M. A. (1995). Fluometuron sorption and transport in Dundee Soil. J . Environ. Qual. 24, 29-36. Gaston, L. A., and Selim, H. M. (1990). Transport of exchangeable cations in an aggregated clay soil. Soil Sci. SOC. Am. J. 54, 31-38. Gerke, H. H., and van Genuchten, M. Th. (1993). A dual-porosity model for simulating the preferential movement of water and solutes in structured porous media. Water Resour. Res. 29, 305-319. Goltz, M. N., and Roberts, P. V. (1986a). Three-dimensional simulations for solute transport in an infinite medium with mobile-immobile zone. Water Resour. Res. 22, 1139- I 148. Goltz, M. N., and Roberts, P. V. (1986b). Interpreting organic solute transport data from a field experiment using physical nonequilibrium models. J. Contamin. Hydrol. 1, 77-93. Goltz, M. N., and Roberts, P. V. (1988). Simulations of physical nonequilibrium solute transport models: Application to a large scale field experiment. J . Contamin. Hydrol. 3, 37-63. Gottschlich, C. F. (1963). Axial dispersion in a packed bed. Am. Instir. Chem. Engin. Journ. 9,88-92. Green, R. E., Yamane, V. K., and Obien, S. R. (1968). Transport of atrazine in a latosolic soil in relation to adsorption, degradation, and soil water variables. fnt. Congr. Soil Sci. Trans. 9th (Adelaide, Aust.) 1, 195-204. Grochulska, J., and Kladivko, E. J. (1994). A two-region model of preferential flow of chemicals using a transfer function approach. J . Environ. Qual. 23, 498-507. Hamlen, C. J., and Kachanoski, R. G . (1992). Field solute transport across a soil horizon boundary. Soil Sci. SOC. Am. J. 56, 1716-1720. Homberger, G. M., Beven, K. J., and Germann, P. F. (1990). Inferences about solute transport in macroporous forest soils from time series models. Geoderrna 46, 249-262. Hutson, J. L., and Wagenet, R. J. (1995). A multiregion model describing water flow and solute transport in heterogenous soils. Soil Sci. SOC.Am. J. 59, 743-75 I. Hutson, J. L., and Wagenet, R. J. (1993). A pragmatic field-scale approach for modeling pesticides. J. Environ. Qual. 22, 494-499. Jarvis, N. J., Jansson, P-E., Dik, P. E., and Messing, I. (1991a). Modeling water and solute transport in macroporous soil. 11. Model description and sensitivity analysis. J . Soil Sci. 42, 59-70. Jarvis, N. J., Bergstrom, L., and Dik, P. E. (1991b). Modeling water and solute transport in macroporous soil. 11. Chloride breakthrough under non-steady flow. J. Soil Sci. 42, 71-81. Jaynes, D. B., Logsdon, S. D., and Horton, R. (1995). Field method for measuring mobile/immobile water content and solute transfer rate coefficient. Soil Sci. SOC.Am. J . 59, 352-356. Jury,W. A., and Fliihler, H. (1992). Transport of chemicals through soil: Mechanisms, models, and field applications. Adv. Agron. 47, 141-201.

APPROACHES TO SOLUTE TRANSPORT IN SOILS

147

Kay, B. D., and Elrick, D. E. (1967). Adsorption and movement of lindane in soils. Soil Sci. 104, 3 14-32 I. Koch, S., and Fliihler, H. (1993). Non-reactive solute transport with micropore diffusion in aggregated porous media determined by a flow-intemption method. J. Conramin. Hydrol. 14, 3954. Kookana, R . S., Schuller, R. D., and Aylmore, L. A. G. (1993). Simulation of simazine transport through soil columns using time-dependent sorption data measured under flow conditions. J . Contamin. Hydrol. 14, 93- 115. Li, Y., and Ghodrati, M. (1994). Preferential transport of nitrate through soil columns containing root channels. Soil Sci. Soc. Am. J. 58, 653-659. Li, Y.. and Ghodrati, M. (1995). Transport of nitrate in soils as affected by earthworm activities. J. Diviron. Qual. 24, 432-438. Lindstrom, F. T., and Boersma, L. (1971). A theory on the mass transport of previously distributed chemicals in a water saturated sorbing porous medium. Soil Sci. 111, 192-199. Logsdon, S. D. (1995). Flow mechanisms through continuous and buried macropores. Soil Sci. 160, 231-242. Logsdon, S. D., Allmaras, R. R., Wu, L.. Swan, J. B., and Randall. G. W. (1990). Macroporosity and its relation to saturated hydraulic conductivity under different tillage practices. Soil Sci. Soc. Am. J . 54, 1096-1101. Luxmoore, R. J. (1981). Micro-, meso-,and macroporosity of soil. Soil Sci. Soc. Am. J. 45, 671672. Ma, L., and Selim, H. M. (1996). Atrazine retention and transport in soils. Rev. Environ. Contamin. Toxicol. 145, 129-173. Ma, L., and Selim, H. M. (1995). Transport of a nonreactive solute in soils: A two-flow domain approach. Soil Sci. 159, 224-234. Ma, L., and Selim, H. M. (1994a). Tortuosity, mean residence time and deformation of tritium breakthroughs from soil columns. Soil Sci. Soc. Am. J. 58, 1076-1085. Ma, L., and Selim, H. M. (1994b). Predicting atrazine adsorption-desorption in soils: A modified second-order kinetic model. Water Resour. Res. 30, 447-456. Ma, L., and Selim, H. M. (1994~).Predicting the transport of atrazine in soils: Second-order and multireaction approaches. Water Resour. Res. 30,3489-3498. Mansell, R. S., Bloom, S. A., Selim, H. M., and Rhue, R. D. (1988). Simulated transport of multiple cations in soil using variable selectivity coefficients. Soil Sci. Soc. Am. J . 52, 15331540. McCoy, E. L., Boast, C. W., Stehouwer, R. C., and Kladivko, E. J. (1994). Macropore hydraulics: Taking a sledgehammer to classical theory. I n “Soil Processes and Water Quality” (R. La1 and B. A. Stewart, Eds.), pp. 303-348. CRC, Boca Raton, FL. Munyankusi, E., Gupta, S. C., Moncrief, J. F., and Berry, E. C. (1994). Earthworm macropores and preferential transport in a long-term manure applied typic hapludalf. J. Environ. Qual. 23, 773784. Nicholls, P. H., Walker, A,, and Baker, R. J. (1982). Measurement and simulation of the movement and degradation of atrazine and metribuzin in a fallow soil. festic. Sci. 12, 484-494. Nkedi-Kizza, P., Rao, P. S. C., Jessup, R. E., and Davidson, 1. M. (1982). Ion exchange and diffusive mass transfer during miscible displacement through and aggregate Oxisol. Soil Sci. SOC. Am. J. 46,471-476. Nkedi-Kizza, P., Biggar. J. W., van Genuchten, M. Th., Wierenga, P. J., Selim, H. M., Davidson, J. M., and Nielson, D. R. (1983). Modeling tritium and chloride 36 transport through an aggregated Oxisol. Water Resour. Res. 19, 691-700. Nkedi-Kizza. P., Biggar, J. W., Selim, H. M., van Genuchten, M. Th., Wierenga, P. J., Davidson, J. M., and Nielson, D. R. (1984). On the equivalence of two conceptual models for describing

148

L. M A AND H. M. SELIM

ion exchange during transport through an aggregated Oxisol. Water Resour. Res. 20, 11231130. Oostrom, M . , Dane, J. H., and Gwen, 0. (1992). Dispersivity values determined from effluent and nonintrusive resident concentration measurements. Soil Sci. Sue. Am. J . 56, 1754- 1758. Othmer. H., Diekkruger, B., and Kutilek. M. (1991). Bimodal porosity and unsaturated hydraulic conductivity. Soil Sci. 152, 139- 150. Parker, J. C., and Valocchi, A. J. (1986). Constraints of the validity of equilibrium and first-order kinetic transport models in structure soils. Wafer Resour. Rcs. 22, 399-407. Parrish, R. S., Smith, C. N., and Fong, F. K. (1992). Tests of pesticide Root Zone Model and the aggregate model for transport and transformation of aldicarb, metolachlor, and bromide. J. Environ. Qual. 21, 685-697. Passiourd, J. B. (1971). Hydrodynamic dispersion in aggregated media. I . Theory. Soil Sci. 111, 339- 344. Pignatello, J. J., Ferrandino, F. J., and Huang, L. Q. (1993). Elution of aged and freshly added herbicides from a soil. Environ. Sei. Techno/. 27, 1563-1571. Quisenberry. V. L., Philips, R. E., and Zeleznik, J. M. (1994). Spatial distribution of water and chloride macropore flow in a well-structured soil. Soil Sci. Soc. Am. J . 58, 1294-1300. Rao, P. S. C., Green, R. E., Ahuja, L. R., and Davidson, J. M. (1976). Evaluation of a capillary bundle model for describing solute dispersion in aggregated soils. SuilSci. Soc. Am. J . 40,815820. Rao, P. S. C., Davidson, J. M., Jessup, R. E., and Selim, H. M. (1979). Evaluation of conceptual models for describing nonequilibrium adsorption-desorption of pesticides during steady-flow in soils. Soil Sci. Sue. Am. J. 43, 22-28. Rao. P. S. C., Jessup, R. E., Ralston. D. E.. Davidson, J. M., and Kikredse, D. P. (1980a). Experimental and mathematical description of nonadsorbed solute transfer by diffusion in spherical aggregate. Soil Sci. Soc. Am. J . 44, 684-688. Rao, P. S. C., Ralston. D. E., Jessup, R. E.. and Davidson, J. M. (1980b). Solute transport in aggregated porous media: Theoretical and experimental evaluation. Soil Sri. Soc. Am. J . 44, 1139-1 146. Rao, P. S. C., Jessup, R. E., and Addiscott, T. M. (1982). Experimental and theoretical aspects of solute diffusion in spherical and nonspherical aggregates. Soil Sci. 133, 342-349. Roseberg, R . J., and McCoy, E. L. (1990). Measurement of soil macropore air permeability. Soil Sci. Sue. Am. J . 54, 969-974. Sardin, M., Schweich, D., Leij, F. J., and van Genchten, M .Th. (1991). Modeling the nonequilibrium transport of linearly interacting solutes in porous media: A review. Water Resour. Res. 27, 2287-2307. Scotter, D. R., and Ross, P. J. (1994). The upper limit of solute dispersion and soil hydraulic properties. Soil Sci. Sue. Am. J . 58, 659-663. Selim H. M. (1992). Modeling the transport and retention of inorganics in soils. Adv. Agron. 47, 33 1-384. Selim, H . M., and Amdcher, M. C. (1988). A second-order kinetic approach for modeling solute retention and transport in soils. Wurer Rrsour. Res. 24, 2061 -2075. Selim, H. M., and Ma, L. (1995). Transport of reactive solutes in soils: A modified two-region approach. Soil Sci. Sue. Am. J . 59, 75-82. Selim, H. M.. Davidson, J. M., and Mansell. R. S. (1976). Evaluation of a two-site adsorptiondesorption model for describing solute transport in soils. In “Proceedings of the Summer Computer Simulation Conference Washington, DC, July 12- 14, 1976” pp. 444-448. Simulation Councils, La Jolla, CA. Selim. H. M.. Schulin, R., and Fliihler, H. (1987). Transport and ion exchange of calcium and magnesium in an aggregated soil. Soil Sci. Sue. Am. J . 51, 876-884.

APPROACHES TO SOLUTE TRANSPORT IN SOILS Selim, H. M., Buchter. B.. Hinz, C., and Ma, L. (1992). Modeling the transport and retention of cadmium in soils: Multireaction and multicomponent approaches. Soil Sci. Soc. Am. J. 56, 1004-1015. Shipitalo, M. J., Edwards, W. M.. Dick. W. A.. and Owens, L. B. (1990). Initial storm effects on macropore transport of surface applied chemicals in no-till soil. Soil Sci. Soc. Am. J. 54, 15301536. Singh, P., Kanwar. R. S., and Thompson, M. L. (1991). Macropore characterization for two tillage systems using resin impregnation technique. Soil Sci. Sue. Am. J. 55, 1674-1679. Skopp. J.. and Gardner, W. R. (1992). Miscible displacement: An interacting flow region model. Soil Sci. Soc. Am. J. 56, 1680-1686. Skopp. J., and Warrick. A. W. (1974). A two-phase model for the miscible displacement of reactive solutes in soils. Soil Sci. Soc. Am. J. 38, 545-550. Skopp. J., Gardner, W. R.. and Tyler, E. J. (19811. Solute niovement in structured soils: Two-region model with small interaction. Soil Sci. Soc. Am. J. 45, 837-842. Smettem, K. R. J . , and Kirkby. C. (1990). Measuring the hydraulic properties of a stable aggregated soil. J. Hvdrol. 117, 1-13. Sposito, G. (198 I ) . “The Thermodynamics of Soil Solutions.” Oxford University Press, New York. Steenhuis. T. S.. Parlange, J.-Y.. and Andreini, M. S. (1990). A numerical model for preferential solute movement in structured soils. Ceodermu 46, 193-208. Steenhuis. T. S.. Boll. J.. Shalit, G., Selker. J. S., and Merwin, I . A. (1994). A simple equation for predicting preferential flow solute concentrations. J. Giviron. Qual. 23, 1058- 1064. Toride, N., Leij, F. 3.. and van Genuchten, M. Th. (1993). A comprehensive set of analytical solution for nonequilibrium solute transport with first-order decay and zero-order production. Wurer Resorir. Res. 29, 2167-2182. Trojan. M. D.. and Linden, D. R. (1992). Microrelief and rainfall effects on water and solute movement in earthworm burrows. Soil Sci. Soc. Am. J. 56. 727-733. Turner. G. A. (1958). The flow structure in packed beds. Chem. Eng. Sci. 7, 156-165. Utermann. J., Kladivko, E. J., and Jury, W. A. (1990). Evaluating pesticide migration in tile-drained soil with transfer function model. J. Environ. QNuI. 19, 707-714. Valocchi, A. J. (1990). Use of temporal moment analysis to study reactive solute transport in aggregated porous media. Geodermu 46, 233-247. van Eijkcren, J. C. M., and Loch. I . P. G. (1984). Transport of cation solutes in sorbing porous medium. Wurer Resow. Res. 20, 7 14-71 8. van Genuchten, M. Th. (1980). A closed-form equation for predicting the hydraulic conductivity of unsaturated soils. Soil Sci. Sue. Am. J. 44, 892-898. van Genuchten. M. Th. (1985). A general approach for modeling solute transport in structured soils. I n “Proceedings of the 17th International Congress. IAH. Hydrogeology of Rocks of Low Permeability.“ January 7- 12. 1985, Tucson. AZ. IMem. I n / . A.s.soc. Hvdrogeol. 17, 512-5261 van Genuchten, M. Th., and Dalton, F. N. (1986). Models for simulating salt movement in aggregated field soils. Gerodcrma 38, 165-183. van Gcnuchten. M. Th.. and Wagenet. R. J. (1989). Two-siteitwo-region models for pesticide transport and degradation: Theoretical development and analytical solutions. Soil Sci. Sue. Am. J. 53, 1303-1310. van Genuchten, M. Th.. and Wierenga. P. J. (1976). Mass transfer studies in sorbing porous media: I . Analytical solutions. Soil Sci. Soc. Am. J. 40, 473-480. van Genuchten, M. Th., and Wierenga, P. J. (1977). Mass transfer studies in sorbing porous media. 11. Experimental cvaluation with Tritium (’H,O). Soil Sci. Soc. Am. J. 41, 272-277. van Genuchten. M. Th.. Wierenga, P. J.. and O’Connor. C. A. (1977). Mass transfer studies in sorbing porous media: 111. Experimental evaluation with 2.4.5-T. Soil Sci. Soc. Am. J. 41, 278284.

I50

L. MA AND H. M. SELIM

Wang, D., Norman, J. M., Lowery, B., and McSweeney, K. (1994). Nondestructive determination of hydrogeometrical characteristics of soil macropores. Soil Sci. Soc. Am. J. 58, 294-303. Ward, A. L., Kachanoski, R. G., and Elrick, D. E. (1994). Laboratory measurements of solute transport using time domain reflectometry. Soil Sci. Soc. Am. J . 58, 1031-1039. Warner, G . S . , Nieber, J. L., Moore, 1. D., and Geise, R. A. (1989). Characterizing macropores in soils by computed tomography. Soil Sci. Soc. Am. J. 53, 653-660. Watson, K. W., and Luxmoore, R. J. (1986). Estimating macroporosity in a forest watershed by use of a tension infiltrometer. Soil Sci. Sue. Am. J. 50, 578-582. Wilson, G. V., and Luxmoore, R. J. (1988). Infiltration, macroporosity, and mesoporosity distributions on two forested watersheds. Soil Sci. Suc. Am. J. 52, 329-335. Wilson, G. V., Jardine, P. M., and Gwo, J. P. (1992). Modeling the hydraulic properties of a multiregion soil. Soil Sci. Sor. Am. J. 56, 1731-1737. Yasuda, H., Berndtsson, R., Bahri, A., and Jinno, K. (1994). Plot-scale solute transport in a semiarid agricultural soil. Soil Sci. Soc. Am. J . 58, 1052-1060. Zimmerman, R. W., Chen, G., Hadgu, T., and Bodvarsson, G. S. (1993). A numerical dual-porosity model with semianalytical treatment of fracturdmatrix flow. Water Resour. Res. 29, 21272137.

SILICON MANAGEMENT AND SUSTAINABLE RICEPRODUCTION N. K. Savant', G. H. Snyder2, and L. E. D a m o F ' IStaSav International, Florence, Alabama 35630 ZDeparanents of Soil-Water Science and 'Plant Pathology, University of Florida, Everglades Research and Education Center, Belle Glade, Florida, 33430; 'Corresponding author

I. Introduction 11. Silicon Nutrition in Rice 111. Silicon in Soil and Water IV Silicon Management Agenda A. Si Fertilization B. Plant Si Recycling V. Potential Benefits of Silicon Management A. Agronomic Benefits B. Induced Resistance to Stress C. Increased Productivity of Problem Soils VI. Agronomic Essentiality of Silicon Management VII. Determining Need for Silicon Fertilization A. Soil Testing B. Plant Testing C. Si Availability in Si Sources VIII. Suggestions for Research IX.Summary References

I. INTRODUCTION An integrated approach of nutrient management will play an important role in sustainable rice (Oryza sariva L.) production, especially as the land used for crop production decreases and more areas become unavailable for farming. For properly exploiting yield potentials of high-yielding rice cultivars, agronomists generally consider improving the management of 13 essential elements, namely 6 macroelements (N, P, K, S , Ca, and Mg) and 7 microelements (Fe, Mn, Zn, B, 151 Advunces in Apnmny, Volrrmc F8

Copyright 0 1997 by Academic Press, Inc. All nghts of reproduction in any form resemed

152

N. K. SAVANT ET A L

Mo, C1, and Cu). These elements are essential because deficiency of any one of them adversely affects physiological plant function, resulting in abnormal growth and/or an incomplete life cycle. Rice is a known silicon (Si) accumulator (Takahashi et al., 1990), and the plant benefits from Si nutrition (Yoshida, 1975; Takahashi, 1995). Consequently, there is a definitive need to consider Si as an agronomically essential element for increasing and/or sustaining rice production (Takahashi and Miyake, 1977; Yoshida, 1981). Si is required for healthy and productive development of the rice plant (Lewin and Reimann, 1969; Yoshida, 1975). This element is absorbed by rice from the soil in large amounts that are several-fold greater than those of the other macronutrients. On an average, it is estimated that a rice crop producing a total grain yield of 5 t/ha will normally remove from 230 to 470 kg Si/ha (500-1000 kg SiO,/ha) from soil (Amarasiri and Perera, 1975; CRRI, 1976; IFA, 1992; N. K. Savant, unpublished data). Recent research and advances in physiology, biochemistry, and genetics suggest that Si interacts with other native or applied nutrients and has the potential to induce resistance/tolerance in the rice plant to biotic (insect pests and diseases) and abiotic stresses (Al, Fe, and Mn toxicity; salt injury; lodging, etc.) (Okuda and Takahashi, 1965; Yoshida, 1975; Epstein, 1994). Although agricultural soils are largely composed of silicate minerals, many soils contain an inadequate supply or are naturally low in plant available Si (intensity factor). Most likely, the Si content in some regions might be limiting to sustainable rice production. In addition, Si depletion can occur in traditional rice soils from the continuous monoculture of high-yielding cultivars with intensive cultivation practices (Kawaguchi, 1966; Miyake, 1993), especially if farmers are not replacing the Si removed by rice. Many soils in Africa, Asia, and Latin America are highly weathered and dedicated, and Si management may be important for increasing and sustaining rice productivity in these regions (D’Hoore and Coulter, 1977; Juo and Sanchez, 1986; Winslow, 1992; Friessen et al., 1994; Correa-Victoria et al., 1994). This chapter summarizes the past and current literature on Si nutrition of rice. A Si management agenda is presented, and its potential beneficial role in increasing and sustaining rice productivity in the future is discussed. A few suggestions for future research on Si are indicated that should help to meet a critical need for Si to increase rice yields on a sustained basis.

11. SILICON NUTRITION IN RICE The rice plant absorbs Si from soil solution in the form of monosilicic (monomeric) acid, also called orthosilicic acid [H4Si04] (Barber and Shone, 1966; Lewin and Reimann, 1969; Yoshida, 1975). The large amounts of Si absorbed by

SILICON MANAGEMENT

153

the rice plant cannot be fully explained by passive absorption (diffusion and mass flow). Rice is a Si accumulator (Takahashi et a / ., 1990); its mode of absorption seems to be actively linked with aerobic respiration, and the rice roots have a specific ability to concentrate Si from the external solution (Okuda and Takahashi, 1965; Tanaka and Park, 1966; Takahashi, 1995). A study conducted using metabolic inhibitors, such as NaF, NaCN, monoiodoacetate, and 2,4dinitrophenol, suggests that active uptake of Si by rice requires energy from aerobic respiration and also seems to be connected with anaerobic glycolysis (Okuda and Takahashi, 1962, 1965). According to van der Vorm (1980), two features are associated with active metabolic absorption of Si by rice, namely, a preferential absorption at low concentrations and an exclusion at high concentrations. After Si is absorbed by the plant as monosilicic acid, the absorbed water is lost through transpiration and the Si stays in the plant tissue. When Si concentration increases in the plant water, monosilicic acid polymerizes (Yoshida, 1975). The chemical nature of polymerized Si has been identified as Si gel (Yoshida et al., 1959; Yoshida, 1975) or biogenetic opal-as amorphous SiO,nH,O (Lanning, 1963). It seems to be mainly associated with polysaccharide fractions (cellulose and hemicellulose) and not with the lignin in the plant (Liu and Ho, 1960). Inanaga ef a / . (1995) provided evidence suggesting a potential role of Si in the formation of cross-links between lignin and carbohydrates via association with phenolic acids or aromatic rings. Of the polymerized Si within the plant, 87-99% exists as a very slightly soluble form in hulls, leaf blades, and leaf sheaths (Yoshida et af., 1962b). In these tissues, Si tends to be deposited as a 2.5-p thick layer in the space immediately beneath the thin (0.1 p) cuticle layer (Fig. 1). It has been proposed that the location and mechanical strength of this cuticle-,% double layer helps to maintain erect leaves, minimize transpiration, and protect the rice plant from fungal diseases and insect pests (see Section V). The reports on chemical studies suggest that generally Si is not very mobile in the rice plant (Yoshida, 1965, 1975), and therefore a continued supply of this element would be required during practically all growth stages for healthy and productive development of the plant. Although the active Si absorption by rice seems to start after the tillering stage (Kato and Owa, 1990) or after stem elongation (Chen, 1990), the most effective period for Si application for increasing yield seems to be during the reproductive stage in which its uptake and dry matter production are most vigorous (Fig. 2). Ultraviolet-B (UV-B) radiation seems to increase total leaf Si content in rice cultivars that are UV-B sensitive, but not in tolerant rice cultivars (IRRI, 1991). In a nutrient culture experiment, the application of Si (up to 200 ppm Si) generally decreased the uptake of nitrogen, potassium, iron, and manganese in transplanted rice and increased the uptake of phosphorus, calcium, magnesium and the formation of carbohydrate (Islam and Saha, 1969). In another greenhouse experiment, the application of 200 mg Si/kg soil (Hapludalf) as Na,SiO,; 5 H 2 0 decreased the translocation of

N. K. SAVANT ET AL.

154

Figure 1 Schematic representation of the rice leaf epidermal cell (Yoshida, 1975).

iron and increased that of manganese from rice (var. RT 42) straw to grain (Verma and Minhas, 1989). Some Russian studies suggest a possible role of Si in protein synthesis (Aleshin and Avakyan, 1983; Aleshin, 1988). Si is not yet considered an essential element for plant growth (Takahashi er al., 1990; Epstein, 1994). However, its deficiency has been reported to be associated with the following symptoms in the rice plant (Mitsui and Takatoh, 1963; Nishihara et al., 1970; Elawad and Green, 1979; Bergmann, 1992): (a) lower leaves becoming yellow or brown and necrotic, (b) poor tillering and retarded growth, (c) leaf tips wilting (probably due to impaired transpiration) and drying out, and (d) smaller panicles with increased sterility.

400

200

I Growth Stage V Re Ri O+Si -Si -Si I -Si +Si -Si -Si -Si +Si

m

100

Plant Height

Straw Weight

agure 2 Effect of silicon supply at different plant growth stages on the growth of rice. (Values relative to the control, i.e., silicon-free treatment during the whole growth period.) (Ma et al., 1989) *V, vegetative; Re, reproductive; Ri, ripening growth stages.

SILICON MANAGEMENT

155

Improves supply of carbohydrates

Silicon mechanical of culm

in soil solution

light penetration Into plant

Increases oxidation nutrient uptake

Plant uptake in large amounts in leaf blades, culrn and

Figure 3

Detoxication of reduced substances

Physiological role of silicon in the rice plant (Takahashi et a / . , 1990; Takahashi, 1995).

By and large, there is a general consensus that Si is a beneficial element for rice. The potential beneficial functions of Si are summarized in Fig. 3 (see Section V).

111. SILICON IN SOIL AND WATER Understanding the basic chemistry of Si in soil and water would be helpful for its effective management for increasing and/or sustaining rice yields. Si is abundant in the earth’s crust (constitutes up to 28%) and occurs in various complex mineral forms. From an agronomic management viewpoint, the main forms of Si in soil are dissolved Si mainly as monosilicic acid (H,SiO,) and polysilicic acids, including Si adsorbed on and precipitated with oxides of Al, Fe, and Mn and crystalline (well-ordered) and noncrystalline (amorphous or disordered) silicate minerals. Excellent review papers on Si forms, amounts and dynamics of their dissolution in soil are published elsewhere (McKeague and Cline, 1963b; Jones and Handreck, 1967; Lindsay, 1979; Hallmark et al., 1982; Drees et al., 1989). The solubility of soil Si minerals is variable and is influenced by temperature,

N. K. SAVANT ET AL.

156

pH, particle size, chemical composition, and the presence of disrupted layers. Their dissolution kinetics is affected by soil factors such as organic matter, water content, redox potential, and sesquioxides. Many investigators believe that sesquioxides apparently act as a soluble Si sink or source depending on the pH of the soil (Drees et al., 1989). The collection, analysis, and interpretation of soil solutions could be a powerful technique for investigating equilibrium and kinetics of Si dissolution in soils. Dahlgren (1993) compared five soil solution extraction procedures: high-speed centrifugation, immiscible displacement, column equilibrium/vacuum extraction, and 1:1 and 2: 1 soi1:water extraction. Si concentrations ranged from 222 pmol %/liter (column procedure) to 605 pmol %/liter (centrifuge procedure), indicating the widely varying effect of selected methods of soil solution extraction on Si concentration. Thermodynamically, the solubility of silicate minerals in terms of H,SiO, is expected to range from lO-*.74 M (amorphous Si) to 10-4 M (quartz) with soil Si having solubility corresponding to 10-3.'0 M (Lindsay, 1979). However, the Si concentration as H,Si04 in soil solution is generally 0.1-0.6 m M (Drees et al., 1989; Epstein, 1994), which is less than that in saturated monosilicic acid solution and is largely due to pHdependent adsorption of silicon by sesquioxides (Jones and Handreck, 1967; McKeague and Cline, 1963a,b). The pH dependency of the sorption of Si by sesquioxides is illustrated by the following equations (Drees er al., 1989): %(OH), [Si 0 (OH),]-

[Si 0 (OH),]-

+ Fe (OH),

+ H+

Fe (OH), OSi (OH),

(1)

+ OH-.

(2)

Insofar as wetland rice soils are concerned, soil Eh and organic matter are important factors that influence Si in soil solution (Ponnamperuma, 1965a; IRRI, 1992). Ponnamperuma (1965a) reported a marked decrease in soil Eh and concurrent increase in solubility of soil Si with submergence time. In one soil (pH 4.8; organic matter, 4.4%), the concentration of Si increased from 24 to 41 ppm in less than 50 days after submergence. This increase in Si has been attributed to its release from ferrisilica complexes under reducing soil conditions. In warm subhumid and humid tropical ecoregions (IRRI, 1993a), a high degree of weathering, mainly as dedication (Fig. 4), has resulted in the development of soils rich in iron and aluminum oxides and low in nutrient bases and Si (Takijima and Gunawardena, 1969; Juo and Sanchez, 1986). As a result of Si leaching, the soluble Si content of tropical soils, such as Ultisols and Oxisols, is generally 5-10 times less than in most temperate soils (McKeague and Cline, 1963b; Juo and Sanchez, 1986, Foy, 1992). This might be one of the unidentified causes of lower rice productivity of many tropicallsubtropical soils compared with that of temperate soils. Present trends toward intensive rice cultivation by growing more than one crop of fertilizer-responsive high-yielding cultivars in a year may result in a decrease in plant available Si in these soils (Miyake, 1993).

SILICON MANAGEMENT

157

I Mollisols

Weathering] Less

Feldspars, Vermiculites

1

SiO,

Vertisols

Smectites

si4

h

lnceptisols

Smectites, Kaolinites SiO,

Kaolinites. Smectites

Alfisols

Ultisols

Kaolinites. Sesquioxides

Oxisols

SiO, Sesquioxides, Kaolinites

h

M&e

Figure 4 Simplified acid weathering sequence in soils (Friesen et a / ., 1994).

This possible decreasing trend in plant available Si could be due to (a) low solubility and/or slow dissolution kinetics of soil Si (Lindsay, 1979; Drees et a l., 1989), (b) high uptake of Si by rice crops (Fig. 51,and (c) limited or no concerted

m f

250

g 200 Y Q

150 3 +

;.

2

100 50

0

Si

K

N

P

Ca

Mg

S

Fe

Mn

Figure 5 Estimated nutrient uptake (kg/ha) by a rice crop (HYV)producing 5 tlha grain yield (IFA, 1992).

N. K. SAVANT ET AL.

158

minerals (crystalline noncr stalline

I hydroxides

Figure 6 Main transforrnations/processes influencing silicon concentration in soil solution.

attempts by farmers to recycle Si in rice crop xesidues and/or to apply balanced fertilizers that include Si sources. Irrigation water from rivers, wells, and runoffs collected in catchment areas (dams) may supply some Si to rice crops. The mean dissolved Si concentration of most river waters seems to vary from 4.7 to 16.4 ppm Si (McKeague and Cline, 1963b). Well water from the state of Kerala in India contained less Si (2.4-3.2 ppm Si) than irrigation water from a dam (5.6 ppm Si) (Nair and Aiyer, 1968). According to Sadzawka and Aomine (1977), Si in river water of central Chile was related to the chemical nature of rocks in the catchment area and was relatively high in the rivers in the volcanic ash area. A summary of the main reactions/transformations influencing Si concentration in soil solution is shown in Fig. 6.

N.SILICON MANAGEMENT AGENDA Rice requires large amounts of Si for healthy plant growth and development (Lewin and Reimann, 1969; Ishizuka, 1973; Yoshida, 1975). For example, under the warm subhumid tropical conditions of India, Si removed by 12 rice cultivars

159

SILICON MANAGEMENT

(90-140 days duration) grown on an Inceptisol during the dry season varied from 205 to 61 1 kg Si/ha when grain yields ranged from 4.6 to 8.4 t/ha (Nayar et al., 1982). Because of the high Si requirement, rice generally responds well to applied Si sources. In the Japanese literature, Si has been regarded as an agronomically essential elemenr (Takahashi et a l . , 1990); consequently, management requires special attention in sustainable rice production. The agenda for Si management in rice cultivation includes Si fertilization and plant Si recycling.

A. SI FERTILIZATION 1. Sources of Si

Wollastonite ore [native calcium metasilicate, CaSiO, (24% Si, 34.5% Ca), and different kinds of slags from iron, ferronickel, and manganese ore smelters have been used as Si sources for rice (IRRI, 1978b). These sources vary in their composition (Table I), and their Si-supplying value depends on relative amounts of their basic constituents (soluble in 0.5 M HCl at 30°C) such as dicalcium silicate, monocalcium silicate, and moticellite (Lian, 1976). In field experiments, Korean scientists have used, besides wollastonite, different slags such as slowly cooled slag (13% Si,) air-cooled slag (12.6% Si), quenched slag (16% Si), and fused phosphate (1 1.5% Si) (Kim er al., 1985; Lee ef al., 1985). Electric furnace calcium silicate slags produced as a by-product of elemental phosphorus production and cement have also been used as Si sources (Elawad and Green, 1979; Snyder et a l . , 1986). Since 1990, Chinese researchers have been testing a Table I Chemical Composition of Slags Chemical composition (%) Type of slag

Country

High furnace High furnace Electric furnace Iron industry

Japan Taiwan Japan Japan

Iron industry

Japan

Electric furnace

United States

K Ca Si 15.6 14.8 15.0 12.1

Reference

Mg

Mn

Fe

A1

28.5 30.2 35.7 28.6

3.7 0.6 4.0 3.6

0.7 0.8 0.8

0.6 0.8 3.6 0.8

11.8 17.1 3.9

-

Lian (1976) Ota (1964) Takijima er

-

Ota ( 1964)

al. (1970) 9.4

1.7

17.9

4.2

7.7

2.1

-

Takijima et

22.0

0.1

33.0

0.3

0.2

TR

-

Snyder et al.

al. (1970)

(1986)

Note. -, Not available; TR, traces

160

N. K. SAVANT ET AL.

new silicate fertilizer containing 230 g water-soluble Si/kg fertilizer for rice with promising results (Liang et al., 1994). There are other materials that contain relatively small amounts of Si and may have minor importance as Si sources. A sample of lignite fly ash from an Indian thermal power plant containing 23% Si (141 ppm available Si) has been evaluated as a Si source for rice (Raghupathy, 1993). Rhenania phosphate prepared by Kali-Chemie Company of Germany contained 4.2% soluble Si and has been tested as a source of Si for rice with good results (Anonymous, 1976).

2. Rate of Application The rate of application of the Si source may depend on its chemical and physical nature and soil factors. Results of several field experiments suggest that, in general, 1.5-2.0 t/ha of calcium silicate slag may be adequate for lowland rice grown in Japan, Korea, and Taiwan (Kono, 1969; Lian, 1976; DeDatta, 1981). Based on the direct and residual effects of silicate slag applications on rice grown on acid soils of Taiwan (a 5-year experiment with 10 rice crops), a rate of 2 t slag/ha to the first crop every 2 years is recommended (Lian, 1992). Recent results indicate a possibility of reducing rates of silicate slag added by using finer-grade materials (100% less than 0.15 mm) (Datnoff et al., 1992).

3. Time and Method of Application About two-thirds of Si in the whole rice plant and nearly three-fourths of Si in the leaf blades are absorbed during the reproductive stage (Ma et al., 1989). According to another study, the absorption of Si starts from transplanting and continues to flowering (CRRI, 1976). Japanese workers found that basal application of Si fertilizer and topdressing to tillering stage were equally effective (Ota, 1964; Kono, 1969). Similar results for broadcasting and topdressing silicate sources have been reported on Taiwan soils (Su, 1972). As far as broadcast application of silicates is concerned, Chiu and Huang (1971) observed broadcasting to be somewhat better than banding in the furrow. Potassium silicate fertilizers developed in Japan are reported to have slow-release properties that make them more suitable for broadcast application in rice, and their agronomic performance is encouraging (Kubota, 1984; Tokunaga, 1991).

4. Reactions of Silicate Fertilizers in Soil Bioavailability of Si in silicate slag is influenced by its dissolution and sorption reactions in soil. Kato and Owa (1990) conducted a series of pot experiments and elucidated the effect of PCO, on dissolution of silicate slags applied to lowland rice soils in the presence of plants. They found that, at the early plant growth

SILICON MANAGEMENT

161

stage, the addition of slag increased pH and Ca*+ concentration in the soil, which in turn reduced the solubility of the slag. However, after the tillering stage a significant amount of C02 produced due to root respiration tended to reduce soil pH, and probably CaZ+ concentration, and therefore seemed to enhance solubility of the added slag. Russian workers investigated the changes in concentrations of monosilicic and polysilicic acids in soil solution as influenced by additions of amorphous Si and proposed the following phases (Dr. V. V. Matichenkov, Institute of Soil Science and Photosynthesis, Russian Academy of Science, Moscow Region, Pushchino, Russia, 1993, personal communication): Phase a-With a gradual initial addition of amorphous Si, concentration of monosilicic acid increases while that of polysilicic acids remains practically unchanged. Phase b-New polysilicic acids are formed with some decrease in monosilicic acid concentration. Phase c-With the further additions of Si, there is a gradual increase in the amounts of monosilicic and polysilicic acids. The relevance of these phases in soil systems to Si uptake by rice plants needs to be investigated. Most rice farmers of south and southeast Asian countries are small scale and may not be able to afford regular application of Si fertilizers at 1 or 2 t/ha to their small rice fields in the future mainly because of their high cost and limited availability. There is a need to develop efficient fertilization practices that will use lower and affordable rates of Si source without sacrificing their agronomic benefits. In this regard, better understanding of the dissolution kinetics and overall fate of applied silicates in soil would be very useful.

B. PLANTSI RECYCLING 1. Plant Si Sources

For recycling plant Si, farmers can use rice straw and rice hull. In 1993, world production of unmilled rice grain was 519 million tons (FAO, 1993). Assuming an average 1:I .2 grain:straw ratio and 4% Si in rice straw, 623 million tons of rice straw containing roughly 25 million tons of Si was available for recycling. Similarly, assuming an average of about 20% rice hull in unmilled rice and nearly 8% Si in rice hull, probably 104 million tons of rice hull containing 8 million tons of Si was available for recycling. Despite the nutrient value of these plant residues, proper recycling of rice straw and rice hull is not common among most rice farmers of developing countries in Asia probably for the following reasons (Ponnamperuma, 1984): bulkiness of the material; additional labor requirement

N. K. SAVANT ET AL.

162

for recycling (mainly for transport and spreading); difficulty in incorporating uncut straw; insect, disease, and rat problems associated with rice straw; low benefit:cost ratio; and the demand for nonagricultural uses. The Si content of rice straw (in leaf blades where transpiration occurs) and rice hull is relatively high compared with other parts of the rice plant (Table 11). Si content of rice straw shows large variations, from 1.7 to 9.3% Si (Yoshida, 1978), and is influenced by several factors such as soil, irrigation water quality, amounts of fertilizers applied, rice cultivars, and season (Ponnamperuma, 1984). For practical purposes in farmers’ fields in the tropics, the rice straw corresponding to l ton of sun-dried paddy is about 1.5 tons and may contain about 70 kg Si (or about 150 kg SiO,). Rice absorbs Si in the form of monosilicic acid along with water (Yoshida, 1975; Takahashi, 1978). However, 99% of the total Si content in rice straw is in the form of polymerized silicic acids that are difficult to solubilize, with less than 1% ionic plus colloidal Si (Yoshida et al., 1962b; Yoshida, 1975). Zhang (1984) could not extract Si from straw with acetate buffer at pH 4, but could extract some with dilute NaOH, probably because of the polymerized nature of Si and its association mainly with the polysaccharide fraction (Liu and Ho, 1960). Ma and Takahashi (1989, 1991b) observed its slow release in flooded soil during a short period. Under tropical field conditions, with floodwater temperature ranging from 27 to 40°C and soil temperature at 1- to 10-cm depth ranging from 24 to 39°C (dry season) in Thailand, Snitwongse et al. (1988) reported initial faster decomposition of ‘%-labeled rice straw during the first 4 weeks in planted flooded soil. However, they did not monitor the release of Si from the added straw. The incorporation of 5.0 t/ha of rice straw for four seasons at a location in Table 11

Summary of Observations on Silicon Deposition in the Rice Planta Silicon content Plant part

(% s i ) b

Observations on silicon deposition

Hull

7. I

Leaf blade

5.6

Leaf sheath

4.7

Stem

2.4

Root

0.9

Heaviest deposition in the interspace between the cuticle and the epidermal cells Same as above (see Fig. 1) and heavy deposition in the vascular bundle plus the bundle sheath and sclerenchyma Heavy deposition in the epidermis and along the cell walls in the parenchyma Deposition in the epidermis, sclerenchyma plus bundle sheath, and the cell walls in the parenchyma No localization, widely distributed in all tissue ~~

______________

From Yoshida et a!., 1962a,b; Yoshida, 1975. Estimated average.

SILICON MANAGEMENT

163

Java, Indonesia, has increased Si from 50 ppm in soil without straw to 150 ppm in a straw-amended lateritic paddy soil (probably an Oxisol) (Diamond et al., 1986). These observations suggest that, for flooded fields, Si in straw may be released much faster and in relatively larger amounts under tropical conditions than under temperate conditions. More laboratory and field research is needed for better understanding of the dynamics of Si release during decomposition of rice straw in soil as influenced by temperature, water, and time and method of straw incorporation.

2. Plant Si Cycling In the past, rice straw has been incorporated into soils mainly as a source of C and/or N, and at times for the improvement of soil physical conditions. Although it contains about 4.0% Si, practically no reports are seen in the literature on its recycling as a source of Si in the subtropical and tropical rice culture. In a mechanical rice cultivation, the crop is harvested by combine, and chopped straw is spread on the land. It is then incorporated into the soil by disking or plowing, thus recycling nearly all Si in the straw. However, in many south and southeast Asian countries where rice crops are harvested manually, only stubbles can be plowed in and the amount of straw recycled depends on the manner and height at which the crop is harvested. For example, in rain-fed regions of Indonesia, only panicles are harvested; the straw is slashed, heaped, or spread in the field and then plowed in wet at the beginning of the next season (Ismunadji, 1978). This practice has the potential to recycle nearly 50% of the Si removed by the previous crop. In the Philippines, most farmers generally harvest the stalks 60-90 cm from the base of the panicles (DeDatta, 1981). This practice also leaves a lot of straw in the fields and helps considerably in recycling of Si. In India, however, rice straw has many off-farm uses. Because harvesting is done almost at ground level, practically no stubbles are left in the fields for recycling of plant Si. Rice straw hauled away from rice fields and used for various purposes, such as animal feed/bedding, biogas production, or mushroom cultivation, may retain its nutrient value as a source of Si; thus, the end-products of these uses should be recycled. Rice compost prepared by aerobic or anaerobic process could be a good source of Si because the plant Si associated with cellulose and hemicelluose (Liu and Ho, 1960) would be released as they degrade during composting. Cellulolytic fungi, namely Trichoderma harzianum (Cuevas, 1993) or Trichoderma reesei and Pieurotus sajor-caju (Kanotra and Mathur, 1994), could be used for rapid composting of straw and may be of some advantage for recycling plant Si. In general, composting of rice straw offers a potential way of recycling plant Si because it reduces the bulk of straw to be handled. Rice farmers who have few animals can use straw for their bedding, then save the bedding and recycle it as

N. K. SAVANT ET AL.

164

such or after composting. The straw bedding used for the cultivation of ricestraw mushrooms or oyster mushrooms may also have good recycle value as a source of Si, and this practice merits investigation. In Australia, France, Indonesia, Italy, Malaysia, Myanmar (Burma), the Philippines, Spain, Thailand, and in California (United States), rice straw is generally burned (Tanaka, 1973, 1978). It is a minor practice in Japan, India, and Sri Lanka. The burning of straw does not reduce the recycle value of plant Si provided it produces a black to gray ash (up to 600°C) containing amorphous Si; pink-white ash produced at 1000°C has no recycle value because the Si in it is crystallized mainly as crystabolite (Fig. 7). Rice hull has some value as a soil amendment and nutrient source, but its use as a Si source has been very limited. Hull bedding used in poultry production might be a good source of Si, and information is needed on the potential of recycling this material. Incorporation of rice hull into soil normally occurs near rice mills and/or their disposal sites. Past research efforts have been largely confined to demonstrating that the hull is not harmful to the soil but is slightly beneficial as a fertilizer (IRRI, 1966; Shanna et al., 1988; Sahu, 1990). Traditionally, some rice farmers apply rice hull and/or its ash mainly to seedbeds, but proper information on this practice of recycling of plant Si is needed. Recently, it has been shown that the application of black to gray rice hull ash (32.9-40% Si) at 0.5- 1.O kg/m* to seedbed produced healthy and strong rice seedlings (Sawant et al., 1994). In order to facilitate the collection of black to gray rice hull ash for

12000

10000-0 C

002

8000.

L

(u

2

6000

c

c)

3 0

0

y

1ooo*c

4000-

2000-

600°C

0.

10

I

30

I 40

I

50

I

60

Degrees 2-theta EFgure 7 Effect of the ashing temperature on XRD pattern of rice hull ash (N. K . Savant, unpublished results).

SILICON MANAGEMENT

165

the preparation of healthy rice seedlings, farmers can use hull-fired stoves (in place of wood-burning stoves), namely the Vietnamese Lo Trau (Tran, 1991), IRRI’s fpu-Qalun (IRRI, 1993b, 1994a), the Indonesian Porkurn (RT 2) (Rachmat et al., 1991), or the Maligya cooking stove (PhilRice, 1994), for daily domestic cooking. The Si supply through irrigation water at times may be substantial (Amarasiri, 1973) and may influence Si management of irrigated rice. Kobayashi and associates (Kobayashi et al., 1955, Kobayashi, 1957; Kobayashi and Shinagawa, 1957) demonstrated a remarkable efficacy of irrigation for the increase of crop yields in Ando soils in Kyushu. They attributed this to the soil ameliorating effect of irrigation water, partly due to the accumulation of Si and partly due to increase in Si:AI ratio in soil colloids and the decline of phosphate sorption and exchangeable Al. The soluble Si content of river water tends to be high in the catchment area with allophanic colloids and of volcanic origin (Sadzawka and Aomine, 1977). When used for irrigation, such water may benefit the rice crop by supplying Si and eventually increasing and/or sustaining yields. This also implies that rainwater in catchment areas with Si-deficient highly weathered soils, such as Ultisols and Oxisols, may supply small or negligible amounts of Si to rain-fed rice. Under such agroclimatic conditions, Si management (Si fertilization or plant Si cycling) would be more important for sustaining high yields over a long time.

V. POTENTIAL BENEFITS OF SILICON MANAGEMENT The effective management of Si can offer several potential benefits for rice cultivation including agronomic benefits, induced resistance to stress (abiotic and biotic), and increased productivity of problem soils.

A. AGRONOMIC BENEFITS 1. Improved Plant Growth

From several studies, it appears that Si nutrition has direct and indirect beneficial effects on the rice growth largely due to its unique physiological role (Okuda and Takahashi, 1965; Yoshida, 1975; Takahashi er a f . , 1990; also see Fig. 3). In the tropics, excessive vegetative growth of rice can cause low yields. In such an environment, the modifying positive effect of an adequate supply of Si on leaf erectness can be beneficial, especially when rice plant density is high and low light intensity is likely to limit photosynthesis (Yoshida et uf., 1969; Kang,

166

N. K. SAVANT ET AL.

1985). According to Agarie et al. (1992), the maintenance of photosynthetic activity due to Si fertilization could be one of the reasons for the increased dry matter production. They also observed an increase in water use efficiency in Siamended rice plants probably due to prevention of excessive transpiration. Si fertilization benefits rice plants in the nursery (at seedling stage) as well as in the field after transplanting. In field experiments, various slags applied to nursery plants increased the number of leaves and dry matter produced by rice plants (Lee et al., 1985). The application of black to gray ash of rice hulls (at 0.5-2.0 kg/m2) made rice seedlings healthy and strong and increased their biomass (Sawant et al., 1994; Savant and Sawant, 1995). After transplanting, Si fertilization increases the number of tillers and panicles (IRRI, 1965; Kim ef al., 1985; Liang et al., 1994). A similar beneficial effect of applied Si on tiller number and grain filling has been reported by Burbey et al. (1988) in upland rice (var. Danaubawah) grown on an Ultisol in West Sumatra. Poor tillering and wilting of leaf tips due to an impaired transpiration in Si-deficient rice plants have been reported by Bergmann (1992). The effect of Si supply on the growth of rice plants seems to be most remarkable during the reproductive growth stage (Ma et al., 1989). Si also has a positive effect on the number of spikelets on secondary branches of panicles and the ripening of grains (Seo and Ota, 1983; Lee et al., 1990). Apparently, Si plays an important role in hull formation, which in turn seems to influence grain quality. The hulls of poor-quality milky-white grains (kernels) are generally low in Si content, which is directly proportional to the Si concentration in the straw (Mizuno, 1987). Rice plants suffering from certain nutritional disorders, such as Akiochi or Bronzing, are often low in Si, which may not be a direct cause of this disorder. However, in Japan, 1-3 tons of slag per hectare is generally recommended for alleviation of such nutritional disorders (Tanaka and Yoshida, 1970).

2. Increased Yield Since 1955, Japanese farmers have increased and sustained average rice yields up to 6 tiha (IRRI, 1993a). This could be due to adoption of balanced integrated nutrient management that includes Si fertilization. Silicate slag application at an optimum rate of 1.5-2.0 t/ha is now widely used in degraded paddy fields and peaty paddy fields in Japan (Kono, 1969; Takahashi and Miyake, 1977). Yield increases of 10% are common when Si is added and at times exceed 30% when leaf blast is severe (Yoshida, 1981). Lian (1976) and Elawad and Green (1979) have published excellent review papers on rice yield responses to Si fertilization, mostly in temperate countries such as Japan, Korea, and Taiwan. Therefore, the results of Si fertilization in rice under subtropical to tropical conditions are mainly reviewed herein. It has been reported that the Si supply from paddy soil and its uptake by rice

167

SILICON MANAGEMENT

plants are temperature-dependent (Sumida and Ohyama, 1990; Sumida, 1992b). Therefore, the need for proper Si management to increase and/or to sustain rice productivity appears to be greater in temperate countries than in tropical countries. An acid weathering sequence of soils suggests that, due to dedication, soils (minerals) lose Si as a result of leaching (see Fig. 4). Subtropical and tropical soils (Inceptisols, Alfisols, Ultisols, and Oxisols) in subhumid to humid climates are generally low in plant-available Si and would benefit from Si fertilization (Kawaguchi, 1966; Takijima and Gunawardena, 1969; Tanaka and Yoshida, 1970; Juo and Sanchez, 1986; Foy, 1992). In 1963, IRRI scientists conducted a field experiment during the dry season to investigate the effect of Si fertilization on yields of two rice varieties (Chianung and Peta). The Si applied basally at 47 kg/ha as calcium magnesium silicate significantly increased grain yields by more than 500 kglha. This surprised the scientists because the Maahas clay was believed to have sufficient Si (IRRI, 1964). Since then, IRRI continued investigations on the effects of different forms of Si sources, rates, and methods of application on agronomic and entomological aspects of flooded rice, and with some exceptions obtained encouraging results (IRRI, 1965, 1966, 1967). For example, in a 1966 dry season experiment on a farmer's field [a Louisiana clay, a humic latsol, pH 5.1 (probably an Oxisol)], with calcium silicate (18.8% Si) application, grain yield of IR 8 was significantly greater than that from nitrogen treatments only (Fig. 8). Increases in flooded rice yields with Si fertilization have been reported in Sri Lanka (Rodrigo, 1964; Takijima et al., 1970), Thailand (Takahashi et al., 1980), Indonesia (Burbey et al., 1988), India (Datta et al., 1962; Sadanandan and Verghese 1968; Subramanian and Gopalswamy, 19901, China (Ho et al., 1980;

fi

+

IL '

"0

+ 1

115

235 352 470 Si Applied (kglha)

705

940

Figure 8 Effect of silicon rates on yield of IR 8 grown on a humic latsol, 1966 dry season (IRRI, 1967).

168

N. K. SAVANT ET AL.

Liang et al., 1994), and Florida (Snyder et al., 1986; Datnoff et al., 1991, 1992). In Thailand, Takahashi et hl. (1980) reported particularly striking rice yield responses to Si application especially when application rates of other conventional fertilizers were rather high. Snyder et al. (1986) showed that calcium silicate application increased rice yield on Histosols mainly due to the supply of plantavailable Si and not due to supply of other nutrients. The results of field trials on rice soils with different levels of available Si in Jinhua, Zhejiang, (South China) suggest a synergistic effect of added N on performance of Si fertilizer (Ho et al., 1980). In a long-term trial in Taiwan, rice yield responses were obtained by Si additions even after continued application and available Si exceeded the critical level for Si deficiency (Su et al., 1983). At times, Lian (1988) observed responses of rice to added Si higher than those to added P and K and concluded that repeated Si applications are required to maintain this effect. In field trials conducted on 10,800 ha during 1990-1993, Liang et al. (1994) reported additional rice yields from 4.6 to 20.7% with an average increase of 10% (P < 0.01) due to the basal application of a new silicate fertilizer (containing 230 g water-soluble Si/kg) to calcareous soils (up to 8.9% CaCO,). Beneficial effects of applied Si on yields of upland rice have been observed in the Philippines (Garrity et al., 1990), Colombia (Correa-Victoria et al., 1994; Friesen et al., 1994), China (Liang et af., 1994), and West Africa (Yamauchi and Winslow, 1989; Winslow, 1992). Garrity et al. (1990) found that Si application as rice hulls (9.4% Si) or its ash (44.4%Si) increased yield of rice on an Ultisol with very high P-fixing capacity at Cavinti, Laguna, Philippines. An application of 18.7 g Si/m2 as sodium metasilicate doubled plant Si concentration and increased grain yield of upland rice by 48% (Winslow, 1992). Recent results of field studies clearly indicated that Si deficiency is a major soil nutrient constraint limiting upland rice yields by about 40% (600-900 kg/ha) on highly weathered savanna soils in Colombia (Friesen et al., 1994). Increased flooded rice yields have been reported due to recycling of Si in rice hulls and straw andlor. their ash in the Philippines (IRRI, 1966), Sri Lanka (Amarasiri and Wickramasinghe, 1977; Amarasiri, 1978), Madagascar (IRRI, 1991), India (Sharma et al., 1988; Sahu, 1990), and Thailand (Ditsathaporn et al., 1989). Field experiments conducted with IR 8 at Polonnaruwa in Sri Lanka have shown that rice hull ash application of 0.74 t/ha with the recommended level of fertilizers (69:20: 18 kg NPK/ha) yielded an additional 1.O- 1.4 t/ha, whereas further increase in ash additions decreased the yield (Amarasiri, 1978). Amendment of a saline-acid soil by rice hull ash increased rice yield only in the first year of application (Ditsathaporn et al., 1989).

3. Positive Interactions with Applied N, P, and K Fertilizers a. Nitrogen Fertilizers Rice straw biomass generally increases with an increase in N rate, whereas

169

SILICON MANAGEMENT

brown rice yield does not always do the same. In many cases, yield decreases when N rates are more than optimum. Due to a synergistic effect, the application of Si has the potential to raise the optimum N rate, thus enhancing productivity of existing lowland rice fields (Kono, 1969; Elawad and Green, 1979; Ho et a l . , 1980). This tendency to increase grain yields due to N application in the presence of Si fertilization is illustrated in Fig. 9. Application of N tends to decrease Si uptake in rice, and fertilizers containing NHi-N may decrease it more than NOF-N (Kono and Takahashi, 1958; Wallace, 1989). At the earlier growth stages, high amounts of NHJ-N in soil resulting from high N rates may limit Si uptake by the rice plants. At later growth stages, Si management for response of rice to Si application may be more important because Si uptake is mainly dependent on Si-supplying ability of the soil (Sumida, 1992a,b). Fertilizing with N tends to make rice leaves droopy, whereas Si

8.0 -

Japan (Ota, 1964)

'

7.0 -

6.0

m

E

.-

$

5.0

.

Taiwan (Lian, 1976)

P N , 120 kglha

6, 60 kglha

N, 0 kglha

2.3

2.8

3.2

3.7

4.2

Si content of straw (YO) Figure 9 Relation between silicon content of straw and grain yield as influenced by rates of N application.

170

N. K. SAVANT ET AL.

keeps them erect. By adopting proper Si management, erect leaves can easily account for a 10%increase in the photosynthesis of the canopy and consequently a similar increase in yield (Yoshida et al., 1969; Yoshida, 1981). Therefore, the maintenance of erect leaves by proper Si fertilization for higher photosynthetic efficiency becomes more important when rice is grown with liberal applications of N fertilizers in lowland rice fields having highly weathered tropical soils with low Si-supplying capacity (Yoshida et al., 1969). b. Phosphate Fertilizers Reports on the effects of Si on P availability to plants are encouraging, but explanations for this effect vary. With no P applied as a fertilizer, Si fertilization increased the P content of the rice straw and grain during the 1964 wet and 1965 dry seasons (IRRI, 1965, 1966). This increased P content has been attributed to better availability of soil P and/or enhanced mobility of P from the roots to the stems. Savant and Sawant (1995) also observed increased P content in rice seedlings (var. Jaya) when black-gray ash of rice hull at 0.5-2.0 kg/m* was applied to rice seedbeds. The application of calcium silicate to highly weathered savanna soils enhanced upland rice response to applied phosphate (Friesen et al., 1994). Under low P-adsorbing conditions, application of Si has been found to reduce the P requirement, but on highly weathered soils the results have been more variable and less promising (Blair er al., 1990). The efficiency of phosphate fertilizer seemed to be enhanced when it was applied along with Si. Fertilizer P absorbed by the rice crop increased from 26 to 34% when P as single superphosphate (at 26 kg/ha) was applied along with a silicate fertilizer (IARI, 1988). Results of laboratory experiments suggest that the Si in Rhenania phosphate (thermally treated phosphate rock and Si mixture) decreased the P-retention capacity of soil and thus increased the level of watersoluble P in the soil (Chien, 1978). Others have reported a decreasing effect of silicate on P-retention capacity of soil, thus helping to increase the level of watersoluble P (intensity factor) in soil (Silva, 1971; Roy er al., 1971; Adams, 1980; Syouji, 1981; Kundu et al., 1988; Subramanian and Gopalswamy, 1991). Some Russian work suggests that the addition of highly mobile Si compounds increased mobile phosphates from 30 to 80%in podzolic, chestnut, and chernozem soils (Dr. V. V. Matichenkov, Institute of Soil Science and Photosynthesis, Russian Academy of Science, 1994, personal communication). However, recent results of laboratory and greenhouse experiments conducted by Ma and Takahashi (1990a,b,c, 1991a) suggest that silicic acid does not increase P availability in soil. According to them, the overall beneficial effect of Si may be attributed to a higher P:Mn ratio in the shoot due to the decreased Mn and Fe uptake, and thus indirectly to improved P utilization within the rice plants. They have concluded that the interaction between P and Si is indirect in P-deficient soil.

SILICON MANAGEMENT

171

c. Potassium Fertilizers Interactions of applied K and Si in soil seem to have beneficial effects on rice yields. Si application (140 and 280 kg %/ha) increased upland rice yield response to applied K on a Ultisol in West Sumatra (Burbey ef al., 1988). Ota ( 1988) studied the effect of application of K and Si at the spikelet-differentiation stage and observed an increase in the number of spikelets/m*, percentage of ripened grains, and 1000-grain weight. Silicification of cell walls seems to be linked with K nutrition. According to Nogushi and Sugawara (1966), K deficiency reduces the accumulation of Si in the epidermal cells of the leaf blades, thus increasing the susceptibility of the plant to rice blast. Burbey et af. (1988) reported decreased neck blast incidence in upland rice on a Ultisol due to the possible K X Si interaction. Low soil moisture and high humidity are environmental conditions that are common in upland rice regions. These environmental conditions can reduce Si and K absorption by rice plants (Baba et af., 1956) and may decrease their ability to resist biotic and abiotic stresses. Therefore, Si management integrated with K may be more important for sustaining rice yields in upland areas than in lowland areas.

B. INDUCED RESISTANCE TO STRESS An adequate supply of Si has been reported to alleviate both biotic and abiotic stresses in rice (Okuda and Takahashi, 1965; Yoshida, 1975; Epstein, 1994).

1. Biotic Stresses Si has been shown to suppress fungal diseases such as rice blast, brown spot, leaf scald, sheath blight, and grain discoloration (Table 111). Osuna-Canizales et af. (199 I ) have clearly demonstrated that the Si concentration of 1.4 mourn3 in nutrient solution dramatically reduced incidence of blast in rice (cultivars IR 36 and 50). In field trials, Datnoff et af. (1991) reported the reduction in blast ranging from 17 to 30% and in brown spot from 15 to 32% in rice grown on Histosols in the subtropical climate of Florida. The finer grades of a calcium silicate slag used were associated with lower severities of blast and brown spot because of the increase in Si uptake associated with the reduced particle size of the slag (Datnoff et a l . , 1992). For example, blast severity was 20.8% with the application of a fine-grade Si material (100% <0.15 mm) in comparison to 31.8% for the standard-grade Si material (90% c2.36 mm) or 41.4% for the nonamended control. Disease severity or incidence tends to be reduced with increasing tissue concentration of Si (Datnoff et al., 1990a, 1991, 1992; OsunaCanizales et af., 1991). It seems that the residual effects of Si fertilization will

N. K. SAVANT ET AL.

172

Table 111 Rice Diseases Suppressed by Improved Silicon Nutrition Disease

Pathogena

Reference

Leaf and neck blast

Magnaporfha grisea (Pyricularia grisea or Pyricularia oryzae)

Suzuki (1935a,b), Volk el al. (1958), Kozaka (1965). Kim et al. (1986), Kim and Lee (1982). Aleshin er al. (1987). Yamauchi and Winslow (1987). Datnoff er al. (1991, 1992), Osuna-Canizales et al. (1991). Winslow (1992). Correa-Victoria er a1 (1994), Kumbhar et al. (1995). Seebold et af. (1995)

Brown spot

Cochliobolus miyabeanus (Bipolaris oryzae)

Takahashi (1967). Ohata ef al. (19721, Yamauchi and Winslow (1989), Datnoff et al. (1991, 1992). Lee er al. (1981). Nanda and Gangopadhyay (1984)

Sheath blight

Thanatephorus cucumeris (Rhizoctonia solani)

Elawad and Green (1979), Datnoff er al. (1990a)

Leaf scald

Monographella albescens (Gerlachia oryzae)

Elawad and Green (1979). Yamauchi and Winslow (1987). Winslow (1992), Correa-Victoria et al. (1994)

Grain discoloration

Bipolaris, Fusarium, Epicoccum, etc.

Yamauchi and Winslow (1987), Winslow (1992), Correa-Victoria et al. (1994)

Stem rot

Magnaporihe salvanii (Sclererotiumoryzae)

Elwad and Green (1979)

a Pathogens in parentheses are the anamorph (imperfect state), whereas the others are the teleomorph (perfect state) of the fungus.

also reduce the disease intensities of blast and brown spot, although recent applications of Si are more effective (Datnoff el al., 1991; Correa-Victoria el al., 1994). For example, in 1988, experiments were designed to evaluate the effect of a 1987 residual application of Si, a 1987 residual plus a new application of Si in 1988, and a new application of only Si in 1988. Neck blast incidence for the 1987 residual Si application was 29.1% less than the control. The 1987 residual Si plus a new application of Si in 1988 reduced blast incidence by 32.4%, whereas only a new application of Si in 1988 reduced blast by 27.0% over the control (Datnoff et al., 1991). Although the application of Si in 1988 generally suppressed neck blast more than did the 1987 residual applications on the 1988

I73

SILICON MANAGEMENT

rice crop, the residual applications were very effective. In addition, the effect of Si in suppressing disease incidence or severity seems to be comparable with the effect of fungicides (Datnoff et ul., 1990b; Datnoff, 1992, Datnoff and Snyder, 1994). Where Si was not applied, neck blast incidence decreased from 73% in the nonsprayed plots to 27% in the benomyl-treated plots (Datnoff, 1994; Datnoff and Snyder, 1994). Where Si was applied, blast incidence decreased to 36% in the nonsprayed plots to 13% in the benomyl-treated plots. Significant differences (P = 0.05) were not detected for neck blast incidence between plots only sprayed with benomyl and only amended with Si. Hence, it appears that Si can control diseases such as blast to the same degree as a fungicide. In upland rice grown on highly weathered tropical soils (Ultisols) of West Africa, the application of 18.7 g Si/m* as sodium metasilicate doubled plant Si uptake and significantly reduced the severities of husk discoloration, neck blast, sheath blight, and leaf scald (Winslow, 1992). Leaf scald severity and neck blast incidence were also reduced from about 26 and 53%, respectively, in nonamended plots to 15% in Si-amended plots on highly weathered savanna soils (Oxisols) in Colombia (Correa-Victoria et al., 1994). When black-gray ash from rice hulls containing amorphous Si was applied to rice seedbeds (about 1 kglmz), seedlings of rice cultivar susceptible to leaf blast were resistant up to 45 days after sowing (Kumbhar et ul., 1995). Several economically important insect pests, such as stem borer, brown planthopper, rice green leafhopper, and whitebacked planthopper, and noninsect pests, such as leaf spider and mites, have been suppressed by improving Si concentration of rice plants (Table IV). When the effect of Si fertilization on Si content of plant and penetration time of yellow stem borer larvae in rice plants (var. LMN 111) grown in nutrient Table IV Rice Pests Suppressed by Improved Silicon Nutrition Pest

Insect

Reference

Stem borer

Chilo supprrssalis. Srirpophaga incerrulas

Stem maggot Green leafhopper

Chlorops oryzae Nephotettix bipunctarus cinticeps Nilaparvata lugrns Sogetella furcifera Tetranychus spp. -

Ota er al. (1957) Yoshida (1975), IRRI (1965). Sawant er al. (1994) Maxwell et a / . (1972) Maxwell et a / . (1972)

Brown planthopper White-backed planthopper Leaf spider" Mites" Noninsect pests.

Sujatha er a!. (1987) Salim and Saxena (1992) Yoshida (1975) Tanaka and Park ( 1966)

174

N. K. SAVANT ET AL.

solution was studied, the penetration time of first instars increased from 2.8 min (for 0.47 ppm Si) to 21.2 min (for 47 ppm Si) (IRRI, 1990). Infestation of stem borer was reduced considerably by addition of Si to the soils with low available Si, but there was no change in the infestation when Si was added to the soils with high available Si (Nakano er al., 1961). Sawant et af. (1994) observed decreased incidence of deadhearts (stem borer damage) after transplanting rice seedlings (var. Jaya) that were fertilized with Si as black-gray ash of rice hulls at nursery stage, In a field study, a positive relationship between Si content of rice plants and resistance to brown planthopper (Nilaparvata lugens) has been observed (Sujatha et af., 1987). Increased Si concentration in some rice genotypes has been correlated with disease and/or insect resistance (Kozaka, 1965; Datnoff etal., 1991, 1992). It is possible that certain genotypes are more efficient than others in their accumulation of Si, thus making them more resistant (Deren et af., 1992, 1994; Winslow, 1992). However, some cultivars with low Si content were resistant to some diseases, whereas others with high Si content were susceptible (Ou, 1985). In addition, some genotypes with greater Si in plant tissue are not necessarily more disease resistant than those with lower Si levels for the same application rate of Si (Kozaka, 1965; Deren et al., 1994). Evidently, there are other genetic and/or modifying factors that seem to influence disease and possibly insect resistance in addition to Si levels in plant tissue. These factors should be identified and considered when developing or selecting genotypes for resistance. The Si-induced resistance of rice plants to biotic stresses can be attributed to one or more of the following physical, physiological, and/or biochemical factors. It is believed that the silicated epidermal layer formed by the filling of apertures of cellulose micelle constituting cell walls prevents physical penetration by insects and/or makes the plant cell less susceptible to enzymatic degradation by fungal pathogens (Sasamoto, 1961; Yoshida, 1975; Ou, 1985). Entomologists have found that the incisor region of the mandibles of stem borer larvae fed on rice plants with a high Si content were more damaged (Djamin and Pathak, 1967). In a resistant wild rice, 0. brachyanrha, Si cells in epidermal layers of leaves were more closely packed than those in susceptible rice (IR 31917) and hybrid rice and contributed to physical resistance through enhanced mandibular wear in leaf folder larvae (IRRI, 1991). Tasugi and Yoshida (1958) noticed the accumulation and deposition of Si around cells injured by the fungal penetration at a very early stage of infection. They assumed that this Si accumulation served as a barrier against fungal infection. The pest insect's behavior and responses seem to be affected by the presence of high levels of Si in the plant. Salim and Saxena (1992) investigated the effect of Si on the establishment of whitebacked planthopper on resistant IR 2035 and susceptible TN 1 rice cultivars cultured in hydroponics at 29/21' C (dayhight temperature). They observed decreased food intake, fewer nymphs becoming

SILICON MANAGEMENT

175

adults, decreased female and male longevity and fecundity, and eventually decreased insect population on susceptible TN 1 plants due to an increase in the level of Si in the nutrient solution. In another investigation, the presence of silicated cells on leaves inhibited scraping by leaffolder larvae of the green tissue on Si-treated rice plants; consequently, larval weight gain was significantly less on Si-treated plants than on the control plants (IRRI, 1991). Soluble silicic acid (as low as 0.01 mg/mi) in the sap of the rice plant acts as an inhibitor of the sucking activity of the brown planthopper (Yoshihara et af., 1979). Recently, another mechanism of resistance was reported in which Si stimulates chitinase activity and rapid activation of peroxidases and polyphenoxidases after fungal infection (Cherif et af., 1994). Glycosidcally bound phenolics extracted from Siamended plants when subjected to acid or B-glucosidase hydrolysis displayed strong fungistatic activity.

2. Abiotic Stresses Si influences water loss from plants largely by reducing cuticular transpiration (Jones and Handreck, 1967; Lewin and Reimann, 1969; Yoshida, 1975). Okuda and Takahashi (1965) grew rice plants (lowland rice variety Norin No. 22) in nutrient culture solutions ranging from 0 to 46.6 ppm Si (100 ppm SiO,) and measured the transpiration rate at certain intervals for a 2-month period. With increasing Si in the solutions, they found a consistent decrease in transpiration rate; in one instance, the rate decreased from 5.1 to 3.6 ml/g fresh wt/24 h for the nutrient solution containing 0 and 46.6 ppm Si, respectively. Horiguchi (1988) reported higher transpiration rates of -Si (control) plants than of +Si (treated) plants. These findings and other observations published elsewhere (Yoshida, 1975) demonstrate that by increasing Si content of rice plants, it may be possible to reduce their internal water stress and thereby make them withstand salt stress. Yoshida (1965) showed that rice plants without a Si supply could not grow in the culture solution containing a Na,SO, equivalent to an osmotic pressure of 5 atm, whereas the plants with a Si supply grew well in the same nutrient solution. Matoh et af. (1986) confirmed the beneficial effect of Si on the growth of NaC1-stressed rice plants grown in nutrient solution culture. They also observed that Si treatment reduced the sodium concentration of the shoots (tops) to 54% of that found in the shoots not receiving Si treatment. The excess absorption of Na, C1, and Mg by the salt-stressed cucumber and tomato plants was markedly decreased by the addition of Si to nutrient solution (Miyake, 1993). These observations may also be applicable to upland rice. Lodging can be defined as bending and/or breaking over of a plant before it is harvested. Lodging results in economic loss to farmers because it lowers quantity and quality of yield. Si application helps to increase resistance in plants to lodging (Jones and Handreck, 1967; Lewin and Reimann, 1969; Takahashi et

176

N. K. SAVANT ET AL.

al., 1990; Liang et al., 1994). For example, Lee et al. (1990) found that Si fertilization reduced lodging in lowland rice that was caused by N applied at 300 kg/ha. Ma et al. (1992) also noticed an increase in resistance to lodging due to the application of Si fertilizer to rice. Liang et al. (1994) reported less than 10% or practically no lodging in rice fields fertilized with a new silicate material and more than 66% lodging in the untreated (control) fields. Evidently, thickening of the cell walls of the sclerenchyma tissue in the culm and/or shortening and thickening of internodes or increase in Si content of the lower internodes (Kido and Yanatori, 1963) provides mechanical strength to enable the plant to resist lodging (Jones and Handreck, 1967; Lee er al., 1990; Liang et al., 1994; Takahashi, 1995). Because Si fertilization generally increases grain yield and culm supporting, heavy panicles can be more susceptible to lodging, therefore Si fertilization in certain tall or susceptible rice genotypes may do little to prevent lodging (IRRI, 1965; Snyder et al., 1986). Nevertheless, under the conditions of heavy application of N fertilizers, Si fertilization merits consideration because it has some potential to reduce lodging (Miyake, 1993).

C. INCRFASED PRODUCTMTY OF PROBLEM SOILS 1. Histosols Histosols or organic soils with high organic matter content are considered adverse or problem soils mainly because of nutrient deficiency problems (Green, 1957). Plant-available Si concentrations are lower in these organic soils than in most mineral soils (Elawad and Green, 1979; Snyder et al., 1986). The Si deficiency of Histosols can be corrected by amending them with silicate slags, and they can be made productive. Since 1964, silicate fertilizers have been used by Japanese farmers for enhancing productivity of peaty lowland rice fields (Kono, 1969). Recent field research in the Everglades Agricultural Area, south of Lake Okeechobee in Florida, demonstrated that amending Histosols with silicate slag increased rice yields by 5 6 4 8 % (Snyder et al., 1986).

2. Acid Soils with At, Fe, and Mn Toxicities Si fertilization may be useful to alleviate Al, Fe, and Mn toxicities in certain acid soils. Application of slag (2 t/ha) to soil with high exchangeable Fe and Mn decreased the contents of the latter elements in rice plants without producing their deficiency symptoms (IRRI, 1965). This could be due to an increase in the oxidizing capacity of the roots of rice plants supplied with adequate amounts of Si (Okuda and Takahashi, 1965). According to Ponnamperuma (1965b), sufficient Si supply facilitates oxygen transport more efficiently from the plant tops to the roots through enlargement or rigidity of gas channels and, as a result, in-

SILICON MANAGEMENT

177

creases oxidation and subsequent deposition of iron and manganese on the root surface, thus excluding them from absorption by the plants. Horiguchi (1988) confirmed that the effect of Si on the alleviation of Mn toxicity in rice seedlings was due to the decreased uptake of Mn by the Si-supplied plants. However, he attributed the beneficial effect of applied Si to three factors: (a) the increased oxidizing power of the rice roots causing Mn deposition on them, (b) the decreased transpiration playing an important role by reducing the Mn absorption, and (c) the increased tolerance to an excess amount of Mn in the plant tissues. Results of some recent studies indicate that Si ameliorates A1 toxicity to plants not only by decreasing the activity of free A13+ in solution but also by reducing the internal toxicity of Al. The A1 toxicity effect on growth of sorghum (Galvez et af., 1987) and maize (Barcelo et al., 1993) was decreased when Si (up to 4 pM concentration) was added to nutrient solution (pH 4).Decreased A1 concentration in sorghum plants grown on acid soils has been found to be associated with increased levels of Si in the plants (Clark and Gourley, 1988; Clark et al., 1988; Galvez and Clark, 1991). Exposure to A1 markedly increased Si levels in the corticle cell walls of roots of spruce seedlings in an acid soil (Hodson and Wilkins, 1991). It seems possible that the formation of aluminosilicate compounds in walls of root cortex cells inhibits the uptake of A1 into the protoplast and reduces A1 toxicity. Based on the nutrient solution culture experiments with maize, Barcelo er al. (1993) concluded that Si has a significant effect on both A1 speciation in the solution and A1 uptake by the plants. They also noticed Si in the solution causing a substantial increase in malic acid concentration in the plant roots, and postulated a chelation of A1 by malic acid as another mechanism of Siinduced A1 tolerance of maize. These results may be relevant to upland rice culture and need to be confirmed using rice plants grown on acid soils under upland conditions. Si has also been suggested as a factor responsible in the A1 and Fe tolerance of some species of gramineae when grown on acid soils because of its possible involvement in cation-anion balance in plants (Wallace, 1993). Silicate (SiO,) supposedly adds to the excess of anion uptake and therefore expels equivalent amounts of OH- from the roots, which can increase soil pH at the root surface and eventually decrease uptake of A1 and Fe. However, this type of cation-anion balance as a possible mechanism for A1 and Fe tolerance in rice grown in acid soil has not been demonstrated under laboratory or field conditions.

VI. AGRONOMIC ESSENTIALITY OF SILICON MANAGEMENT From a plant physiological point of view (see Section II), Si is currently not considered an essential element for plant growth and development (Epstein,

178

N. K. SAVANT ET AL.

1994). However, the several potential benefits of Si nutrition in rice (Section V) indicate the agronomic essentiality of Si management for increasing and/or sustaining rice yields. Additionally, this view point can be supported by the following factors. 1. Soil Si status sign$es potential soil productivity: In a field survey, rice plants suffering from various nutritional disorders were found low in Si content (Tanaka and Park, 1966). Because low Si content per se was not a direct cause of the disorders, it was speculated that other soil factors associated with low Si supply may be required for healthy development of the rice plant. It was also observed that high soil productivity is generally associated with adequate supply of soil Si (Lian, 1976). Maximum grain yields were usually obtained on soils where the Si content of the straw at harvest was as high as 6.1-7.1% Si (Imaizumi and Yoshida, 1958; Lian, 1976). These observations indicate that soil Si supply can be an important factor determining or associated with maximum rice yield ceilings in a given ecosystem. 2. Balanced nutrient management systems need to include Si application for sustained rice production: Japanese and Korean rice farmers are able to sustain high yields in the range of 6 to 9 t/ha (IRRI, 1993a) probably because their nutrient management systems include a practice of silicate slag application at 1.5-2.0 t/ha for degraded lowland rice soils. On the other hand, the annual growth rate of yields of irrigated rice obtained during Green Revolution periods of south and southeast Asia declined from 2.9% in the mid- and late 1970s to 1.9%in the 1980s (Pinstrup-Anderson and Pandya-Lorch, 1994). Mutert et al. (1993) reported some decline in rice yield growth rates between 1980 and 1990 resulting in almost stagnant yields in China, India, Indonesia, Thailand, and Vietnam. In long-term experiments (1968-1991) on continuous irrigated rice systems in the tropics, IRRI scientists have observed a phenomenon of yield decline (Cassman et al., 1995). These declining trends in the rate of yield increases may partially be due to failure to replace mega amounts of Si removed from soils by rice crops every season (CRRI, 1976; Amarasiri and Wickamasinghe, 1977; Miyake, 1993). 3. Si management is critical for intensive rice production: Intensive production (two or more cropdyear) can result in rapid depletion of all nutrients, including that of Si, because the short fallow periods between two successive crops in a year may not be sufficient for replenishing plant-available Si in soil when the dissolution rate of soil Si is very slow (see Section 111). Under these soil conditions, the additions of large amounts of NPK fertilizers alone may gradually lose their agronomic effectiveness. Nitrogen sources applied at higher rates will produce rice plants with soft, droopy leaves that are more susceptible to diseases, pests, and lodging (Yoshida, 1975; 1981) and are photosynthetically less efficient (Yoshida et al., 1969). This unhealthy and unproductive crop condition could be

SILICON MANAGEMENT

I79

prevented by using a proper Si application. In other words, for intensive rice production systems it seems essential to adopt appropriate Si management practices for sustaining additional yield responses to other nutrients applied at higher rates, especially nitrogen (Kawaguchi, 1966; Lian, 1976; Yoshida, 1971). 4. Si management can play a positive role in integrated pest management: In order to protect the environment, some feel there is an urgent need to reduce use of chemical pesticides for managing insect pests and fungal diseases while increasing rice productivity. Research already has demonstrated that Si can reduce the incidences of several important rice diseases, such as leaf blast and brown spot, to the same general degree as a fungicide (Datnoff, 1994; Datnoff and Snyder, 1994). Seebold et al. (1995) showed that the number of fungicide applications could be reduced when Si was applied to soils low in plant-available Si. Consequently, wherever possible, adoption of region-specific effective Si management practices as a part of integrated pest management has that potential. 5. Weathered tropical rice soils can be more productive with proper Si management: Higher temperatures prevailing during the growing season in the subtropics and tropics can help to increase soil Si availability (Sumida, 1990, 1992b), but the temperature effect may help only to some extent because most soils in traditional rice growing areas are highly weathered and dedicated (Kawaguchi, 1966; Foy, 1992; Friesen et a l . , 1994). For substantially increasing and sustaining rice yields of these regions, therefore, Si management should be an essential part of integrated nutrient management systems (Yoshida, 1971; Park, 1976).

VII. DETERMINING NEED FOR SILICON FERTILIZATION The rice crop responds favorably to Si fertilization depending on soil factors, such as the plant-available Si status of the soil, and plant factors such as the Si content of plant tissue. However, traditional sources of Si used as fertilizers (mainly silicate slags) are available in limited quantities and can pose logistic problems; more important, they are expensive. Therefore, in order to give proper guidance to rice farmers for judicious use of available Si sources, there is a need for efficient soil testing and plant testing for Si content.

A. SOILTESTING Since 1950, Japanese researchers have conducted a series of investigations into the relationship between Si in soils and plants and Si fertilization. Imaizumi

180

N. K. SAVANT ET AL.

and Yoshida (1958) compared the amounts of Si absorbed by rice plants from soils and Si extracted from the soils by hot HCI, ammonium oxalate (pH 3.0), 2% Na,CO,, C0,-saturated water, 0.002 N H,SO,, and ammonium acetate (pH 4.0). They concluded that dilute acid-soluble Si was better correlated with Si absorption by plants and the Si-supplying power of the soil, and that dilute alkalisoluble Si did not reflect the Si-supplying power of the soil. They further observed that soil Si was more soluble in the acid than in the neutral solutions and that the extractability of Si from soils by various reagents tested was in the following order: oxalate > citrate > tartarate > acetate and chloride. From this they concluded that the dissolution of Si in soil is increased by chelation, which helps to lower the effective concentration of silicate fixers such as A1 and Fe in the soil. Recognizing the potential role of A1 and Fe compounds in the dissolution kinetics of Si in soil, Kawaguchi et af. (1958) and Kawaguchi and Matsuo (1958) probably have proposed concurrent examination of the Si:Fe and Si:AI ratios and the amounts of Si extracted by the acetate buffer or 0.5 N HCl for evaluating Si-supplying power of the soil. Khalid et u1. (1978) used both waterextractable Si and Si extractable by 0.1 M acetic acid (pH 3.5) containing 50 mg P/liter as Ca(H,PO,), to study residual Si in soils after application of calcium silicate slag. The assumption was that water-extractable Si is a measure of the solution concentration at near equilibrium with the soil system (an intensity factor), whereas the acidified phosphate-extractable Si is an index of the amount of Si that remains in the soil in an adsorbed form (a capacity factor). Thus, both methods of extraction may provide different but useful information. Imaizumi and Yoshida (1958) put forward a criterion for Si fertilizer application that states that if the extractable Si (sodium acetate buffer, pH 4.0) content in the soil is less than 4.9 mg Si /lo0 g soil, beneficial effects of a silicate fertilizer application may be observed. The acetate buffer method has been used in Japan and Taiwan and, with some modifications in the extraction ratio and time and temperature of shaking, for estimating availability of Si in soils for rice plants in Korea (Kim et al., 1971; Park, 1976; Lian, 1976). The acetate buffer is made by diluting 49.2 ml acetic acid and 14.8 g anhydrous sodium acetate to 1 liter and adjusting to pH 4.0 with acetic acid or sodium acetate (K. Nonaka, personal communication). Unfortunately, it appears that no papers have been published in English that specifically describe the development and use of this buffer. The sodium acetate buffer-extractable Si content of calcareous soils of China ranged from 7.1 to 18.1 mg Si/100 g of soil and still rice yields responded to the application of silicate fertilizer to these soils (Liang et al., 1994). Therefore, it seems that the buffer probably overestimates available Si in calcareous soils under field conditions, and there is a need to develop a suitable soil test (probably extractant) for calcareous paddy soils. In general, criteria for using acetate buffer-extractable Si to determine the need for Si application to tropical rice soils have not been adequately investigated.

181

SILICON MANAGEMENT

In India, Nayer et af. (1977) compared N sodium acetate buffer (pH 4.0) with three other chemical extractants (distilled water, 0.2 N HCl, and 0.025 M citric acid) and reported the extracting power of the reagents for Si as: 0.2 N HCl > 0.025 M citric acid > N acetate buffer > water with some exceptions in some soils. In a greenhouse experiment, they found that the 0.025 M citric acidextractable Si in soils showed better correlation with the Si uptake by the rice plants (var. Jaya). For tropical rice soils of Malaysia and Thailand, Kawaguchi (1966) used 3.3 mg Si/IOO g soil and for those of Sri Lanka, Takijima et al. (1970) used 3.8 mg Si/ lOOg soil as tentative criteria of acetate buffer-extractable Si for describing Si deficiency. Dependable criteria for describing Si deficiency in tropical and subtropical soils are not seen in the literature and need investigation (see Section VIII). IRRI workers observed a positive relationship between the Si content of flooded soil percolates and the Si content of rice plants (IRRI,1965). According to Lian (1976), therefore, a percolation method may be more realistic because it will directly measure the Si content in the submerged soil, whereas the acetate buffer method estimates the available Si indirectly in air-dried soil samples. A Si soil test based on extraction with 0.5 M acetic acid has been developed by Snyder (1991) for identifying the need for Si fertilization of Histosols planted to rice in the Everglades Agricultural Area in Florida. This soil test has not, however, been adequately evaluated for soils previously amended with calcium silicate. For improving the dependability and scope of the chemical soil tests for available Si to rice plants, two points relative to soil samples to be used merit consideration: (i) Decreased redox potential (Eh) of flooded lowland rice soil increases water-soluble Si, and therefore air-dried soil samples with high redox potential may not truly represent the soil environment in which rice roots have to absorb Si; (ii) most of these extractants are likely to underestimate the need for Si fertilization of soils that have been previously amended with slags. This is because Si extracted by the acetate buffer evidently increases in soil treated with slag (Nonaka and Takahashi, 1990). It seems that the “too strong acetate buffer dissolves some nonavailable Si from the slag previously added to soil. In order to address these issues, Nonaka and Takahashi (1988, 1990) developed a method for measuring water-soluble Si in rice soil that involves flooded soil incubation. With this method, a 10-g dry soil sample (air-dried and <2mm) is submerged in a 100-ml cylindrical bottle (about 4.5 cm i.d.) with 60 ml water and incubated at 40°C for a week, after which time the supernatant is analyzed for Si content. For soils that had a history of Si fertilization, Si extracted by the new method generally correlated better with Si in straw than did Si extracted with the acetate buffer method, although for some soils neither method produced a good correlation. These researchers considered the 2-week period between soil sampling and soil-test results to be a serious disadvantage of the incubation method. ”

182

N. K. SAVANT ET AL.

Sumida (1992b) developed a soil incubation method that requires even longer submergence time than the method of Nonaka and Takahashi (1990). Sumida (1992b) incubated soil under flooded conditions for 4 weeks at 30“ C using a soi1:water ratio of 1 :4.To eliminate ferrous iron interference in the colorimetric determination of Si, he added 1 g of cation exchange resin (Amberlite IR-120, H type) to a 20-ml aliquot of the supernatant. He also found that the 4-week incubation method provided more reasonable estimates of available Si for various soils, but this did not always correlate well with Si uptake by rice. Therefore, Sumida (1992b) developed another method in which concurrent dissolution and adsorption of Si in a submerged soil were examined by incubating the soil with silicate solutions containing from 0 to 100 ppm SiO, (46.6 ppm Si) at 30” C using a soi1:solution ratio of I :lo. He calculated the so-called “potentially soluble silicate” by the following equation: c = (uc + av)la, where v is mg of silicate dissolved or adsorbed/100 g soil, u is mg of Si0,lliter of supernatant solution, a is constant indicating the SiO, concentration at which neither dissolution nor adsorption occurs, and c is potentially soluble SO,. He concluded that the method provided the most suitable indices of the Si-supplying capacity of lowland rice soils amended with calcium silicate or organic matter and soils with varying clay minerals and texture. The potentially soluble silicate concept may be good, but the method is not suitable for routine soil testing because of its longer incubation period. The long incubation periods required for water-extraction methods may be due to the fact that drying of soil samples leads to changes in the equilibrium between soluble and solid silicon substances (V. V. Matichenkov, personal communication). Soluble silicon compounds (monosilicic and polysilicic acids and organosilicon substances) adsorbed on soil particles are dehydrated when the soil is dried. Soil may have to be immersed in water for up to 1 month in order to restore the natural equilibrium. Matichenkov is developing water-extraction soil-test methods based on field-moist soil samples. The temperature of soil incubation or of shaking (extraction) would be a critical factor in evaluating a soil test for available Si in soils of a given agroecological zone because the Si supply from lowland rice soils and Si uptake by rice depends on the temperature (Sumida and Ohyama, 1990; Sumida, 1992b).

B. PLANTTESTING Although rice with low Si content will usually have soft and droopy leaves (weeping willow-like appearance) (Lewin and Reimann, 1969), it is more accurate to chemically determine the Si content. There is little controversy as to which plant part should be analyzed, and for most practical testing purposes the straw of the whole rice plant at harvest is sampled mainly because it is normally

SILICON MANAGEMENT

183

related to grain yield and the results can be easily compared with those obtained in previous studies. Park el al. (1964) used the Si content of the flag leaf as an index for available Si in soil. In view of a 10% coefficient of variation for the Si content of straw in a uniform lowland rice field, a composite sample of 10 representative hillslfield would be required for a 10% precision (Imaizumi and Yoshida, 1958; IRRI, 1978a; Chang, 1978). Winslow (1995) suggested the use of hull (husk) of paddy rice as a sampling material for the diagnosis of Si status because of the high level of accuracy and sensitivity, low coefficient of variation, and practical convenience. On the basis of a large number of trials conducted in Japan, Korea, and Taiwan to correlate the Si content of straw with the yield response of rice to Si fertilization, Lian (1976) calculated the critical percentage of Si in straw for no yield response to slag application. ,It is >6.1% Si in straw in Japan and Korea, and 5.1% Si (for both the first and second rice crops) in Taiwan. For >5% additional yield, the critical percentage of Si in straw is <5.1% in Japan, <4.7% in Korea, and <4.2% in Taiwan. In cooler ecoregions, japonica rice varieties seem to respond to the applications of silicate slags in which Si content in straw is less than 5.1%. In tropical ecoregions of Sri Lanka and India, indica rice varieties may respond to Si application in which Si content in straw is below 3.7% (Nair and Aiyer, 1968; Takijima er al., 1970). Snyder er al. (1986) have suggested that more than 3% Si (6. 4% SiO,) in straw is needed for good yields of rice grown on organic soils (Histosols) of the Everglades Agricultural Area in Florida. Rice plant tissue may be analyzed for Si by several methods. High-temperature fusion with NaOH, followed by development of a silicomolybdous chromophore and subsequent colorimetry (Kilmer, 1965), provides the most rigorous conditions for solubilization of Si in plant tissue. Plant tissue can be analyzed for Si gravimetrically by freeing Si from the tissue matrix through wet digestion procedures, followed by weighing it as Si (Yoshida et al., 1976; Elliott er al., 1988). Colorimetric determination of Si in aliquots of nitric acid digests of rice plant tissue has been suggested (CRRI, 1974). Hydrofluoric acid may be used to solubilize Si in plant material or ash, with subsequent determination of Si by atomic absorption spectrometry or colorimetry (Novozamsky and Houba, 1984; van der Vorm, 1987). These methods are laborious andtor involve the use of very hazardous chemicals. An autoclave-induced digestion method, which has been developed specifically for determining Si in rice tissue, has none of these drawbacks (Elliott and Snyder, 1991). Plant tissue is digested with NaOH and H,O, in an autoclave; when coupled with automated colorimetry, the method has been found useful for processing large numbers of plant tissue samples. Spical relationships between rice yield responses to Si fertilization in Taiwan and the Si content of straw samples or the available Si in the surface soil samples are shown in Fig. 10. Similar relationships, when developed for other agroclima-

N. K. SAVANT ET AL.

184

"""

1

1.00

3

~

'

0.95

v)

+

Y

5

0.90

2 'Z

$

I

1

0.85

z .2

.-A .8

L

Q)

.->

/

r = 0.67** '

9.3

o

18.7

37.4

46.7

"

7.5

28.0

Available Si in surface soil (pprn) 1.00

0.95-

-m

. I -

8

0.90 -

0.85 -

1,.

0.80

I

2.8

'

'

3.7

"

4.7

"

5.6

6.5

Si content of straw (%) Rgure 10 Relationships between relative rice yield and (top) available Si in surface soil and (bottom) Si content of straw (Lian, 1976).

tic regions, may be useful in determining needs of paddy fields for Si fertilization.

C . SI AVAILABILITY INSI SOURCES Various slag waste products from industrial processes have been used as Si fertilizers. However, there is considerable variation in the composition of these slags and in Si availability to rice. For example, Takahashi (1981) determined

SILICON MANAGEMENT

185

that Si availability was greater for slags that had been cooled slowly, compared to water-cooled slags, and that availability increased with decreasing grain size. He also determined that the mole ratio of Ca to Si influences Si availability, with a ratio of greater than 1.0 favoring Si uptake. Slags also vary considerably in Si content. Among the slags from steel mills, convertor slags that result from the manufacture of very pure steels typically contain considerably less Si (5-10%) than blast furnace slags (1520%) that result from pig iron manufacture. Slags produced during the electric furnace production of P contain approximately 20% Si, just slightly less than mined calcium silicate minerals such as wollastonite. Japanese workers have used a variety of methods of assessing the plantavailable Si in slags and other inorganic materials. Silicon extraction by 0.5 M HCl has become an “official” method of gauging slag effectiveness, although many researchers feel it is of little value in predicting Si uptake by rice (K. Nonaka, H. Sumida, and N. Kato, personal communication). Poor correlations between acetate buffer-soluble (pH 4) Si and straw Si have been reported (Nonaka and Takahashi, 1986). Water-soluble Si also can be correlated with Si uptake by rice, but several factors have been shown to influence Si dissolution in the laboratory. For example, in incubation studies, Ca accumulation in the supernatant represses further dissolution of calcium silicates (Kato and Owa, 1990). Increasing the solution pH and salt concentration also diminishes Si dissolution. Because rice in the field removes Ca from solution and produces CO, in the soil through root respiration, these factors become important when trying to simulate Si dissolution in the laboratory. N . Kato (personal communication) developed a laboratory method for analyzing Si availability that takes into account such factors as Ca dissolution and pH changes. He uses Amberlite IRC-50 (pK 6.1) resin to adsorb dissolved Ca and to stabilize pH in a 4-day period at 35°C incubation of slag in water and with continuous shaking at 35°C. The correlation between Si dissolved by this method and Si uptake by rice was found to be better than that for dissolution in water or in acetate buffer.

VIII. SUGGESTIONS FOR RESEARCH Research information available on Si nutrition ofjaponica rice, soil Si, and Si management in cooler agroclimatic zones may be adequate for sustainable rice cultivation in Japan, Korea, and Taiwan. There are, however, several gaps in information on Si nutrition of indica rice and its Si management that need to be filled before the potential benefits of Si nutrition can be fully realized in fields by rice farmers of subtropical and tropical Asia, Latin America, and Africa. Some suggestions for research are given.

186

N. K. SAVANT ET AL.

1. Regional Si status surveys of soils, waters, and plants: As a result of dedication (Fig. 4), highly weathered subtropical and tropical rice soils (Alfisols, Ultisols, Oxisols, etc.) may contain insufficient plant-available Si for healthy growth of rice and sustained high levels of yields. So far, the overall nutritional role of soil Si in rice production has received adequate attention in Japan, Korea, and Taiwan but not in south and southeast Asia, Latin America, and Africa. Therefore, extensive regional Si status surveys of the subtropical and tropical soils, plant (straw) samples, and irrigation waters used (if any) are needed to identify the nature and magnitude of the problem of Si availability in these soils. Demarcation of areas having low, medium, and high levels of plantavailable Si in the soils should help in developing appropriate Si management practices and in eventually increasing or sustaining regional rice productivity. 2. Development of cheaper and eficient Si sources and their use: Silicate slags are expensive Si sources, and therefore most rice farmers of tropical and subtropical regions probably will be unable to use them at the rates of 1 or 2 t/ha/year. There is a need to identify and/or develop cheaper, efficient sources of Si and to determine their appropriate use integrated with other nutrient management practices. Findings of this research should enable rice farmers to efficiently use slag at affordable rates. In the future, if Si fertilization is required as a part of a balanced integrated nutrient management system for sustaining rice production in developing countries, sufficient amounts of silicate slags may not be available to many rice farmers in those countries. In order to address the potential problem of a shortage of slags in the 2 1st century, agronomic and policy research should be started now to promote environmentally friendly recycling of Si in rice straw and hull (both on-farm sources) alone or integrated with silicate slag application. Recycling of rice straw as a source of Si needs to be done in a way that will ensure minimum damage to the environment because its incorporation into soil increases emission of greenhouse gases (such as methane) from paddy fields. 3 . Si management for reducing methane emission from lowland rice fields: It has been demonstrated that increased Si absorption by rice plants improves the oxygen transport (supply) from the tops to the roots thus increasing the oxidizing power of the roots (Okuda and Takahashi, 1965; Ponnaperuma, 1965b). It is also known that methane emission from lowland rice soil is largely related to methane produced minus methane oxidized (chemically and/or microbiologically) in a soil-rice plant system (Minami, 1994). Thus, these observations imply that developing efficient Si management practices may help to increase the oxidizing power of the roots, which in turn may cause more oxidation and eventually less emission of methane from lowland rice fields, thus helping to partially mitigate the greenhouse gas problem of paddy fields. Preliminary results of a greenhouse pot trial suggested this possibility (N. K. Savant, G . H. Snyder, and L. E. Datnoff, unpublished data), but additional research is needed.

SILICON MANAGEMENT

187

4. Si management for reducing pesticide use: Si uptake induces resistance in rice plants to insect pests and fungal diseases (see Section V). Si can also control diseases to the same degree as a fungicide (Datnoff, 1994; Datnoff and Snyder, 1994). Recent research has demonstrated that fungicide applications can be reduced when used in combination with Si (Seebold et a l . , 1995). Therefore, different Si sources and their appropriate management practices should be developed and, wherever possible, investigated in integrated pest management programs. 5 , Development of Si-amended phosphate sources and their use: The presence of silicate as Ca,SiO, in Rhenania phosphate containing P as CaNaPO, decreased P retention by soil and increased the level of water-soluble P in soil in laboratory experiments (Chien, 1978) and increased rice yields by 23% compared with superphosphate alone as a P source in a pot culture trial (Anonymous, 1976). Unfortunately, rhenania phosphate is no longer produced, mainly because of its high-cost energy-intensive production. However, the results of rhenania phosphate experiments and other observations published elsewhere (Adams, 1980) illustrate that it may be possible to develop silicate-amended phosphate sources and/or integrated management practices of using different sources of Si and conventional phosphate sources in different combinations for improving efficiency of applied P while supplying Si to rice. 6. Genetics research for identifiing e$cienr Si-accumulating rice cultivars: Genetics plays an important role in Si uptake by rice plants. Genotypes differ in their Si contents and respond differently to applied Si (Garrity et a l . , 1984; Majumder et a l . , 1985; Winslow, 1992; Deren et a l . , 1992;). In the genetics of Si uptake, it seems that additive as well as nonadditive genes are involved (Majumder et a l . , 1985). While developing and selecting genotypes for other desirable traits, it will be prudent to consider rice genotypes with greater Si concentration. 7. Si management for breaking yield barriers: For raising the yield ceiling in the future, attempts are under way to redesign and develop a new rice plant type (with yield potential of 13-15 t/ha) (IRRI, 1994b) for intensively cropped irrigated ecosystems and hybrid rice for the tropics. This new rice plant type and the hybrids will reach their yield potentials on farmers’ fields only when their large nutritional needs are adequately met. It will be necessary to address agronomic field issues relative to new rice plants’ nutritional requirements, including that of Si, before they are released for use by farmers. 8. Foliar Si application-A potential management option? Okamoto (1 963) demonstrated that spraying of soluble Si (0.1-0.2 ppm solution of Na,SiO,) on plants on alternate days improved overall plant growth (except root system) and silicified cells in the leaf blade. He also observed that the spraying facilitated NPK absorption through roots and translocation of N and P within the plants. More research in this area may lead to the development of a Si management option for irrigated areas.

188

N. K. SAVANT ET AL.

M.SUMMARY Rice is a Si accumulator; therefore, adequate attention should be given to the beneficial role that Si nutrition and its management can play in a balanced integrated nutrient management system for increasing and sustaining rice yields. Large amounts of Si as monosilicic acid [H,SiO, or Si(OH)",] are absorbed by the rice plant. Subsequently, Si is deposited as amorphous SiO, and is mainly associated with cellulose and hemicellulose in the rice plant tops (leaves and hulls) in the form of so-called cuticle-Si double layer. The solubility of soil Si is low. Si in soil solution (intensity factor) is largely influenced by its dissolution kinetics, which in turn is influenced by various soil factors such as Al, Fe oxides, organic matter, redox potential, and moisture. In general, highly weathered soils of the tropics and subtropics (Oxisols, Ultisols, etc.) are low in available Si mainly due to a dedication (leaching) process. Si management agenda includes Si fertilization (mainly application of silicate slags) and recycling of Si in rice crop residues (straw and hull). Si has several potential benefits. Its sufficient supply in soil is required for healthy growth and productive development of the rice crop. Apparently, applied Si seems to interact favorably with other applied fertilizer nutrients (namely N, P, and K) and offers the potential to improve their agronomic performance and efficiency in terms of yield response. Si-amended rice plants possess varying degrees of ability to resist or to tolerate biotic stresses, such as attack of insect pests and fungal diseases, and abiotic stresses due to toxicity of soil Al, Fe, and Mn and excessive salts. Si supply also helps to reduce cuticular transpiration and to some extent crop lodging caused by excessive N supply. Agronomically, Si management is essential for increasing andlor sustaining rice productivity in temperate, tropical, and subtropical soils. So far, Si nutrition and management in rice have received adequate attention in Japan and Korea and to some extent in Taiwan. Research needs include extensive Si status surveys of soil, water, and plant in traditional rice growing areas of the tropics and subtropics. This initial step should help in identifying the nature and magnitude of the Si-deficiency problem of regional rice ecosystems of developing Asia, and then in developing a suitable Si management agenda for obtaining or sustaining yield potentials of improved rice cultivars in given ecoregions. Because yield responses of rice to Si application are related to available Si in soil and the Si content of rice plants, appropriate soil and plant testing methods should be developed for deciding the need for Si application to soils in given agroclimatic regions. Research work is needed to identify or develop relatively inexpensive and efficient Si sources and methods of use and environmentally safe ways of recycling Si in rice crop residue for use by small farmers. Research on Si management

SILICON MANAGEMENT

189

for making problem soils (organic, acid, or saline soils) more productive and for decreasing use of insecticides/fungicides also has merit.

ACKNOWLEDGMENTS We thank Dr. Paul J. Stangel for reviewing the manuscript and making useful suggestions. We also thank Ms. Dalia Stubblefield and Mrs. Elizabeth Roth for helping with corrections and Mr. Norman Harrison for final preparation of the figures.

REFERENCES Adams, F. ( 1980). Interactions of phosphorus with other elements in soils and in plants. In “The Role of Phosphorus in Agriculture,” pp. 655-680. American Society of Agronomy, Madison, WI. Agarie, S.. Agata, W., Kubota, F., and Kaufman, P. B. (1992). Physiological roles of silicon in photosynthesis and dry matter production in rice plants. Jpn. J. Crop Sci. 61, 200-206. [in Japanese with English summary] Aleshin, N. E. (1988). On the biological role of silicon in rice. Vesrnik Sel’skokhozyaistvennoi Nauki. Moskva. 10, 77-85. [In Russian; Soils Fert. 53, 5050, 19901 Aleshin, N. E., and Avakyan, E. R. (1983). Absorption of silicon by rice. Izv. Akad. Nauk SSSR, Ser. Biol. 3, 451-454. [In Russian; Chem. Absr. 99, 52472u, 1983). Aleshin, N. E., Avakyan, E. R., Dyakunchak. S . A.. Aleshkin. E. P.. Baryshok, V. P., and Voronkov, G. (1987). Role of silicon in resistance of rice to blast. Doklady Akademii Nuuk SSSR. 291(2), 217-219. Amarasiri. S. L. (1973). Water quality of irrigation tanks in Sri Lanka. Trop. Agric. 129, 19-25. Amarasiri, S. L. (1978). Organic recycling in Asia-Sri Lanka. FA0 Soils Bull. 36, 119-133. Amarasiri. S. L.,and Perera, W. R. (1975). Nutrient removal by crops in the dry zone of Sri Lanka. Trop. Agric. 131, 61-70. Amarasiri. S. L., and Wickramasinghe. K. (1977). Use of rice straw as a fertilizer material. Trop. Agric. 133, 39-49. Anonymous ( 1976). “Rhenania-Phosphate Multipurpose Fertilizer with Universal Applicability.” Kali-Chemie AG., Hannover, Federal Republic of Germany. Baba, I., Iwata, I., and Takahashi, Y. (1956). Studies on the nutrition of rice plants with reference to Helminrhosporium leaf spot (preliminary report). XI. Absorption and translocation of nutrients as influenced by soil moisture and air humidity. Proc. Crop Sci. Sac. Jpn. 24, 169-172. [in Japanese] Barber, D. A., and Shone, M. G. T. (1966). The absorption of silica from aqueous solutions by plants. J . Exp. Bot. 17, 569-578. Barcelo, J., Guevara, P., and Poschenrieder, Ch. (1993). Silicon amelioration of aluminum toxicity in teosinte (Zea mays L. ssp. mexicana). Plunr Soil 154, 249-255. Bergmann, W. (1 992). “Nutritional Disorders of Plants: Development, Visual and Analytical Diagnosis.” Gustav Fisher Verlag Jena, Stuttgart/New York. Blair, G. J., Freney. J. R., and Park, J. K. (1990). Effect of sulfur, silicon, and trace metal interactions in determining the dynamics of phosphorus in agricultural systems. In “Proceedings of the Symposium on Phosphorus Requirements for Sustainable Agriculture in Asia and Oceania,” March 6-10, 1989, pp. 269-280. International Rice Research Institute, Los Banos, Laguna, Philippines.

190

N. K. SAVANT E T AL.

Burbey, A,, Rizaldi, B., and Yulizar, Z. (1988). Response of upland rice to potassium and silicate application on Ultisol. Pemberitaan Penelitian Sukarami 15, 26-31, [In Indonesian with English summary; Regional Info. Support Service 14, 517, 19901 Cassman, K. G., DeDatta, S. K., Olk, D. D., Alcantara, J., Sanson, M., Descalsota, J., and Dizon, M. (1995). Yield decline and the nitrogen economy of long-term experiments on continuous, irrigated rice systems in the tropics. In “Soil Management-Exprimental Basis for Sustainability and Environmental Quality” (R. La1 and B. A. Stewart, Eds), pp. 181-222. CRC Press, Boca Raton, FL. Central Rice Research Institute (CRRI) (1974). Annual report 1973. Cuttack, Orissa, India. Central Rice Research Institute (CRRI) (1976). Annual report 1975. Cuttack, Orissa, India. Chang, S. C. (1978). Evaluation of the fertility of rice soils. In “Soils and Rice. International Rice Research Institute,” pp. 52 1-541. Los Banos, Laguna, Philippines. Chen, Y. (1990). Characteristics of silicon uptaking and accumulation in rice. J. Guizhou Agric. Sci. 6 , 37-40. [In Chinese with English summary] Cheng, B. T. (1982). Some significant functions of silicon to higher plants. J. Plant Nutr. 5 , 13451353. Cherif, M., Asselin. A,, and Belanger, R. R. (1994). Defense responses induced by soluble silicon in cucumber roots infected by Pyrhium spp. Phytopatho1og.v 84, 236-242. Chien, S. H. (1978). Reactions of phosphate rocks, rhenania phosphate and superphosphate with an acid soil. Soil Sci. Soc. Am. J . 42, 705-708. Chiu, T. F., and Huang, W. L. (1971). Expe;iinents on effects of variously applied siliceous slag to rice. Annual report, Taiwan Agriculture Research Institute for 1970, Taiwan. Clark, R. B., and Gourley, L. M. (1988). Mine;dl element concentrations of sorghum genotypes grown on tropical acid soil. Commun. Soil Sci. Plant Anal. 19, 1019-1029. Clark, R. B., Flores, C. I., and Gourley, L. M.(1988). Mineral element concentration in acid soil tolerant and susceptible sorghum genotypes. Comrrurr. Soil Sci. Plant Anal. 19, 1003-1017. Correa-Victoria, F. J., Datnoff, L. E., Winslow, M. D., Okada, K., Friesen, D. K, Sanz, J. I., and Snyder, G. H. (1994). Silicon deficiency of upland rice on highly weathered Savanna soils in Colombia. 11. Diseases and grain quality. 1X Conferencia Internaciaonal de arroz para a America Latina e para o Caribe, V Reuniao Nacional de Pesquisa de Arroz, Castro’s Park Hotel. 21-25 Marco. Goiania, Goias, Brasil. Cuevas, V. C. (1993). Rapid composting fits rice farmers. Inform. Cenfr.for Low External Inputs and Susfainable Agric. Newsletr. 9(2), 11-12. Dahlgren, R. A. (1993). Comparison of soil solution extraction procedures: Effect on solute chemistry. Commun. Soil Sci. Plant Anal. 24, 1783-1794. Datnoff, L. E. (1992). Influence of silicon fertilization on rice diseases in Florida. Newslett. Florida Phyropathol. Soc. 3, 10-11. Datnoff, L. E. (1994). Influence of mineral nutrition of rice on disease development. In “Rice Pest Science and Management” (P. S. Teng, K. L. Heong, and K.Moody, Eds.), pp. 89-100. IRRI, Manila, Philippines. Datnoff, L. E., and Snyder, G . H. (1994). Comparison of silicon and benomyl alone and in combination for reducing blast incidence. B i d . Cultural Tesfs Control Plant Dis. 9, 113. Datnoff, L. E., Raid, R. N., Snyder, G. H., and Jones, D. B. (1990a). Evaluationof calcium silicate slag and nitrogen on brown spot, neck rot, and sheath blight development on rice. B i d . Cultural Tests Control Planf Dis. 5 , 65. Datnoff, L. E., Snyder, G. H., and Jones, D. B. (1990b). Influence of calcium silicate slag and fungicides on brown spot and neck rot development and yields of rice. In “1990 Rice Growers Seminar,” pp. 26-33. Belle glade EREC, Research Report EV-1990-2 (F. J. Coale, Ed.). University of Florida, Belle Glade, FL.

SILICON MANAGEMENT

191

Datnoff, L. E., Raid, R. N., Snyder, G. H., and Jones, D. B. (1991). Effect of calcium silicate on blast and brown spot intensities and yields of rice. Plant Dis. 75, 729-732. Datnoff, L. E., Snyder. G. H., and Deren, C. W. (1992). Influence of silicon fertilizer grades on blast and brown spot development and on rice yields. Plant Dis. 76, 1182-1184. Datta. N. P., Shinde. J. E., Kamath, M. D., and Datta, S. K. (1962). Effect of sodium silicate on the uptake of soil and fertilizer phosphorus by wheat, rice and berseem. Indian J. Agric. Sci. 32, 2 19-227. DeDatta, S. K. (1981). “Principles and Practices of Rice Production.” Wiley, New York. Deren, C. W., Datnoff, L. E., and Snyder. G. H. (1992). Variable silicon content of rice cultivars grown on Everglades Histosols. J. Plant Nutr. 15, 2363-2368. Deren, C. W., Datnoff, L. E., Snyder, G. H., and Martin, F. G. (1994). Silicon concentration, disease response and yield components of rice genotypes grown on flooded organic Histosols. Crop Sci. 34, 733-737. D’Hoore J., and Coulter, J. K. (1977). Soil silicon and plant nutrition. In “Soil of the Humid Tropics,” pp. 163- 173. National Academy of Sciences, Washington, DC. Diamond, R. B., Sri Adiningsih, J., and Sudjadi, M. (1986). Maintaining yields of lowland rice with reduced fertilizer applications in Indonesia. Paper presented at the INSFFER workshop, September 22-24, 1986, Hangzhow, China. Ditsathaporn, C., Rungsangchan, P., Yuvaniyom, A., Suttavas, A., and Arunin, S. (1989). Effect of rice hull ash and plowing depths on saline-acid soil improvement. Research Report 1989, pp. 184- 194. Agriculture Development Research Center, Bangkok, Thailand. Djamin, A.. and Pathak, M. D. (1967). Role of silica in resistance to Asiatic rice borer, Chilo suppressalis (Walker) in rice varieties. J. Econ. Enromol. 60, 347-35 I . Drees, L. R., Wilding, L. P., Smeck, N. E., and Senkayi. A. L. (1989). Silica in soils: Quartz and disordered silica polymorphs. In “Minerals in Soil Environments” (J. B. Dixon and S. B. Weed, Eds.), pp. 913-974. Soil Science Society of America, Madison, WI. Elawad. S. H., and Green, V. E., Jr. (1979). Silicon and the rice plant environment: A review of recent research. I / Riso 28, 235-253. Elliott, C. L., and Snyder, G. H. (1991). Autoclave-induced digestion for the colorimetric determination of silicon in rice straw. J. Agric. Food Chem. 39, 1118-1 I 19. Elliott, C. L., Snyder, G . H., and Jones, D. B. (1988). Rapid gravimetric determination of Si in rice straw. Commun. Soil Sci. Plant Anal. 19, 1543- 1550. Epstein, E. (1994). The anomaly of silicon in plant biology. Proc. Nazl. Acad. Sci. USA 91, 11-17. Fageria, N. K., Baligar, V. C., and Jones, C. A. (1991). “Growth and Mineral Nutrition of Field Crops.” Dekker, New York. Food and Agriculture Organization (FAO) (1993). FA0 Bull. Slat. 6(4), 22. Foy, C. D. (1992). Soil chemical factors limiting plant root growth. Adv. Soil Sci. 19, 97-149. Friesen, D. K., Sanz, J. I , Correa, F. J, Winslow, M. D., Okada, K., Datnoff, L. E., and Snyder, G. H. (1994). Silicon deficiency of upland rice on highly weathered Savanna soils in Colombia. 1. Evidence of a major yield constraint. IX Conferencia lnternaciaonal de arroz para a America Latina e para o Caribe, V Reuniao Nacional de Pesquisa de Arroz, Castro’s Park Hotel. 21-25 Marco. Goiania. Goias, Brasil. Galvez, L.. and Clark, R. B. (1991). Effect of silicon on growth and mineral composition of sorghum (Sorghum bicolor) grown with toxic levels of aluminum. In “Plant-Soil Interactions at Low pH* (R. J. Wright. V. C. Baligar. and R. P. Munman. Eds.), pp. 815-823. Kluwer Academic, Dordrecht. The Netherlands. Galvez, L., Clark, R. B., Gourley, L. M., and Maranville, J. W. (1987). Silicon interactions with manganese and aluminum toxicity in sorghum. J. Plant Nutr. 10, 1139-1 147. Garrity, D. P., Mamaril, C. P., and Soepardi, G. (1990). Phosphorus requirements and management

192

N. K. SAVANT ET AL.

in upland rice-bdsed cropping systems. In “Proceedings of the Symposium on Phosphorus Requirements for Sustainable Agriculture in Asia and Oceania,” March 6-10, 1989, pp. 333347. International Rice Research Institute, Los banos, Laguna, Philippines. Gmity, D. P., Vidal, E. T., and 0’Toole, J. C. (1984). Genotype variation in the thickness of silica deposition on flowering rice spikelets. Ann. Boi. 54, 413-421. Green, V. E., Jr. (1957). The culture of rice on organic soils-A world survey. Agron. J. 49, 468472. Hallmark, C. T., Wilding, L. P., and Smeck, N. E. (1982). Silicon. In “Methods of Soil Analysis,” Part 2, pp. 263-273. Agronomics Monograph No. 9, 2nd Ed. American Society of Agronomy, Madison, WI. Ho, D. Y., Zang, H. L., and Zhang, X. P. (1980). On the silicon supplying ability of some important paddy soils in South China. In “Proceedings of the Symposium on Paddy Soils,” 19-24 October, 1980, p. 95. Nanjing, China. [Abstracts] Hodson, M. J., and Wilkins, D. A. (1991). Localization of aluminum in the roots of nonvay spruce (Picea abies L. Karst) inoculated with Paillus involutus Fr. New Phytol. 118, 273-278. Horiguchi, T. (1988). Mechanism of manganese toxicity and tolerance of plants. IV. Effect of silicon on alleviation of manganese toxicity of rice plants. Soil Sci. Plant Nutr. 34, 65-73. Irnaizumi, K., and Yoshida, S. (1958). Edaphological studies on silicon supplying power of paddy soils. Bull. Nail. Ins!. Agric. Sci. (Jpn) B 8, 261-304. Inanaga S., Okasaka, A,, and Tanaka, S. (1995). Does silicon exist in association with organic compounds in rice plants? Soil Sci. Plant Nutr. 41, 1 1 1-1 17. Indian Agriculture Research Institute (IARI) (1988). Annual report 1987. New Delhi, India. International Fertilizer Industry Association (IFA) (1992). Rice. In “IFA World Fertilizer Use Manual,” p. 38. Paris, France. International Rice Research Institute (IRRI) (1964). Annual report 1963. Los Banos. Laguna, Philippines. International Rice Research Institute (IRRI) (1965). Annual report 1964. Los Banos, Laguna, Philippines. International Rice Research Institute (IRRI) (1966). Annual report 1965. Los Banos, Laguna, Philippines. International Rice Research Institute (IRRI) (1967). Annual report 1966. Los Banos, Laguna, Philippines. International Rice Research Institute (IRRI) (l978a). “Soils and Rice.” IRRI, Los Banos, Laguna, Philippines. International Rice Research Institute (IRRI) (1978b). Annual report 1977. Los Banos, Laguna, Philippines. International Rice Research Institute (IRRI) (1990). Program report for 1989. Los Banos, Laguna, Philippines. International Rice Research Institute (IRRI) (1991). Program report for 1990. Los Banos, Laguna, Philippines. International Rice Research Institute (IRRI) (1992). Program report for 1991. Los Banos, Laguna, Philippines. International Rice Research Institute (IRRI) (1993a). “IRRI Rice Almanac.” Los Banos, Laguna, Philippines. International Rice Research Institute (IRRI) (1993b). Rice hulls fuel new cook stove; design wins national prize. IRRI Reporter 2, 8. International Rice Research Institute (IRRI) (1994a). Free-fuel stove a success and saves forests. IRRI Reporter 2, 7. International Rice Research Institute (IRRI) (1994b). “Rice in Crucial Environment.” IRRI, Los Banos. Laguna, Philippines.

SILICON MANAGEMENT

193

Ishizuka, Y. (1973). Physiology of the rice plant. Food Fert. Tech. Cent. Tech. Bull. 13, 241-315. Islam, A., and Saha, R . (1969). Effects of silicon on the chemical composition of rice plants. f l a n t Soil 30,446-458. Ismunadji. M. (1978). Utilization of cereal crop residues and its agricultural significance in Indonesia. Contr. Center Res. Insr. Agric. Bogor, Indonesia 37, 1-14. Jones, L. H. P., and Handreck. K. A. (1967). Silica in soils, plants and animals. Adv. Agron. 19, 107- 149. Juo, A. S. R., and Sanchez. P. A. (1986). Soil nutritional aspects with a view to characterize upland rice environment. In “Upland Rice Research.” pp. 8 1-95. International Rice Research Institute, Los banos, Laguna, Philippines. Kang, Y. S. (1985). The influence of silicon on growth of rice plants. Res. Rep. Rural Dev. Admin., Plant Environ., Mycol.. Farm Product Utilization, Korean Republic 27( I), 57-72. [In Korean; Rice Abstr. 10, 2504, 19871 Kanotra, S., and Mathur, R. S. (1994). Biodegradation of paddy straw with cellulolytic fungi and its application on wheat crop. Biores. Technol. 47, 185-188. Kato, N . , and Owa, N. (1990). Dissolution mechanism of silicate slag fertilizers in paddy soils. XIV Inr. Congr. Soil Sci. 4, 609-610. Kawaguchi. K. (1966). Tropical paddy soils. Jpn. Agic. Res. Q . 1, 7-1 I . Kawaguchi, K., and Matsuo, Y. (1958). A study of available silica in paddy soils. 11. J. Sci. Soil, Tokyo 29, 9-12. [Soils Fert. 22, 2052, 19591 Kawaguchi, K., Ojima, T., and Matsuo, Y. (1958). A study of available silica in paddy soils. I. J. Sci. Soil. Tokyo 28, 493-496. [SoilsFert. 22, 2053, 19591 Khalid, R. A . , Silva, J. A., and Fox, R. L. (1978). Residual effects of calcium silicate in tropical soils: 1. Fate of applied silicon during five years of cropping. Soil Sci. Soc. Am. J. 42, 8994. Kido, M., and Yanatori, S. (1963). Physiological & ecological studies on the rice plant in welldrained and poorly-drained fields. 8. Distribution of silicified tissue of rice culm & silicate content in each internode as affected by different cultural conditions. Proc. Crop Sci. Soc. Jpn. 31, 237-240. [In Japanese] Kilmer, V. J. (1965). Silicon. In “Methods Of Soil Analysis,” Part 2, pp. 959-962. Chemical and Microbiological Properties, Agronomy Series No. 9. American Society of Agronomy, Madison, WI . Kim, C. B.. Kim, S. H.. Park, N. K., Park, S. D., andChoi, D. U. (1985). Effectsofapplicationof slag as silicate materials on rice yields in normal paddy soil. Res. Rep. Rural Dev. Admin.. Plant Environ., Mvcol.. Farm Product Utilization, Korea Republic 27(1), 41-60. [In Korean; Rice Abstr. 10, 2503. 19871 Kim, C. H.. Rush, M. C., and MacKenzie, D. R. (1986). Food-mediated resistance to the rice blast disease. In “The Wetlands and Rice in Subsaharan Africa” (A. S. R. Juo and J. A. Lowe, Eds.), pp. 151- 169. IITA. Ibadan, Nigeria. Kim, C. K., and Lee, S. (1982). Reduction of the incidence of rice neck blast by integrated soil improvement practice. Korean J . Plant frot. 21, 15-18. Kim, Y. K., et al. (1971). Research Report of the Office of the Rural Development (Plant Environment), vol. 14, pp. 39-44. Kobayashi, T. (1957). Studies on the improvement of the upland soil by means of irrigation practice. 11. Improvement of the volcanic ash soil in Kuma District Kumamoto Prefecture. Bull. Far. Agric. Kagashima Univ. 6, 1-22. [In Japanese] Kobayashi. T., and Shinagawa, A. (1957). Studies on the improvement of the upland soil by means of irrigation practice. 111. On the mechanisms of the improvements of the upland fields of volcanic ash soil by irrigation. Bull. Far. Agric. Kagashima Univ. 6 , 23-43. [In Japanese] Kobayashi, T., Ogata, T., and Yoshida. Y. (1955). Studies on the improvement of the upland soils by

194

N. K. SAVANT ET AL.

means of imgation practice. I. Improvement of volcanic ash soil in Goshi Clearing, Kumamoto Prefecture. Bull. Kyushu Agric. Exp. Sta. 3, 1-30. [In Japanese] Kono, M. (1969). Effectiveness of silicate fertilizer to japonica varieties. Trop. Agric. Res. Ser. No. 3, 241-247. Kono, M., and Takasashi, J. (1958). Selective absorption of silica and calcium by rice and tomato plants. J . Sci. Soil, Tokyo. 29, 63-66. Kozaka, T. (1965). Control of rice blast by cultivation practices in Japan. In “The Rice Blast Disease,” pp. 421-438. John Hopkins Press, Baltimore, MD. Kubota, S. (1984). Slow-release potassium fertilizer with special reference to potassium silicate fertilizer. Paper presented at the International Symposium on Fertilizer Quality & EfficiencyProspect and Problem, October 23-25, 1984, Tokyo, Japan. Kumbhar, C. T., Nevase, A. G., and Savant, N. K. (1995). Rice hull ash applied to soil reduces leaf blast incidence. Int. Rice Res. Notes 20, 23-24. Kundu, S., Kamath, M. B., and Goswami, N. N. (1988). Effect of sulfate, silicate and fluoride anions. I. Phosphate fixation in soil. J. Indian Soc. Soil Sci. 36, 43-47. Laming, F. C. (1963). Silicon in rice. J . Agric. Food Chem. 11, 435-437. Lee, K. S., Ahn, S. B., Rhee, G. S., Yeon, B. Y., and Park, J. K. (1985). Studies of silica application to nursery beds on rice seedling growth. Res. Rep. Rural Dev. Admin., Plant Environ., Mycol., Farm Product Utilization, Korea Republic 27( I ) , 23-27. [In Korean; Rice Absfr. 10, 2502, 19871 Lee, D. B., Kwon, T. O., and Park, K. H. (1990). Influence of nitrogen and silica on the yield and the lodging related traits of paddy rice. Res. Rep. Rural Dev. Admin., SoiIs Ferr. 32(2), 15-23. [In Korean; Soils Fert. 55, 2428, 19921 Lee, T. S., Hsu, L. S . , Wang, C. C., and Jeng, Y. H. (1981). Amelioration of soil fertility for reducing brown spot incidence in the paddy fields of Taiwan. J. Agric. Res. China 30, 35-49. Lewin, J., and Reimann, B. E. F. (1969). Silicon and plant growth. Annu. Rev. Plant Physiol. 20, 289-304. Lian, S. (1963). The effect of silica on paddy rice. Agric. Res. Taiwan 12(3), 16-28. [Soil Fen. 27, 1181, 19641 Lian, S. (1976). Silica fertilization of rice. I n “The Fertility of Paddy Soils and Fertilizer Applications for Rice,” pp. 197-220. Food Fertilizer Technology Center, Taipei, Taiwan. Lian, S. (1988). Fertility management of rice soils in R. 0. C. on Taiwan. In “Classification and Management of Rice Growing Soils.” Food Fertilizer Technology Center, Taipei, Taiwan. Lian, S. (1992). Effect of continued application of silicate slag under the double rice-based cropping system. In “Proceedings of the International Symposium on Paddy Soils,” September 15- 19, 1992, pp. 222-223. Academia Sinica, Nanjing, China. [Abstracts] Liang, Y. C., Ma, T. S., LI, F. J., and Feng, Y. J. (1994). Silicon availability and response of rice and wheat to silicon in calcareous soils. Commun. Soil Sci. Plant Anal. 25, 2285-2297. Lin, K. C., Lian, S., and Lee, L. T. (1962). Experiments on the effect of silica on paddy. Agric. Res. Taiwan 11(3), 45-47. [Soils Fert. 27, 1180, 19641 Lindsay, W. L. (1979). “Chemical Equilibria in Soils.” Wiley, New York. Liu, S. L., and Ho, C. H. (1960). Study in the nature of silicon in rice hull. I. Solubility of the silicon part. J. Chinese Chem. Soc. 6, 141-153. Ma, J. F., and Takahashi, E. (1989). Release of silicon from rice straw under flooded conditions. Soil Sci. Plant Nutr. 35, 663-661. Ma, J. F., and Takahashi, E. (1990a). Effect of silicon on growth and phosphorus uptake of rice. Plant Soil 126, 115- 119. Ma, J. F., and Takahashi, E. (1990b). The effect of silicic acid on rice in a P-deficient soil. Plant Soil 126, 121-125. Ma, J. F., and Takahashi, E. (1990~).New aspects of silicon nutrition in rice plants. XIVInr. Congr. Soil Sci. Kyoto, Japan 4, 158-163.

SILICON MANAGEMENT

195

Ma, 3. F., and Takahashi, E. (1991a). Effect of silicate on phosphate availability for rice in a P-deficient soil. Plant Soil 133, 151-155. Ma, J. F., and Takahashi, E. (1991b). Availability of rice straw Si to rice plants. SoilSci. Plant Nutr. 37, 111-116. Ma, J. F., Nishimura, K.. and Takahashi, E. (1989). Effect of silicon on the growth of rice plant at different growth stages. Soil Sci. Plant Nutr. 35, 347-356. Ma. T. S . , Wang, D. P., Liung, Y. T.. Chen. S. H., Zhang, F. S . , Wang, S. F., Chien, 2. Z., and Liu, L. 2. (1992). Effect of high efficiency silicate fertilizer on rice. Turang 24(2), 168-169. [In Chinese] Majumder, N. D., Rakshit, S. C., and Borthakur, D. N. (1985). Genetics of silica uptake in selected sativa L.). Plant Soil 88, 449-453. genotypes of rice (0. Matoh, T.. Kairusmee, P., and Takahashi. E. (1986). Salt-induced damage to rice plants and alleviation effect of silicate. Soil Sci. Plant Nutr. 32, 295-304. Maxwell, F. G.. Jenkins, J. N., and Parrott, W. L. (1972). Resistance of plants to insects. Adv. Agron. 24, 187-265. McKeague, J. A., and Cline. M. G. (1963a). Silica in soil solutions. 11. The absorption of monosilicic acid by soil and by other substances. Can. J . Soil Sci. 43, 83-96. McKeague, J. A., and Cline, M. G. (l963b). Silica in soils. Adv. Agron. 15, 339-396. Minami, K. (1994). Methane from rice production. Fert. Res. 37, 167-179. Mitsui, S . , and Takatoh, H. (1963). Nutritional study of silicon in graminaceous crops. Soil Sci. Plant Nutr. 9, 49-58. Miyake, Y. (1993). Silica in soils and plants. Sci. Rep. Fuculry uf Okayuma Univ. 81, 61-79. [In Japanese with English summary] Mizuno, N. (1987). Effects of silica on hull weight and ripening of rice plants. J . Plant Nutr. lO(916), 2159. Mutert, E., Portch, S . , and Woo, Y. C. (1993). Asia’s present and future agronomic problems. AgroChemi. News Brief 15(4). 20-30. Nair, P. K., and Aiyer, R. S. (1968). Status of available silica in the rice soils of Kerala State (India). 11. Silicon uptake by different varieties of rice in relation to available silica contributed by soil and irrigation water. Agric. Res. J . Keralu 6 , 88-94. Nakano, K., Abe, G., and Hirano, C . (1961). Jpn. J . Appl. Enromol. Zool. 5, 17-27. Nanda, H. P.. and Gangopadhyay. (1984). Role of silicated cells in rice leaf on brown spot disease incidence by Bipolaris Otyzae. Inr. J . Trop. Plant Dis. 2, 89-98. Nayar, P. K., Misra, A. K . . and Patnaik. S. (1977). Evaluation of silica-supplying power of soils for growing rice. Plant Soil 47, 487-494. Nayar, P. K., Misra, A. K., and Patnaik, S . (1982). Silica in rice and flooded rice soils. 11. Uptake of silica in relation to growth of rice varieties of different durations grown on an Inceptisol. Otyzu 19, 82-92. Nishihara, T., Tsuneyoshi, T.. and Nakamura, H. ( 1970). Effect of silicon on phosphorus metabolism in rice plant. Bull. Fur. Agric. Kugoshimu (lniv. 20, 85-103. [In Japanese; Soils Ferl. 34, 1705, 19711 Nogushi, J., and Sugawara. T. ( 1966). “Potassium and Japonica Rice.” International Potash institute Bern, Switzerland. Nonaka, K, and Takahsahi, K. (1986). Available silicates in paddy soils. Part I . Available silicates in various soils and slags. Bull. Shikoku Natl. Agrir. Exp. Stn. 47 (March), 15. Nonaka, K., and Takahashi, K. (1988). A method of measuring available silicates in paddy soils. Jpn. Agric. Res. Q. 22, 91-95. Nonaka, K., and Takahashi, K. (1990). A method of assessing the need of silicate fertilizers in paddy soils. XIV Int. Congr. Soil Sci. Kyoto, Japan 4, 513-514. Novozamsky, R. van Eck, and Houba, V. J. G. (1984). A rapid determination of silicon in plant material. Commun. Soil Sci. Plant Anal. 15, 205-21 1.

196

N. K. SAVANT ET AL.

Ohata, K., Kubo, C., and Kitani, K. (1972). Relationship between susceptibility of rice plants to Helmithosporium blight and physiological changes in plants. BuII. Shikoku Agric. Exp. Sta. 25, 1-19. Okamoto, Y. (1963). Effect of silicic acid absorbed from the surface of leaves on the growth of rice plants. Proc. Crop Sci. Soc. Jpn. 32, 61-65. [Soils Ferr. 27, 1182, 19641 Okuda, A , , and Takahashi, E. (1962). Effect of various metabolic inhibitors on the silicon uptake by rice plants. Part 9. J. Sci. Soil Manure, Jpn. 33, 453-455. Okuda, A,, and Takahashi, E. (1965). The role of silicon. In “The Mineral Nutrition of the Rice Plant,” pp. 126-146. John Hopkins Press, Baltimore, MD. Osuna-Canizales, F. I., DeDatta, S. K., and Bonman, J. M. (1991). Nitrogen form and silicon nutrition effects on resistance to blast disease of rice. Planr Soil 135, 223-231. Ota, M. (1964). “Studies of the Utilization of Slag Fertilizers.” Kazama, Tokyo, Japan. [In Japanese] Ota, Y. (1988). Role of the hull in ripening of rice plant. Taichung Dist. Agric. Improvement Sfa. 13, 173-188. [Special publication; Field Crop Abstr. 43, 301, 19901 Ota, M., Kobayashi, H., and Kawaguchi, Y. (1957). Effect of slag on paddy rice. 2. Influence of different nitrogen and slag levels on growth and composition of rice plant. Soil Plant Food 3, 104-107. Ou, S . H. (1985). “Rice Diseases,” 2nd ed. Commonwealth Mycological Institute, Kew, Surrey, UK. Park, C. S. (1976). Silicate responses to rice in Korea. In “The Fertility of Paddy Soils and Fertilizer Applications for Rice,” p. 221. Food Fertilization Technology Center, Taipei, Taiwan. Park, Y. S., Oh, W. K., and Park, C. S. (1964). A study of the silica content of the rice plant. Res. Rep. Of.Rural Dev. Suwon 7(1), 31-38. [In Korea] Pinstrup-Anderson, P., and Pandya-Lorch, R. (1994). Alleviating poverty, intensifying agriculture, and effectively managing natural resources. Food, Agriculture, and the Environment, Discussion Paper No. 1. International Food Policy Research Institute (IFPRI), Washington, DC. Philippine Rice Research Institute (PhilRice) ( 1994). Almost smokeless rice hull stove developed. PhilRice Newsletter 7(2), 5 . Ponnamperuma, F. N. (1965a). Dynamic aspects of flooded soils and the nutrition of the rice plant. In “The Mineral Nutrition of the Rice Plant,” pp. 295-328. John Hopkins Press, Baltimore, MD. Ponnamperuma, F. N. (1965b). Review of the symposium on the mineral nutrition of the rice plant. In “The Mineral Nutrition of the Rice Plant,” pp. 461-482. John Hopkins Press, Baltimore, MD. Ponnamperuma, F. N. (1984). Straw as a source of nutrients for wetland rice. In “Organic Matter and Rice,” pp. 117-136. International Rice Research Institute, Los Banos, Laguna, Philippines. Rachmat, R., Thahir, R., and dan Sutrisno, S. (1991). Rice hull stove development for household. Agrimek 3( 1 ), 30-34. [In Indonesian with English summary] Raghupathy, B. (1993). Effect of lignite fly ash (LFA) on rice. Inr. Rice Res. Notes 18(3), 27-28. Rodrigo, D. M. (1964). Response of rice to silica. Trop. Agric. 120, 219-226. Rossman, A. Y., Howard, R. J., and Valent, B. (1990). Pyricularia grisea, the correct name for the rice blast disease fungus. Mycologia 82, 509-521. Roy, A. C., Ah, M. Y., Fox, R. L., and Silva, 1. A. (1971). Influence of calcium silicate on phosphate solubility and availability in Hawaiian latosols. Proc. Int. Symp. Soil Ferf. Evaluation,New Delhi. India 1, 757-165. Sadanandan, A. K., and Verghese, E. J. (1968). Studies on silicate nutrition of rice in the laterite soil of Kerala. I . Effect on growth and yield. Madras Agric. J. 55, 260-263. Sadzawka, R. M. A., and Aomine, S. (1977). Adsorption of silica in river waters by soils in Central Chile. Soil Sci. Plant Nutr. 23, 297-309. Sahu, S. K. (1990). Effect of silica and phosphorus application on yield and phosphorus nutrition of rice. Int. Rice Res. Notes 15(1), 25. Salim, M., and Saxena, R. C. (1992). Iron, silica, and aluminum stresses and varietal resistance in rice: Effects on whitebacked Planthopper. Crop Sci. 32, 212-219.

SILICON MANAGEMENT

197

Sasamoto, K. (1961). Resistance of rice plant applied with silicate and nitrogenous fertilizers to the rice stem borer Chilo Suppressalis Walker. Proceedings of the Faculty Lib. Art Education, Yamanashi University, No. 3. Savant, N. K . , and Sawant, A. S . (1995). Nutrient composition of rice seedlings as influenced by rice hull ash application to seedbed. O y a , in press. Sawant, A. S., Patil, V. H., and Savant, N. K. (1994). Rice hull ash applied to seedbed reduces deadhearts in transplanted rice. Inr. Rice Res. Notes 19(4), 21-22. Seebold, K., Datnoff, L., Correa-V., F., and Snyder, G. (1995). Effects of silicon and fungicides on leaf and neck blast development in rice. Phvtoparhology 85, 1168. Seo, S. W., and Ota, Y. (1983). Role of the hull in the ripening of rice plant. VII. Effect of supplying of silica and potassium during reproductive growth stage on the form and function of hulls. Nippon Sakumorsu Gakkai Kiji 52(1), 73-79. [In Japanese; Chem. Absrr. 99, 2146911, 19831 Sharma, S. K., Sharma, C. M., and Chakor, 1. S. (1988). Effect of industrial organic wastes and Lantana incorporation on soil properties and yield of rice. Indian J . Agron. 33(2), 225226. Silva, J. A. (1971). Possible mechanisms for crop response to silicate applications. Proc. Intern. Svmp. Soil Ferr. Eva/.. New Delhi, India I, 805-814. Snitwongse, P.. Pongpan. S., and Neue, H. U. (1988). Decomposition of I4C labelled rice straw in a submerged and aerated rice in Northeastern Thailand. Paper presented at the First International Symposium on Paddy Soil Fertility, December 6-13. 1988, Chaiang Mai, Thailand. Snyder, G. H. (1991 ). Development of a silicon soil test for Histosol-grown rice. Belle Glade EREC Research Report EV-1991-2 University of Florida, Belle Glade. FL. Snyder, G. H., Jones, D. B., and Gascho, G. J. (1986). Silicon fertilization of rice on Everglades Histosols. Soil Sci. Soc. Am. J . 50, 1259-1263. Su, N. R. (1972). The fertility status of Taiwan soils. Technical Bulletin No. 8, Food Fertilizer Technology Center, Taipei, Taiwan. Su, N. R., Shen, L., and Lee, T. S. (1983). Availability of the native soil and residual fertilizer silica to rice plants. J . Agric. Assoc. China 122, 46-62. [Rice Absrr. 7, 2371, 19841 Subramanian, S., and Gopalswamy, A. (1990). Influence of silicate and phosphate materials on availability and uptake of silicon and phosphorus in acid soil. O y z a 27, 267-273. Subramanian, S . , and Gopalswamy, A. (1991). Effect of moisture, organic matter, phosphate, and silicate on availability on silicon and phosphorus in rice soils. J . Indian Soc. Soil Sci. 39, 99103. Sujatha, G., Reddy, G. P. V., and Murthy. M. M. K. (1987). Effect of certain biochemical factors on expression of resistance of rice varieties to brown planthopper (Nilaparvuia lugens Stal). J . Res. APAU (Andhra Pradesh Agric. Univ.) 15(2), 124-128. Sumida, H. (1992a). Effects of nitrogen nutrition on silica uptake by rice plant. Jpn. J . Soil Sci. Plant Nurr. 63, 633-638. Sumida, H. (1992b). Silicon supplying capacity of paddy soils and characteristics of silicon uptake by rice plants in cool regions in Japan. Bull. Tohoku Natl. Agric. Exp. S m . 85, 1-46. [In Japanese with English summary] Sumida, H., and Ohyama, N. (1990). The influence of temperature on supply of silica from paddy soil and its uptake by rice plant. Jpn. J . Soil Sci. Planr Nurr. 61, 253-259. [In Japanese] Suzuki, H. (1935a). Studies on the influence of some environmental factors on the susceptibility of the rice plant to blast and Helminrhosporium diseases and on the anatomical characters of the plant. 11. Influence of differences in soil nioisture and in the amount of nitrogenous fertilizer given. J . Coll Agric. XIII, 236-275. Suzuki, H. (1935b). Studies on the influence of some environmental factors on the susceptibility of the rice plant to blast and Helminrhosporium diseases and on the anatomical characters of the plant. 111. Influence of differences in soli moisture and in the amounts of fertilizer and silica given. J . Coll. Agric. XIII, 278-332.

N. K. SAVANT ET AL. Syouji, K. (1981). Application effect of calcium silicate, rice straw, and citrate on phosphorus availability in soil. Soil Sci. Plant Nurr. 52, 253-259. Takahashi, E. (1978). Effect of the form of silicon on the uptake of silicon by rice plant. Jpn. J . Soil Sci. Plant Nutr. 49, 357-360. Takahashi, E. (1995). Uptake mode and physiological functions of silica. Sci. Rice Planr 2, 58-71. Takahashi, E., Ma, J. F., and Miyake, Y. (1990). The possibility of silicon as an essential element for higher plants. Comments Agric. Food Chem. 2, 99-122. Takahashi, E., and Miyake, Y. (1977). Silica and plant growth. In “Proceedings of the International Seminar on Environment and Fertility Management in Intensive Agriculture,” (SEFMIA) pp. 603-61 I . Tokyo, Japan. Takahashi, J., Kanareugsa, C., Somboondumrongkul, J., and Prasittikhet, J. (1980). The effect of silicon, magnesium and zinc on the yield of rice. In “Proceedings of the Symposium On Paddy Soils,” October 19-24, 1988, pp. 82-83. Nanjing, China. [Abstracts] Takahashi, K. (1981). Effects of slags on the growth and the silicon uptake by rice plants and the available silicates in paddy soils. Bull. Shikoku Agric. Exp. Stn. 38 (December). Takahashi, N. (1968). Silica as a nutrient to the rice plant. Jpn. Agric. Res. Q . 3(3), 1-4. Takahashi, Y. (1967). Nutritional studies on development of Helminrhosporium Leaf Spot. In “Proceedings of the Symposium on Rice Diseases and Their Control by Growing Resistant Varieties and Other Measures,” pp. 157-170. Forestry and Fisheries Research Council, Tokyo, Japan. Takijima, Y., and Gunawardena, S . D. I. E. (1969). Nutrient deficiency and physiological disease of lowland rice in Ceylon. 1. Relationships between nutritional status of soil and rice growth. Soil Sci. Plant Nutr. 15, 259-266. Takijima, Y., Wijayaratna, H. M. S., and Seneviratne, C. J. (1970). Nutrient deficiency and physiological disease of lowland rice in Ceylon. 111. Effect of silicate fertilizers and dolomite for increasing rice yields. Soil Sci Planr Nurr. 16, 11-16. Tanaka, A. (1973). Methods of handling rice straw in various countries. Int. Rice Comm. Newslett. 22, 1-20. Tanaka, A. (1978). Role of organic matter. In “Soils and Rice, International” pp. 605-620. International Rice Research Institute, Los Banos, Laguna, Philippines. Tanaka, A,, and Park, Y. D. (1966). Significance of the absorption and distribution of silica in the rice plant. Soil Sci. Plant Nutr. 12, 191-195. Tanaka, A., and Yoshida, S. (1970). Nutritional disorders of the rice plant in Asia. International Rice Research Institute, Los Banos, Laguna, Philippines. Tasugi, H., and Yoshida, K. (1958). On the relation between silica and resistance of rice plants to rice blast. Monbu-sho Shiken Kenkyu Hokoku 48, 31-36. [In Japanese] Tokunaga, Y. (1991). Potassium silicate: A slow release potassium fertilizer. Ferf. Res. 30, 55-59. Tran, D. T. (1991). The Lo Trau: Taking the heat off Africa’s forest. Ceres 131, 8. van der Vorm, P. D. I. (1980). Uptake of Si by five plant species as influenced by variations in Sisupply. Plant Soil 56, 153-156. van der Vorm, P. D. J. (1987). Dry ashing of plant material and dissolution of the ash in HF for the colorimetric determination of silicon. Commun. Soil Sci. Plant Anal. 18, 1181- 1189. Verma, T. S . , and Minhas, R. S. (1989). Effect of iron and manganese interactions on paddy yield and iron and manganese nutrition in Silicon-treated and untreated soils. Soil Sci. 147, 107-1 15. Volk, R. J., Kahn, R. P., and Weintraub, R. L. (1958). Silicon content of the rice plant as a factor influencing its resistance to infection by rice blast fungus, Pyricularia oryzae. Phytopathology 48, 179-184. Wallace, A. (1989). Relationships among nitrogen, silicon, and heavy metal uptake by plants. Soil Sci. 147, 457-460. Wallace, A. (1993). Participation of silicon in cation-anion balance as a possible mechanism for aluminum and iron tolerance in some gramineae. J . Planr Nutr. 16, 547-553.

SILICON MANAGEMENT

199

Winslow, M. D. (1992). Silicon, disease resistance and yield of rice genotypes under upland cultural conditions. Crop Sci. 32, 1208- I2 13. Winslow, M. D. (1995). Silicon: A new macronutrient deficiency in upland rice. Working Document No. 10. International Center of Tropical Agriculture, Cali, Colombia. Yamauchi, M., and Winslow, M. D. (1987). Silica reduces disease on upland rice in a high rainfall area. Int. Rice Res. Newsleft. 12, 22-23. Yamauchi, M., and Winslow, M. D. (1989). Effect of silica and magnesium on yield of upland rice in humid tropics. Plant Soil 113, 265-269. Yoshida, S. (1965). Chemical aspects of the role of silicon in physiology of the rice plant. Bull. Natl. Inst. Agr. Sci. (Jpn.)Ser. B 15, 1-58. Yoshida, S. (1971). Nutritional disorders of rice in Asia. Exten. Bulletin No. 4. Food Fertilization Technology Center, Taipei, Taiwan. Yoshida, S. (1975). The physiology of silicon in rice. Technical Bulletin No. 25. Food Fert. Tech. Centr., Taipei, Taiwan. Yoshida, S. (1978). The availability of silicon in paddy soil. I n “Paddy Soil Science” (K. Kawaguchi, Ed.), pp. 293-299. Kodansha, Tokyo, Japan. Yoshida, S . (198 I). “Fundamentals of Rice Crop Science.” International Rice Research Institute, Los Banos, Laguna, Philippines. Yoshida, S., Forno, D. A., Cook, J. H., and Gomez, K. A. (1976). “Laboratory Manual for Physiological Studies of Rice,” 3rd ed. International Rice Research Institute, Los Banos, Laguna. Philippines. Yoshida, S . , Navasero. S. A., and Ramirez, E. A. (1969). Effects of silica and nitrogen supply on some leaf characters of the rice plant. Plant Soil 31, 48-56. Yoshida, S., Ohnishi, Y., and Kitagishi, K. (1962a). Histochemistry of silicon in rice plant. SoilSci. Plant Nutr. (Tokyo) 8 , 30-5 1 . Yoshida, S., Ohnishi, Y., and Kitagishi, K. (l962b). Chemical forms, mobility, and deposition of silicon in rice plant. Soil Sci. Plant Nutr. 8 , 107-1 1 I . Yoshida, S., Onshi. Y., and Kitagishi, K. (1959). The chemical nature of silicon in rice plant. Soil Plant Food 5 , 23-27. Yoshihara, T., Sogawa, K., Pathak, M. D.. Juliano, B. O., and Sakamura, S. (1979). Soluble silicic acid as a sucking inhibitory substance in rice against the rice brown planthopper. (Deplhucidae homotera). Enr. Exp. Appl. 26, 314. Zhang, H. L. (1984). Studies on soil silica. I n “Studies on Calcium and Silicon Fertilizer,” p. 96. Nanjiing Institute of Soil Science, Academia Sinica Press, Nanjiing, China. [English summary]

This Page Intentionally Left Blank

TISSUE CULTURE-INDUCED VARIATION AND CROPIMPROVEMENT R. R. Duncan University of Georgia, Department of Crop and Soil Sciences, 1109 Experiment Street, Griffin, Georgia 30223

I. Introduction 11. Causes and Range of Variation A. Occurrence of Somatic Variation B. Somatic Embryogenesis 111. Methodological Basis for Variation A. Explant Source B. Age of Culture C. Hormonal Factor D. Genotypic Factor E. PIoidy Factor F. Karyological Aberrations G. Transposable Elements H. DNA Methylation I. Additional Variation IV.Rate of Variation V. In Vitro Selection A. Selection in Culture B. Field Selection of Variants C. Phenotypic Variation D. Field Screening Techniques VI. Conclusions References

I. INTRODUCTION The applicability of biotechnology to crop improvement involves nonconventional plant breeding methodology. These new technologies will not replace the traditional breeding techniques but should mutually complement them by enhancing efficiency, trait-transfer precision, and recovery of useful, value-added 201 Advances in Agronmny, Volume 58 Copyright (0 1997 by Academic Press, Inc. AII rights of reproduction in any form reserved.

202

R. R. DUNCAN

variation (Lon et al., 1988; Karp, 1995). Useful variability can be generated without sexual recombination by in vitro somaclonal variation (Larkin, 1987). If somaclonal variation is to be effectively used as a selection tool in traditional breeding programs, breeders would benefit from knowing where and how variability is elicited during the in vitro process as well as how cell competence for totipotency, somatic embryogenesis, and regeneration capability are genetically controlled and inherited. However, breeders can exploit the variability without understanding it. Similar to conventional breeding programs, relative heritability estimates help to establish a framework for manipulation of in vitro traits. Determination of expected mutation rates for desirable in vitro traits helps to determine the sizes of plant populations necessary for selecting improved variants in the field. The breeder, when attempting to exploit somaclonal variation, should understand the advantages and disadvantages of in vitro selection (Karp, 1995) and must develop appropriate field selection strategies to effectively identify desirable variants in the field. Finally, as with any breeding program, the variants must be evaluated in multiple environments to ensure that all desirable traits are stable over generations. Useful variants have been released, but compared to conventional breeding methodologies, the results thus far have been limited and disappointing. The reasons for the limited success rate will be discussed in this chapter.

11. CAUSES AND RANGE OF VARIATION Somaclonal variation refers to all variability observed among tissue cultureregenerated plants (Larkin and Scowcroft, 1981). An early application of plant tissue culture was to asexually propagate plants rapidly and reliably with the premise that plants derived from a mother plant would be genetically identical to the original mother plant if the explant contained preformed meristems such as apical or lateral buds. However, clonal nonunifomity was the predominant response when adventitious pathways of in vitro regeneration occurred and this entire process offered another breeding tool for genetic improvement (Skirvin and Janick, 1976a; Shepard el al., 1980, Scowcroft, 1985). Tissue cultureregenerated variants have also been called calliclones (Skirvin and Janick, 1976a), phenovariants (Sibi, 1976), protoclones (Shepard er al., 1980), and subclones (Cassells et al., 1991). The selection of novel genotypes with stable, heritable mutations can be achieved with in vitro technology (Karp,1991). However, the success of this approach to crop improvement has resulted in a limited number of actual releases and subsequent exploitation of the variants in the practicaI production environment. The purpose of this chapter is to examine the origin and range of variation

VARIATION AND CROP IMPROVEMENT

203

generated, to document the progress, and offer realistic strategies for implementation and integration of somaclonal variation with traditional breeding methodology to enhance useful genetic manipulation and selection via tissue culture technologies. The chapter will address variation per se but exclude genetic engineering technologies because they add another layer of variability.

A. OCCURRENCE OF SOMATIC VARIATION In v i m culture per se can be extremely stressful on plant cells and involve highly mutagenic processes during explant establishment or callus induction, maintenance, embryo induction, and plant regeneration (Lorz et al., 1988). Two types of somaclonal variation can result: epigenetic (developmental) and heritable variation (Lorz et al., 1988; Skirvin et al., 1994). Epigenetic variation can be transient or temporary in later generations even when the material is asexually propagated. This variation includes phenotypic changes that involve expression of specific genes (Hartman and Kester, 1983). Because explants adapt to an in v i m environment in stepwise fashion by becoming more juvenile, the resulting calli may vary in maturity from juvenile to fully mature. Plants regenerated from these tissues also vary depending on the developmental stage progression of the tissue when the stimulus to regenerate is applied (Skirvin et al., 1994). Shoot regeneration from dedifferentiated callus can produce an immature, unstable clone that may eventually revert to the original parental clone. Examples of epigenetic variation include partial fertility, male sterility, or transient dwarfism (McPheeters and Skirvin, 1989; Moore et al., 1991) that are associated with a carryover of growth regulators from the tissue culture medium; tissue or cellular habituation (heritable variation in cultured plant cell requirements for hormones) involving the loss of auxin, cytokinin, or vitamin requirements by callus (Skirvin, 1978; Meins, 1982, 1989; Jackson and Lyndon, 1990), extreme vigor ex v i m (Swartz et al., 1981) associated with reversion to juvenility in which tissue culture-derived plants grow vigorously as juvenile seedlings until flower induction, or as a result of virus (Abo E1-Nil and Hildebrandt, 1971) or other microbe elimination. This juvenility vigor trait can be advantageous for rapid transplant establishment in horticultural crops (Swartz et al., 1981). Because broad spectrum variation is available through either nuclear or cytoplasmic sources, all types of variation could potentially be recovered and used for crop improvement (Elkonin et al., 1994; Skirvin et al., 1994). Heritable variation is stable through the sexual cycle and with repeated asexual propagation (Skirvin et af., 1994). Stable genetic alterations involve phenotypic and biochemical traits due to karyological variations (gene or chromosome abberations) (Lorz et al., 1988). Heritable somaclonal variation involves either single or

2 04

R. R. DUNCAN

multiple gene changes, including alterations in DNA bases, chromosomes, or the entire genome (Orton, 1984), single base-pair changes, paracentric and pericentric inversions, deletions, translocations, and ploidy changes (D’Amato, 1985, 1991; Evans and Sharp, 1983, 1986; Garcia et al., 1994) (Table I). Cytoplasmic variation has been found, including mitochondrially controlled male sterility (Gengenbach and Umbeck, 1982; Elkonin et al., 1994). Abnormal microsporogenesis affecting pollen sterility may be common and genotype dependent (Garcia et d., 1994). Somaclonal variation results from both preexisting genetic variation within the explants and variation induced during the tissue culture phase (Evans et al., 1984). Preexisting variation can evolve from nonuniformity in multicellular explants (multiple types of cells such as phloem, parenchyma, cortex, and xylem parenchyma). Explants derived from sources other than protoplasts produce complex cultures with variable ploidy levels (Bright et al., 1983). Axillary buds and shoot-tip cultures evolve from meristems of preformed origin and are much less variable than shoots arising adventitiously through either organogenesis or embryogenesis. Chimeras (Table 1) can result in the proliferation of callus that varies in genetic constitution or that develops from a meristem containing layers of mutated tissue (Dermen, 1960). Adventitious shoot production can cause chimeral dissociation (Marcotrigiano, 1990). If adventitious shoots can be traced to a single-cell derivation, chimeral segregation and induction of somaclonal variation can result (Karp, 1989). Chimera1 plants subjected to tissue culture can produce a high percentage of variants (Skirvin and Janick, I976a; Ramulu et al., 1976; McPheeters and Skirvin, 1983; Skirvin et al., 1994). Chimeric regenerant frequencies range from 10 to 70% in sorghum (Cai et al., 1990)and 54 to 79% in maize (Lee and Phillips, 1987b). However, chimeric regenerants in some plant species are rarely observed (Ammirato, 1983; Vasil, 1983). Variation may represent preexisting variation or variation induced during callus formation and not during the shoot formation process (Skirvin et al., 1994). In tobacco (Nicotiana tabacum L. ), 0.1- 1.8% of the variation was present originally in the mesophyll protoplasts, whereas 1.4-6.0% of the variation was attributed to tissue culture stress (Lon et al., 1988).

B. SOMATICEMBRYOGENESIS Callus initiation in v i m and subsequent somaclonal variation can occur (Morrish etal., 1987;Peschke and Phillips, 1992; Skirvin et al., 1994). Plant regeneration from somatic cells can occur by adventitious shoot formation and subsequent root organogenesis (from meristematic regions) or by somatic embryogenesis (formation of somatic embryos that germinate) (Bhaskaran and Smith, 1990). Variable levels of competence to totipotency are expressed by callus derived from somatic cells. Plants regenerated via somatic embryogenesis are usually

Table 1 Documented Reports on Genome Instability Induced by Culture Media or Environmental Conditions in Vitro That Generate Somaclonal VariationU Event

Cause

Reference

Disruption in mitotic cell cycle

Polyploids, aneuploids, hypo- or hyperdiploids Breakage + fusion --j bridge cycles Centric fusion Pseudodiploids, karyotypic changes

Ahloowalia ( I 982, 1983). Dyer (1976). Karp and Maddock ( 1 9 8 4 ~McCoy et a/. (1982) Sunderland (1977). Toncelli et a/. ( 1985) Murata and Orton (1983) Armstrong et al. (1983), Could (l982), Lapitan et a/. (1984). McCoy er a/. (1982) Burr and Burr (1981). Chandlee (1990). Fedoroff and Baker ( 1989). Groose and Bingham (1986a.b). Larkin and Scowcroft (1981). McClintock (1984). Nevers et a/. (1986). Peschke and Phillips (1991), Peschke et a / . (1987. 1991). Peterson (1993). Vodkin (1989) D'Amato (1952, 1985). Ramulu et a/. (1986). van den Bulk et a/. (1990). Van Harten e t a / . (1981) Ammirato (1983). Cai et a/. (1990). Cassells et a/. (1986). de Boucaud and Gaultier (1983). Lee and Phillips (1987b). McPheeters and Skirvin (1983). Skirvin and Janick (1976), Skirvin et a/. (1994). Ramulu et a/. (1976), Vasil (1983). Zehr et a/. (1987) Arnholdt-Schmitt et a/. (1995). Arnholdt-Schmitt (1995), Anderson et a/. (1990). Bianchi and Viotti (1988). Brown (1989). Brown and Lon (1987). Cedar and Razin (1990). Coulondre et a/. (1978). Hepburn e! a/. (1983), Hershkovitz er a/. (1990). Holliday (1987). Muller et a/. (1990). Munksgaard et a/. (1995). Nick et a/. (1986). Oberle e t a / . (1991). Perschke and Phillips (1992). Phillips et a/. (1990. 19921, Quemadd et a/. (1987), Yisraeli and Szyf (1984)

Chromosome rearrangements Robertsonian fusion Somatic crossovers

Transposable elements

Abberations

Pol ysomatry

Diploid and polyploid cells coexist in the same tissue Abberations

Chimeras

DNA methylation

Genome reoganization

'' Partially summarized from Morrish et a/. (1987).

206

R. R. DUNCAN

less variable than plants regenerated via shoot and root morphogenesis (Lorz et al., 1988). Both developmental pathways have been successful in producing regenerated plants from numerous species. The developmental pathway of somatic embryos is similar to zygotic embryos in plants (Ammirato, 1983; Stange, 1984). In organogenesis, shoots and roots evolving from dedifferentiation (callus) develop independently with respect to localization and timing (shoots preceding roots) (Lorz et al., 1988). Somatic embryos have a bipolar structure in which shoot and root meristems are directly connected with no interruption by nondifferentiated callus tissue (Lorz et a l . , 1988). Plants regenerated by somatic embryogenesis contain fewer mutations than those regenerated by organogenesis primarily because the embryoid is generally derived from a single cell (Haccius, 1978; Vasil and Vasil, 1982; Ho and Vasil, 1983) or from a small group of cells (Browers and Orton, 1982; Vasil et al., 1985). Furthermore, the plants regenerated from somatic embryoids contain few mutations or chimeras due to stringent internal genetic controls imposed during embryoid formation, causing selection pressure against abnormal types (Swedlund and Vasil, 1985). Although plants regenerated through tissue culture in some species are less variable than their original donor or explant sources (Feher et id.,1989; Gmitter et al., 1991), enhanced variability (both phenotypic and cytological) from embryogenically regenerated plants over organogenically regenerated plants from the same explant source have been documented (Armstrong and Phillips, 1988). Other cases of variable embryogenically regenerated plants have been reported (Browers and Orton, 1982; Ahloowalia and Maretzki, 1983; Karp and Maddock, 1984). The variability may be caused by a constant mutation rate per cell generation with a multiplicative effect due to an increased number of generations in vitro (Peschke and Phillips, 1992). Pnmary somatic embryogenesis from vegetative explants is an indirect process in which embryos are formed up to the preglobular stage after exposure to auxin or auxin plus cytokinin-supplemented media (Raemakers et al., 1995). Direct somatic embryogenesis from zygotic embryos in certain dicot species is initiated by cytokinin-supplemented or growth regulator-free media. These embryos develop to mature stages before being subjected to secondary somatic embryogenesis (embryos formed from embryos). Secondary embryogenesis requires no growth regulators in species with cytokinin-driven primary embryogenesis (Raemakers er al., 1995) but may require wounding of the primary embryos (Smith and Knkorian, 1989). In some species, wounding may not be obligatory, but it may increase the number of secondary embryos formed (Baker and Wetzstein, 1992). Compared to primary embryogenesis, secondary somatic embryogenesis has the advantageous capability for high multiplication rates, independence of explant sources, repeatability, maintenance of ernbryogenicity for prolonged periods of time, and reduced probability for variation (Raemakers et al., 1995).

VARIATION AND CROP IMPROVEMENT

207

111. METHODOLOGICAL BASIS FOR VARIATION

A. EXPLANTSOURCE Explant sources vary in their ability to generate somaclonal variation (Skirvin et al., 1994). Highly differentiated tissue (roots, leaves, and stems) produce

more variation than explants with preexisting meristems (axillary buds and shoot tips) (cited in Bayliss, 1980; Shepard et al., 1980; cited in D’Amato, 1985; cited in Karp and Bright, 1985; Potter and Jones, 1991; Tsai et a/., 1992; Gui et al., 1993). Cultures give rise to normal or near-normal regenerated plants when no significant dedifferentiation (callus) step occurs (Peschke and Phillips, 1992). Organogenesis can involve more than one cell (Springer et al., 1979). Sectored and nonsectored maize (Zea mays L.) plants regenerated via organogenesis from one explant source contained the same chromosomal aberration (Benzion and Phillips, 1988), indicating the variant was produced in the callus and that the sectored explant was multicellular in origin. Stability differences in tissue cultures originating from different explant sources often are caused by preexisting variability in the explant source (Peschke and Phillips, 1992). For example, polysomaty (Murashige and Nakano, 1966, 1967) (coexistence of diploid and polyploid cells in the same tissue) can be found in over 90% of plant species (D’Amato, 1952, 1985). Because callus are maintained in subculture, polyploidy may increase (Murashige and Nakano, 1966, 1967). Hypocotyls are polysomatic, whereas cotyledons and leaves are predominantly diploid in tomato (Lycopemicon esculentum Mill.) (van den Bulk et al., 1990). When protoplasts are developed from chromosomally variable cell suspensions rather than freshly isolated diploid tissues, protoplast-derived potato (Solanum tuberosum L.) regenerants had a higher frequency of phenotypic and chromosomal aberrations (Ramulu et al., 1986). More chlorophyll-deficient variants are recovered from immature embryo cultures than from mature embryo cultures (Cai et al., 1990). The culture of juvenile, meristematic tissues are generally more efficient for plant regeneration.

B. AGEOF CULTURE Culture age enhances variability of regenerated plants (Shepard et al., 1980; McCoy et al., 1982; Thomas et al., 1982; Lee and Phillips, 1987a,b; Armstrong and Phillips, 1988; Benzion and Phillips, 1988; Muller et al., 1990; Symillides et al., 1995). This age effect (Peschke and Phillips, 1992) can be attributed to (i) increased mutation rate per cell generation over time (Murashige and Nakano, 1965) and accumulation of mutations over time, (ii) a lag period of apparent culture stability caused by early generation mutations that are not detected until

208

R. R. DUNCAN

sufficient mutant cells have accumulated (Benzion and Phillips, 1988), (iii) active selection of early culture mutations that increase in number over time (Armstrong and Phillips, 1988),or (iv) increased polyploidy (Murashige and Nakano, 1966, 1967;Colijn-Hooymans et af., 1994). Regeneration competence of cucumber (Cucumis sativus L.) may decrease and genetic instability may increase due to the simultaneous increase in DNA content and age of cotyledonary cells (Colijn-Hooymans et al., 1994).Many mutations occur after short culture periods but are difficult to detect in regenerated plants (Armstrong and Phillips, 1988). As callus induction time increases, the morphogenesis potential decreases, whereas the frequency of albino shoots and callus producing only roots increases (Wen et al., 1991).Some mutations may occur in sequence throughout the callus phase rather than occurring randomly or all together at an early culture stage (Fukui, 1983). The highest percentage of chlorophyll variations found in regenerated plants from inflorescence cultures came from 2 10+ day old cultures (Cai et al., 1990).

C. HORMONAL FACTOR Callus is initiated in vitro on cut or exposed cell surfaces in contact with a growth medium. Callus proliferation is a wound response. Excision of the explant stimulates the wound responses in vivo (induction of stress-induced enzymes and other chemical by-products, and activation of transposable elements), which can be enhanced by growth regulators (McClintock, 1984). Most plant growth regulators, and specifically 2,4-dichlorophenoxyaceticacid (2,4-~) and 6-benzylaminopurine have been implicated in tissue culture-induced variability (D’Amato, 1985;Evans, 1988;Griesbach et af., 1988;Shoemaker et al., 1991). Regeneration competence in Gramineae may be rapidly lost during differentiation and senescence (Ozias-Akins and Vasil, 1988). This loss may be due to the endogenous concentrations of growth regulators (Vasil, 1987).Tissue cultureinduced variability rates,can increase as growth regulator concentrations increase (Skirvin et al., 1994). The primary event causing tissue culture-induced variability may be cell cycle disturbance (Peschke and Phillips, 1992)caused by exogenous hormone effects (Bayliss, 1977;Bhaskaran and Smith, 1990;Liscum and Hangarter, 1991)or nucleotide pool imbalances (Jacky et al., 1983). Auxins can produce rapid disorganized growth during callus induction that may lead to genetic instability through asynchronous cell divisions (Gould, 1984; Lee and Phillips, 1988). Increased thymidine can enhance chromosome breakage (Ronchi et al., 1986), but the nucleotide pool balance may be more important than the absolute quantity of a specific precursor or nucleotide (Meuth et al., 1979;Weinburg et al., 1981). Sister chromatid exchange frequency can increase with a low concentration of

VARIATION AND CROP IMPROVEMENT

209

2 , 4 - ~(Dolezel et al., 1987). Chromosome breakage could lead to aneuploidy (via chromosome fragment loss), activation of transposable elements (pieces of DNA that move within and between chromosomes), methylation changes (Grafstrom et al., 1984), chromosome breakage (McClintock, 1978), initiation of SOS responses resulting in single base changes (Walker, 1984), and an error-prone repair system (Burr and Burr, 1988). Deamination of 5-methylcytosine causes its conversion to thymine and subsequent production of point mutations (Coulondre et al., 1978). Tissue culture competence may be due to genes involved in plant hormone metabolism (Henry er al., 1994b), which may explain why some cultivars were difficult to regenerate in one laboratory, but a different laboratory (with a different environment or hormonal level in the media) was successful. Some cultivars that supposedly contained no favorable alleles for regeneration have subsequently been regenerated in other studies (Bhaskaran and Smith, 1990). In vitro genetic control of regeneration could be functioning specifically on responses to plant growth regulators (Close and Gallager-Ludeman, 1989; Komamine et al., 1990). Genes controlling phytohormone signals are directly involved in plant regeneration competence (Henry et al., 1994b), negating the concept that “dominant” genes govern regeneration and that all cultivars have genes for regeneration.

D. GENOTYPIC FACTOR Cultivars vary in their ability to produce regenerable cultures, with differences enhanced by the degree to which the in v i m environment disrupts the unique cellular environment (Table I) (Peschke and Phillips, 1992). Some cultivars may have genes that control tissue culture regeneration; others may not have the genes for regeneration or the genes controlling phytohormone signals, but the trait can be transferred via traditional breeding methods into these genotypes (Smith and Quesenberry, 1995). Inheritance of somatic and microspore tissue culture responsiveness can be traced to nuclear genes (Pechan et al., 1991; de Buyser et d . , 1992). Genes on different chromosomes are involved in control of callus growth (Shimada and Makino, 1975; Baroncelli et al., 1978) and in regeneration of shoots (Galiba eta/., 1986; Mathis and Fukui, 1986; Mathis et al., 1988; Kaleikau et al., 1989), suggesting a polygenic system (Henry et al., 1994a) governing induction of cells involved in embryogenesis. The regeneration process includes a reprogramming of gene expression, control of differentiation of induced cells into embryos (with a few genes controlling the embryo germination mechanism and others triggering the regeneration process by acting on the quality of the embryos), and maintenance of embryogenic capacity and the ability to regenerate plants through successive long-term subculturing. Certain genes have a major effect on somatic embryogenesis and regeneration (Brown and Atanassov, 1985),

2 10

R. R. DUNCAN

whereas the lack of certain genes on other chromosomes may suppress embryogenesis (Henry et al., 1994a). Genetic control of somatic embryogenesis in alfalfa (Medicago sativa L.) was found to be under the control of two dominant loci (Crea et al., 1995), whereas some nonembryogenic genotypes when crossed with embryogenic plants did not regenerate. Additional genes may determine the time required for regeneration. Different polygenes are involved in the rye tissue culture response from different explants (Rakoczy-Trojanowska and Malepszy, 1995). Embryogenic callus production, shoot regeneration, and root regeneration are controlled by recessive genes and are inhibited by dominant suppressors. Production of nonembryogenic callus is determined by dominant genes that can give an additive effect. The lack of in vitro response is caused by at least two interacting genes acting in a suppressive fashion in rye. Genetic analysis of regeneration ability has also revealed that the trait can be controlled by a few genes (one to three loci) with quantitative and highly heritable effects (Reish and Bingham, 1980; Charmet and Bernard, 1984; Brown and Atanassov, 1985; MacLean and Nowak, 1989; Nadolska-Orczyk and Malepszy, 1989); polygenic control strongly influenced by environment (Tomes and Smith, 1985); polygenic control with little environmental influence (Rakoczy-Trojanowska and Malepszy, 1995); dominant genes (Komatsuda et al., 1989; Nadolska-Orczyk and Malepszy, 1989; Reish and Bingham, 1980) having positive heterotic effects (Flehinghaus et al., 1991); complementary gene control of callus production and regeneration suppression; plant regenerative ability determined by numerous loci; and recessive control of regeneration (Rakoczy-Trojanowska and Malepszy, 1993, 1995). The range of variable gene control for tissue culturability among 17 different plant species has been summarized (Bhaskaran and Smith, 1990; Henry et al., 1994b). Competencies for induction of embryogenic calli and eventual plant regeneration are both determined by genes with major and minor effects in a polygenic system. Narrow sense heritability estimates range from 40 to 50% (Quesenberry and Smith, 1993), and broad-sense heritability of regeneration ability ranges from 36 to 50% (RakoczyTrojanowska and Malepszy, 1993). Heritability ranges of 9 to 60% (embryo germination), 30 to 55% (callus induction), and 15 to 49%(embryogenic callus formation) have been documented in wheat (Chevrier et al., 1990). Broad-sense heritability for embryogenic callus-plant regeneration in rye ranged from 0.59 to 0.93, depending on genotype (Rakoczy-Trojanowska and Mdpeszy, 1995). Additive genetic effects for callus morphology and differentiation also have been documented (Keyes et al., 1980; Chevrier et al., 1990).

E. PLOIDY FACTOR Regenerated plant variability is higher among polyploid and high-chromosome number explant sources (Heinz and Mee, 1969, 1971; Creissen and Karp, 1985;

VARIATION AND CROP IMPROVEMENT

211

Skirvin et af., 1994) than among lower ploidy and low-chromosome number species. Spontaneous doubling is a common type of chromosome aberration in many diploid dicotyledonous regenerated plants (Murashige and Nakano, 1966, 1967), whereas tetraploids regenerated from callus or protoplasts often have chromosome structural changes, aneuploidy, and chromosome doubling (Bingham and McCoy, 1986). Polyploidy in tissue culture is generally the product of either endopolyploidization or nuclear fusion (Sunderland, 1977; Bayliss, 1980). Aneuploidy may be caused by nondisjunction, aberrant spindles, lagging chromosomes, chromosome breakage that produces dicentric and acentric chromosomes (Sunderland, 1977), or polyploidy followed by chromosome elimination with lobed (constricted) nuclei (Balzan, 1978). Aneuploid regenerants are more common in polyploids of some plant species than in diploid versions in which the trait is rarely transmitted to progeny generations (D’Amato, 1985; Peschke and Phillips, 1992). In some plant species, chromosome rearrangements are more common than ploidy changes (Ashmore and Could, 1981; Murata and Orton, 1983).

E KARYOLOGICAL ABERRATIONS Chromosome abnormalities of in v i m regenerated plants can include numerous karyological changes (Karp and Bright, 1985; Lorz et al., 1988). Chromosome breakage and its consequences (deletions, duplications, inversions, and translocations) cause common aberrational events (Sacristan, 1971 ; cited in Lee and Phillips, 1988). The breakage positions are not random, but involve latereplicating chromosome regions characterized by heterochromatin (Lima-de-Faria, 1969; Sacristan, 1971; McCoy et al., 1982; Lapitan et al., 1984; Murata and Orton, 1984; Johnson et af.,1987; Lee and Phillips, 1987; Benzion and Phillips, 1988). The role of heterochromatin in causing chromosome breakage may be due to later-than-normal chromosome replication in tissue culture, with the chromatids being held together and creating a stress that results in breakage between the centromere and late-replication region (McCoy et al., 1982; Lee and Phillips, 1988). “Genomic stress” caused by a broken chromosome could activate transposable elements in the cell (McClintock, 1978, 1984). Chromosome breakage may create mutations directly through “position effect” or an alteration in gene expression due to chromosomal rearrangement and placement in close proximity or distanced from a specific heterochromatic region (Catcheside, 1939; Spofford, 1976; cited in Peschke and Phillips, 1992). Tissue culture has also resulted in increased frequencies of sister chromatid exchange (Dolezel and Novak, 1986; Dimitrov, 1987; Dolezel et al., 1987) and somatic crossing-over involving two homologs instead of two chromatids (Lorz and Scowcroft, 1983). Homozygous variability was stable only when the plants were selfed and in the homozygous condition, but was unstable after crossing to create heterozygosity

2 12

R. R. DUNCAN

(Oono, 1985; Nowich et a / . , 1988; Xie, 1990). Somaclonal regenerants can produce homozygous variants for certain traits (height, disease resistance, and seed yield) during the sexual F, generation (Sun et al., 1983; Xie, 1990; Cai er al., 1990) when crossed to the donor parents or to normal-type somaclones (OORO,1985; Nowick et al., 1988; Larkin et al., I 989). Homozygous somaclonal variants have been reported in rice (Oryza sativa L.) (Xie, 1990; Xie et al., 1995), wheat (Triticurn aestivurn L.) (Larkin et al., 1984; Larkin, 1985; Maddock et al., 1985), lettuce (Latuca sativa L.) (Engler and Grogan, 1984), potato (S. tuberosurn L.) (Gavazzi er al., 1987), and tomato (L. esculenturn Mill.) (Evans and Sharp, 1983). A higher mutation frequency at a particular locus could significantly increase the variability of both alleles at that locus (Xie et al., 1995). Somaclonal variation may not be random because specific loci may have higher mutation rates than others during the in vitro process (Xie et al., 1995).

G. TRANSPOSABLE ELEMENTS A transposable element is a DNA sequence with the capability for movement throughout the genome by a process of excision and reintegration (Chandlee, 1990). Activation of transposable elements may generate stable or unstable single-gene mutations, methylation changes in and eventual activation of transposable elements, as well as additional chromosome breakage (Peschke and Phillips, 1992). Autonomous transposable element activation may be a symptom of genetic instability in tissue culture-regenerated plants or their progeny. Transposable elements move from one position to another, interrupting gene function when inserted into genes (Peterson, 1993). Insertion of a transposable element in a host site adds nucleotides to the genome (induces a target site duplication). Altered nucleotide sequences result from an excision process (a protein-DNA complex in which endonuclease enzymes excise the element), affecting the gene reading frame and altering the template enough to change the protein coding (Fedoroff and Baker, 1989; Vodkin, 1989; Peterson, 1993). Imprecise excisions are a key to genetic variability.

H. DNA METHYLATION Chromosome structure, methylation, and phenotype are complexly interconnected (Oberle er al., 1991). Gene expression may be altered during somatic embryogenesis (Goldberg, 1986; Borkind et al., 1988; Reinbothe et al., 1992). Methylation can enhance quantitative trait variation because several genes can be affected simultaneously (Phillips et al. 1990). Increases in methylation in vitro

VARIATION AND CROP IMPROVEMENT

213

potentially escalate gene activity (Hepbum et al., 1983; Bianchi and Viotti, 1988) and regulation. Methylation of a gene inactivates its transcription (Cedar and Razin, 1990; Palmgren et al., 1991) and thereby controls gene expression during somatic embryogenesis (Amasino et al., 1990; Cedar and Razin, 1990). A hypomethylated state of DNA is essential for the acquisition and release of somatic cell embryogenic potential (Munksgaard et al., 1995). Hypomethylation of DNA in somatic cells induces an early zygotic embryo state of differentiation (Okkels, 1988). The DNA in the first developmental stages of somatic embryogenesis contains lower levels of methylated DNA than that in later embryo stages (Lo Schiavo et af., 1989; Munksgaard et al., 1995). S-adenosylmethionine (SAM) donates the methyl groups involved in the reaction leading to nucleic acid methylation, as well as the aminopropyl moiety for polyamine synthesis. SAM is metabolized to S-adenosylhomocysteine (SAH), which inhibits DNA methyltransferase (Guitton et al., 1988). Polyamine (Feirer et al., 1984; Fienberg et af., 1984; Mengoli et al., 1989) and DNA methylation levels increase during somatic embryogenesis, indicating SAM regulation of the synthesis of these metabolites (Munksgaard et al., 1995). The methylation index (SAM/SAH ratio) controls cellular utilization of SAM for methyltransferases (Guitton et al., 1988; Yarlett and Bacchi, 1988). The decrease of the methylation index (SAM decrease and SAH increase) after 2 , 4 - ~removal (for subsequent differentiation into embryos and eventual plantlets) may cause nucleic acid hypomethylation because a low SAMlSAH ratio inhibits the SAM-dependent methylation processes. Consequently, both SAM and SAH levels are critical for control of hypomethylation, which in turn may affect differential changes in gene activation. SAM can increase five-fold during somatic embryogenesis and may be a prerequisite for progression of embryogenesis and eventual development of complete embryos (Munksgaard et al., 1995). Homozygous (nonsegregating) variants do not occur randomly and tissue culture may enhance specific mutations involving certain DNA sequences (Xie et af., 1995). Gene mutations induced by nonrandom DNA methylation (Coulondre et al., 1978; Barker et af.,1984; Brown, 1989) cause homozygous variants in the F, sexual generation, but altered gene expression can result from other sectorspecific methylations during tissue culture (Xie et al., 1995). Genome demethylation of carrot occurs during the transition to stationary growth, which indicates differential genomic methylation during different phases of in v i m culture (Amholdt-Schmitt et ul., 1995). Tissue-specific methylation of genomic DNA is influenced by genotype, culture age, plant growth, and primary cultures during the transition from linear to stationary growth. Dedifferenitation at the genome level by nonspecific de n o w methylation is followed by demethylation, which is related to specific tissue reprogramming at the molecular level (Amholdt-Schmitt et al., 1995). A phase of genome instability during culture occurs from the nonrandom, physiological, quantitative variability of

214

R. R. DUNCAN

DNA sequences in the carrot genome during cell division growth of primary cultures (Amholdt-Schmitt, 1995). Active elimination of repetitive sequences could shorten replicon sizes, lower nucleus DNA contents, and preferentially provide stress-induced heritable changes in genome organization. Differential replication (Amholdt-Schmitt, 1995) coupled with DNA methylation pattern changes (Amholdt-Schmitt et a f . , 1995) are normal physiological processes in dividing in vitro cells.

I. ADDITIONAL VARIATION Tissue culture-induced DNA aberrations have resulted from single nucleotide base changes (A conversion to T) and the production of an electrophoretic somaclonal mutant (amino acid change) (Brettell et al., 1986) or a null mutant (production of a premature codon stop) (Dennis et al., 1987). Gene copy product amplification resulting from tissue culture has been observed (Donn et al., 1984; Lapitan et al., 1988; Brown, 1989; Brown et al., 1991). Ribosomal DNA copy number may decrease in v i m (Landsmann and Uhrig, 1985; Brettell etal., 1986; Cullis and Cleary, 1986), and an environmental stress created by a nutrient imbalance can enhance the reduction (Cullis, 1976). Long-term callus culture can be an excellent source of somaclonal variants (Skirvin et al., 1994), and specific cultural conditions, such as maltose or lactose media, can be used to enhance regenerability (Asano et a f . , 1994). Accelerated culture proliferation rate in banana (Musa X paradisiaca L. var. paradisiaca) will also increase the number of mutations (Smith and Drew, 1990).

W. RATE OF VARIATION With normal mutation rates at a specific locus varying from 1:100,000 to 1: 1,000,000(Skirvin et af., 1994) and over 1700 mutant cultivars being officially released involving 154 plant species (Maluszynski et a f . , 1995), pooled random somaclonal variants may total as high as 100%(Orton, 1983; Orton, 1987). A normal variation range of 3 to 26% may be common in some plant species (Larkin et a f . , 1984) when compared to the donor explant chromosomal condition. Variants commonly occur at 1520%frequency, depending on the genotype and the explant source from that genotype (Evans and Sharp, 1983). A more realistic mutation rate in vitro may be 1-3% (Skirvin et a f . , 1994). Single-gene mutations reportedly occur at a frequency of one mutant for every 20-25 regenerated plants (Evans, 1988). In addition, some sector mutations do not appear until the R, generation (Cai et al., 1990).

VARIATION AND CROP IMPROVEMENT

215

Somaclonal variation rate (number of variants in R, families t total number of R, families) in sorghum (Cai et al., 1990) ranged from 1.8 to 6.8% (average 4.6%) for chlorophyll variations, 0.9 to 25% (average 6.7%) for viable variations, and 2.7 to 31.3% (average 11.3%) for total variations. Somaclonal variant total number of R, plants) ranged from 0.39 frequency (number of R, variants i to 1.54% (average 0.72%) for chlorophyll variations, 0.08 to 4.85% (average 0.87%) for lethals, and 1.02 to 6.39% (average 1.59%) for total variations in sorghum (Cai et al., 1990). Compared to documented somaclonal mutation rates in sorghum for the same traits (Ramulu, 1973, the range for chlorophyll mutations was 4-43.6%. The mutant frequency ranged from 0.4 to 9.9% for chlorophyll mutants and 0.55 to 1.37% for viable mutants. Somaclonal variation rates are similar in rice (0.sativa L.; Sun et al., 1983; Adkins et al., 1995), wheat (T. aestivurn L.; Maddock and Semple, 1986), barley (Hordeurn vulgare L.; Pickering, 1989), sorghum (Ma et al., 1987), and maize (2. mays L.) (Zehr et al., 1987; Armstrong and Phillips, 1988).

V. Z N VITRO SELECTION The frequency of total somaclonal variation should be high enough for selection of agronomically desirable variants (with one or two traits different from the original donor parent). However, the ultimate test of any tissue culture-derived progeny in a breeding program is field evaluation under multiple environments. Subjecting somaclonal variants to performance evaluations in the field includes additional genotype X environment interactions. The [in vitro X genotype X explant source] interactions contribute to one level of variation. When in vitro selective agents are added to the protocol, additional interactions are encountered (Handro, 1981). Transformation adds an additional interaction component (Conner and Meredith, 1989). If biotic or abiotic field stresses are used to evaluate the somaclonal variants, these multiple in vitro process X environment interactions will govern the success or failure of selection for useful, positive, agronomically beneficial variants. Breeding programs must adjust to these additional interactions to successfully select and eventually transfer the useful variants to production systems.

A. SELECTION INCULTURE A major challenge for exploitation of somaclonal variation is the identification of useful variation (Chaleff, 1983; Lee et al., 1988; Evans, 1989; Jones, 1990; Smith et al., 1993). Because cells are genetically variable in culture, specific

2 16

R. R. DUNCAN

traits can be selected using in vitro-induced spontaneous mutations (Larkin and Scowcroft, 1981; cited in Maliga, 1984). However, cell cultures do not have to be exposed to in v i m selective agents to develop novel somaclonal variants with useful agronomic traits (Maliga, 1984; Evans and Sharp, 1986). In vitro selection of many abiotic constraints on a cellular level is complex and requires standardization and efficient methodology to exclude artifacts (Wersuhn et al., 1994). Theoretically, a specific selective agent can be used in vitro to select mutationspecific cell lines, thereby reducing the number of plants necessary for screening at the whole plant level under stress in the field (Maliga, 1984; cited in McCoy, 1988). By screening large populations of cells in vim, breeding efficiency would (supposedly) be increased and overhead costs should decrease. The principle criterion necessary for improved genotype selection using this technology depends on expression of a stable target genetic trait while other desirable traits remain unchanged, effective selection strategies in the field that are focused on identification of the traits, and discriminating field environments that allow trait expression (Smith et al., 1993). Cells may be selected in either a positive or a negative manner using direct (cells cultured with selective agents followed by isolation of surviving cells), rescue (cell exposure to culture conditions capable of killing or inhibiting susceptible cells followed by culture under different conditions to recover survivors), and stepwise (gradual increase in selection pressure over time) strategies (Conner and Meredith, 1989). In vitro selective agents, such as NaCl (Nabors et al., 1975, 1980; Hasegawa et al., 1980; Rains et al., 1980; Smith et al., 1983; Bhaskaran et al., 1986; Ye et al., 1987; Yang et al., 1990; Jain et al., 1991; Zapata et al., 1991; Piqueras et al., 1994) for salt tolerance, AlCl, (Smith et al., 1983; Conner and Meredith, 1985; Miller et al., 1992; Rao et al., 1992) for the aluminum toxicity portion of acid soil tolerance, and polyethylene glycol (Bressan et al., 1981, 1982; Handa et al., 1982; Hasegawa et al., 1984; Bhaskaran et al., 1985; Newton et al., 1986; Smith and Bhaskaran, 1988; Waskom et al., 1990; Adkins et al., 1995) as an osmoregulator at the cellular level to simulate drought stress at the whole plant level have been used to develop regenerant plants with salt, acid soil, and drought stress tolerance improvements that exceeded the donor parents. Pathogen toxins can be used as a selective agent on cells in culture (Helgeson et al., 1972; Haberlach et al., 1978; Behnke, 1979; Cassells et al., 1980, 1986, 1991; Meulemans et al., 1987; McCoy, 1988; Jones, 1990; Wenzel and Foroughi-Wehr, 1990; Cao et al., 1991; Oberthur et al., 1993). Pathogen cultures have also been effective in screening regenerated plants (Heath-Pagluiso et al., 1988). Tissue culture can also be used to screen calli for nutrient imbalances: boron-zinc-manganese deficiencies (Graham et al., 1993), phosphorus efficiency (Bagley and Taylor, 1987), and iron deficiency (Stephens et al., 1990; Graham et al., 1992; Kawai and Takagi, 1992). Insect resistance can be improved using in vitro techniques (Croughan and

VARIATION AND CROP IMPROVEMENT

217

Quisenberry, 1989; Lentini et al., 1990; Isenhour and Wiseman, 1991; Isenhour et al., 1991). Cold hardiness can be improved using in vitro selection techniques (Chen and Gusta, 1983; Duncan and Widholm, 1987; Reaney and Gusta, 1987; Robertson et al., 1987; Galiba and Sutka, 1989; Reaney et al., 1989; Kendall et al., 1990; Churchill et a l . , 1992; Nowak et al., 1992; Xin and Li, 1992, 1993; Dorffling et al., 1993). Herbicide resistance can also evolve from somatic cell selection (Chaleff and Ray, 1984; Jordan and McHughen, 1987; Saxena and King, 1988; Swanson et al., 1988; Anderson and Georgeson, 1989; Harms and DiMaio, 1991; Saunders et al., 1992).

B. FIELDSELECTION OF VAFUANTS Haploidization, somatic embryogenesis, cell suspension culture, and protoplast technology all require plant regeneration techniques for genetic manipulation and subsequent selection in plant breeding programs (Henry et al., 1994b). The entire in vitro process from genotype and explant selection to media-induced callus, embryo, and plantlet regeneration is extremely productive in creating variation. However, for the major worldwide effort incorporating in vitro technology in breeding research programs since the 1970s, relatively few agronomically useful cultivars or germ plasm sources have been released for utilization in production programs (Table 11). Horticultural crops were the first (1976) releases, and the agronomic crops (flax, wheat, rice, maize, and tobacco) became available during the mid- 1980s and early 1990s. Nevertheless, somaclonal variation has not been incorporated as a standard protocol in most plant breeding selection schemes. Perhaps this limited integration can be traced to the segregation of biotechnology and traditional breeding laboratories or the limited understanding of molecular events (additional “noise”) that trigger in vitroinduced variation coupled with strong beliefs in traditional breeding methodology. Because the variants are mostly chromosomal abberations, breeders often question the efficiency and effectiveness of using the in vitro system in their breeding programs. Regardless, the successes outlined in Table I1 demonstrate that positive variants can be harnessed and that agronomically useful variation can emerge from selection programs.

C. PHENOTYPIC VARIATION For cross-pollinating forage species, such as Lotus corniculatus L., fieldgrown regenerants demonstrated significant morphological variation during the

Table II Cultivars Derived from Tissue Culture-Induced Somaclonal Variation Crop

O0

Type"

Binomial

Cultivar

Improved traitb

Velvet rose

F

Pelargoniwn graveolens L'Heritier ex. Aiton Rubus laciniarus Willd.

LQganberry Lisianthus (prairie gentian or bluebells) Empress or princess tree Daylily Wishbone flower or bluewings

V F 0

Ipomoea batatas (L.) Lam. Rubus loganobaccus L.H. Bailey Eustom grandgorum L.

Scarlet Lincoln logan Unnamed

Dark skin color Thornless Dwarfness, basal branching

0

Puulownia romentosa (Thunb.) Steud. Hemerocallis lilio-asphodelus L. Toreniu fournieri L.

Somaclonal snowstorm Yellow tinkerbell UConn whte

Celery

V

UC-T3 somaclone

Asparagus

V

Apium graveolens L. var. duke (Miller) Pers. Asparagus oficinalis L.

Flax

A A

Linum usitatissimum L. Sorghum bicolor (L.) Moench

Andro GATCCPe

Irregular variegated foliage Dwarfness White flower color, compact growth habit Fusarium wilt resistance Anther-derived doubled haploid Rust immunity Fall armyworm resistance

Geranium

0

Blackberry sweet potato

Sorghum

0 0

Evergreen

100/101

Sturdiness, vigor, attractiveness Thornless

ReferenceC Skirvin and Janick. 1976b Waldo, 1977 McPheeters and S k i n , 1983, 1989 Moyer and Collins, 1983 Hall er al., 1986 Griesbach and Semeniuk, 1987 Griesbach ef a[., 1988 Marcotrigiano and Jagannathan, 1988 Griesbach, 1989 Brand and Bridgen, 1989

Heath-Pagliuso et al., 1989 Heath-Pagliuso and Rappaport, 1990 Comols et ul., 1990 McHughen and Swartz, 1984 Duncan et al., 1991a Isenhour et al., 1991

Sorghum

Wheat

A

A

Sorghum bicolor (L.) Moench

Triticum aestivum (L.)

GAC 102e

Acid soil tolerance

GC 103/104e

Acid soil tolerance

KS89WGRC9e

Heatldrought stress tolerance High yield Doubled haploid Powdery mildew resistance Barley yellow dwarf virus resistance Anther-derived inbred Unknown Unknown Unknown Unknown Rice blast resistance

Jinghua 1 Florin Anther culture 28 TC5, TC6, TC96 Maize Rice

A

Zeu mays L.

A

Oryza sativa L.

Rice

A

Oryza sativa L.

N

Huayu 1 Hua Yu I, I1 Xin Xiu Late Keng 959 Tung Hua I , 2, 3 Zhonghua 8, 9 Zhonghua 10 Zhonghua 3, 6, 11 Zhonghua 12 Zhonghua 13 Zhonghua 14

“Wide spectrum resistance” “Wide spectrum resistance” Rice blastlstripe virus resistance Bacterial leaf blight resistance Bacterial leaf blight re. sistance

Duncan et al., 1991b Waskom et a / . , 1990 Duncan et a / ., 1992 Foy et a / . . 1993 Miller ef al., 1992 Sears et al., 1992 Daofen, 1986 de Buyser et a/., 1987 Zhao et a / ., 1990 Banks and Larkin, 1995 Wu, 1986 Hu and Zeng, 1984 (1976)f Loo and Xu, 1986 (1976) Loo and Xu, 1986 (1978) Loo and Xu, 1986 (1978) Chen, 1986 (1980) Li et al., 1994 (1980) Li et al., 1994 (1982) Li et a / ., 1994 (1984) Li et 01.. 1994 (1986) Li el a / . , 1994 (1991) Li er a / ., 1994 (1992)

continues

Table 11-continued ~~

Crop

Typea

Binomial

Cultivar Guxing 1 Zixiang 1, 2 Zao D. ShanA

Hua Han Zao Huajian 7902 Nanhua 5

No11

Tobacco

A

Nicotiana tabacum L.

Hua 03 Gum 18 Delfieldd Tan Yuh 1, 2, 3 NC 744e KDH 926, 959, 960e NCBMR 42, 90e

Bermuda grass

T

Cynodon dactylon (L.) Pers.

Braz0s-R3~

Birds-foot trefoil

A

Lotus corniculatus L.

H 401-4-4-2

Red clover

A

Trifolium pratense L.

NJZWRC

0, ornamental; F, fruit; V, vegetable; A, agronomic; T, turfgrass. TC, tissue culture; CMS, cytoplasmic male sterility. The Chinese release reports are summarized in Veilleux (1994). Hybrid. Germ plasm. f Year in parentheses is year of actual release. a

Improved traitb Unknown

Purple color, aromatic, soft kernel CMS Multiple parent traits Unknown Unknown Unknown High protein Unknown Somatic hybrid, high yield Anther-derived haploids Potato virus Y resistance Trichome exudates Blue mold resistance

Fall armyworm resistance Sulfonylurea herbicide resistance Tissue culture regeneration ability

~

ReferenceC Li er al., 1994 (1993) Li ef al., 1994 (1993) Cai et al. 1994 (1993) Chen, 1986 (1982) Chen, 1986 (1983) Loo and Xu, 1986 (1983) Loo and Xu, 1986 (1983) Yang and Fu, 1989 Zhu and Pan, 1990 Pandeya et al., 1991 Hu and Zeng, 1984 (1974) Chaplin et al., 1980 Nielsen, 1989 Rufty er al., 1990 Croughan, 1985 Croughan et al., 1994 Grant and McDougall, 1995 Smith and Quesenbeny, 1995

VARIATION AND CROP IMPROVEMENT

22 1

first year of field trials, and this variation was enhanced by repressive physiological effects on plant vigor (Damiani et al., 1990) resulting from the in vitro process. A similar result occurred with maize (Z. mays L.) (Zehr et al., 1987) and sorghum [Sorghum bicolor (L.) Moench] (Bhaskaran et al., 1987). This heterozygous variation disappeared in the second and/or subsequent years, indicating that tissue culture-induced recessive mutations were expressed after one cycle of sexual propagation, and desirable variants could be identified because the frequency of favorable somaclonal alterations was higher than expected (Damiani et al., 1990). Because a high degree of heterozygosity exists in crosspollinating species and large natural variation is normally present, traits controlled by one or a few genes are more suitable for selection via somaclonal variation than polygenic traits. Phenotypic variation in somaclones resulting from embryogenic calli regeneration increased with increasing time in the callus and the regeneration phases for Paspalum dilatatum Poir. (Davies and Cohen, 1992) and sorghum (Cai et al., 1990). Plant vigor was reduced for most somaclones in the field plantings (Davis and Cohen, 1992). Phenotypically similar variations can segregate in progeny of different callus lines or reappear in later generations (Cai er al., 1990). Somaclonal variation among progenies of regenerants for numerous morphological and agronomic traits have been reported. Positive variations in height (Bhaskaran et al., 1987; Ma et al., 1987; Cai er al., 1990; Encheva et al., 1993), sterility (Ma et al., 1987; Cai et al., 1990; Elkonin et al., 1994; Garcia et al., 1994), maturity (Bhaskaran et al., 1987; Chen et al., 1987; Encheva et al., 1993), biomass (Bhaskaran et al., 1987), grain yield (Sun et al., 1983; Bhaskaran et al., 1987; Mohmand and Nabors, 1990), tillering (Bhaskaran et al., 1987), isozyme instability (Dahleen and Eizenga, 1990; Humphreys and Dalton, 1991, 1992; Garcia et al., 1994), ploidy level (Eapen and George, 1990), and kernel oil content (Encheva et al., 1993) have been documented from various in vitro programs. Significant somaclonal variation can be generated for various morphological and agronomic traits (Mohmand and Nabors, 1990). The differential expression of morphogenic response to tissue culture among genotypes provides the potential for exploiting the genetically heritable variation to develop novel breeding material, particularly for single-gene variants (Brettell et al., 1986; Cai et al., 1990; Damiani et al., 1990; Encheva et al., 1993) or specific traits controlled by one or a few genes (Ahloowalia, 1986).

D. FIELDSCREENING TECHNIQUES Somaclonal variation and its utility as a random, undirected form of mutation breeding has been criticized because few useful variants have been released (Baille et al., 1992). A major failure for the release of more useful regenerants

222

R. R. DUNCAN

may be due to inappropriate germ plasm/segregating progeny selection techniques or improper field screening/evaluation methods. Many of the field tests that have evaluated somaclonal variants have been conducted with small, unreplicated trials because of low seed availability (Jacquemin and Dubois, 1985; Maddock, 1985; Mitchell et a/., 1992). Inadequate or negative phenotypic/ genetic variation between regenerants and their donor parents has discouraged widescale adoption into many plant breeding selection programs (Baille et a/., 1992; Davies and Cohen, 1992). Instability of variant expression over generations and across environments has contributed to nonacceptance (Knauf, 199l). The dilemma is that agronomic performance of somaclonal variants must be confirmed in relevant field tests before they will have an impact in production programs (Jefferson, 1990; McHughen and Holm, 1991; Arnoldo et al., 1992). The number of tissue culture-regenerated genotypes for field evaluations is commonly 5100 R,’s (Damiani et al., 1990; Mohmand and Nabors, 1990; Cassells et al., 1991; Arnoldo et a/., 1992; Baille et al., 1992; Davies and Cohen, 1992; Mitchell et al., 1992; Griga et a / . , 1995); 100-250 regenerated lines have been used in a few studies (Waskom et al., 1990; Miller et al., 1992; Rao et al., 1992; Elkonin et al., 1994; Adkins et al., 1995; Duncan et at., 1995). Other studies have used in excess of 1000 R, families (Cai et al., 1990; Cai and Butler, 1990). Within a species, restrictions in culturability among genotypes may result in variable numbers oi‘ somac:ones available for R2+ generation screening in the field. Ranges of 50 to 120 R, lines (Xie et al., 1995) or 1 1 to 948 R,’s (Cai and Butler, 1990) are common. As miny as 12 replications, depending on the trait and the environment, may be needed in greenhouse (Bhaskaran et al., 1987) or field trials (Miller et al., 1992) for verification and selection of somaclonal variants. The population for selecting somaclonal variants should be as high as possible (>20,000 regenerants per genotype culture) to avoid losing desirable variants (Cai et a/., 1990; Waskom et al., 1990; Miller et a/., 1992; Smith et a/., 1993; Duncan et al., 1995).

VI. CONCLUSIONS The selection of regenerants with good agronomic performance depends on the relative effects of induced and introduced variation during the in vitro process (Caligari et al., 1993). This variation can arise from any number of processes, but the key to successful selection of desirable regenerants with improved traits is the expression of a wide diversity for the traits, trait stability over generations, adequate population density to have a reasonable chance of visually selecting the desirable variants in the field, and the proper field environment that will foster trait expression. These criteria are identical to those required for all conventional and mutation breeding programs.

VARIATION AND CROP IMPROVEMENT

223

Because of the uncontrolled variation induced by the tissue culture process, high plant populations (>20,000 plants per genotype culture) will be needed in field trials for selection of useful regenerants. In v i m selective agents can be used to target specific traits, but the uncontrolled variation induced by the process will require even higher regenerant populations for proper selection. Often, the in virro stressing agents may cause a negative selection pressure by reducing regeneration frequency or by producing weaker or less vigorous seedlings when regenerated. If biotic or abiotic stresses are used in the field selection program, population numbers should be higher in these trials than if no stress is imposed because selection will be based on good agronomic traits plus biotic or abiotic stress tolerance. Until better control of somaclonal variation can be achieved in vitro. breeding programs utilizing this selection tool must adjust their strategy and protocol to maximize identification and subsequent selection of the agronomically useful regenerants. When genes that control quantitative traits (such as drought stress tolerance, grain yield, or tillering ability) are better understood and are located through gene mapping strategies, the efficiency of breeding programs should improve. Strategies for inserting specific genes for specific traits should enhance this efficiency. However, unless breeders learn how to handle somaclonal variation in their selection programs, genetic engineering techniques to transfer single or multiple gene traits into genotypes without altering the desirable agronomic characteristics may parallel the same disappointing results that nontransformational, somaclonal variation has produced; namely, few useful cultivars being utilized in breeding and farmer production programs. If a specific qualitative trait can be selected in v i m , the occurrence of other unrelated variation should not render such a trait unusable in a breeding program. Plant breeders have transferred single genes from exotic sources using backcrossing procedures for many years. Several generations of backcrossing for a valuable trait that could not be easily selected other than in vitro may be an efficient and effective alternative. Perhaps in vitro culture in combination with induced mutations could escalate the breeding programs (Maluszynski et al., 1995). The application of tissue culture regeneration in selection strategies must be advantageous to the breeder before it will be adopted (Kresovich et al., 1987). Only a synergistic and concerted effort by traditional breeders and plant biotechnologists can bring this technology to the potential level of efficiency and application that will result in enhanced breeding efforts and cultivars with improved value-added (Rowland et al., 1995) traits that can be grown profitably by producers.

ACKNOWLEDGMENTS Appreciation is expressed to Dr. Roberta H. Smith (Texas A & M University) and Dr. Steve Kresovich (USDA-ARS-PGCR Lab-Georgia) for their helpful comments and reviews of the manuscript.

224

R. R. DUNCAN

REFERENCES Abo El-Nil, M. M., and Hildebrandt, A. C. (1971). Differentiation of virus-symptomless geranium plants from anther culture. Plant Dis. Rptr. 55, 1017-1020. Adkins, S. W., Kunanuvatchaidach, R., and Godwin, I. D. (1995). Somaclonal variation in ricedrought tolerance and other agronomic characters. A m . J. Eot. 43, 201-209. Ahloowalia, B. S . (1982). Plant regeneration from callus cultures of wheat. Crop Sci. 22,405-410. Ahloowalia, B. S. (1983). Spectrum of variation in somaclones of triploid ryegrass. Crop Sci. 23, 1141-1147. Ahloowalia, B. S. (1986). Limitations to the use of somaclonal variation in crop improvement. In “Somaclonal and Crop Improvement” (1. Semal, ed.), pp. 14-27. Martinus Nijhoff Pub., Boston. Ahloowalia, B. S . , and Maretzki, A. (1983). Plant regeneration via somatic embryogenesis in sugarcane. Plant Cell Rep. 2, 21-25. Amdsino, R. M., Manorama, C. J., Klaas, M., and Crowell, D. N. (1990). Role of DNA methylation in the gene expression in plants. In “Nucleic Acid Methylation” (G. A. Clawson, ed.), pp. 187-198. A. R. Liss, New York. Ammirato, P. V. (1983). Embryogenesis. In “Handbook of Plant Cell Culture. Vol. 1” (D.A. Evans et al.. eds.), pp. 82-123. MacMillian, New York. Anderson, P. C., and Georgeson, M. (1989). Herbicide-tolerant mutants of corn. Genome 31,994999. Anderson, S., Lewis-Smith, A. C., and Smith, S. M. (1990). Methylation of ribosomal RNA genes in Petunia hybrida plants, callus cultures, and regenerated shoots. Plant Cell Rep. 8, 554-557. Armstrong, C. L., and Phillips, R. L. (1988). Genetic and cytogenetic variation in plants regenerated from organogenic and friable, embryogenic tissue cultures in maize. Crop Sci. 28, 363-369. Armstrong, K. C., Nakamura, C.. and Keller, W. A. (1983). Karyotype instability in tissue cultureof regenerates of Triticale (X Triticosecale Wittmark) cv. Welsh from 6-month old callus cultures. 2. Pjlanzenzucht. 91, 233-245. Arnholdt-Schmitt, B. (1995). Physiological aspects of genome variability in tissue culture. 11. Growth-phase dependent quantitative variability of repetitive Bst NI fragments of primary culture of Daucus carota L. Theor. Appl. Genet. 91, 816-823. Arnholdt-Schmitt, B., Herterich, S., and Neumann, K.-H. (1995). Physiological aspects of genome variability in tissue culture. 1. Growth-phase dependent differential DNA methylation of the carnot genome (Daucus carota L.) during primary culture. Theor. Appl. Genet. 91, 809-815. Amoldo, M., Baszcynski, C. L., Bellemare, G . , Brown, G., Carlson, J., Gillespie, B., Huang, B., MacLean, N., MacRae, W. D., Rayner, G., Rozakis, S., Wescott, M., and Kemble, R. J. (1992). Evaluation of ‘transgenic canola under field conditions. Genome 35, 58-63. Asano, Y., lto, Y., Ohara, M., Sugiura, K., and Fujiie, A. (1994). Improved regeneration response of creeping bentgrass and japonica rice by maltose and lactose. Plant Cell Tissue Organ Cufture 39, 101-103. Ashmore, S . E., and Could, A. R. (1981). Karyotype evolution in a amour-derived plant tissue culture analysed by Giemsa-C banding. Proroplasma 106, 297-308. Bagley, P. C., and Taylor, N. L. (1987). Evaluation of phosphorus efficiency in somaclones of red clover. Iowu Sf.J. Res. 61, 459-480. Baille, A. M. R., Rossnagel, B. G., and Kartha, K. K. (1992). Field evaluation of barley (Hordeum vulgare L.) genotypes derived from tissue culture. Can. J. Planf Sci. 72, 725-733. Baker, C. M., and Wetzstein, H. Y. (1992). Somatic embryogenesis and plant regeneration from leaflets of peanut, Arachis hypogaea. Planf Cell Rep. 11, 71-75. Balzan, R. (1978). Karyotype instability in tissue cultures derived from the mesocotyl of Zea mays seedlings. Caryologia 31, 75-87.

VARIATION AND CROP IMPROVEMENT

225

Banks, P. M., and Larkin. P. J. (1995). Registration of three BYDV-resistant wheat germplasms: TC5, TC6, and TC9. Crop Sci. 35, 599-600. Barker, D., Schafer, M., and White, R. (1984). Restriction sites containing CpG show a high frequency of polymorphism in human DNA. Cell 36, 131-138. Baroncelli, S., Buiatti, M., Bennici, A.. Foroughi-Wehr, G., Mix, B., Gaul, H.,Tagliasacchi, A. M., Loiero, M., and Giorgi, B. (1978). Genetic control of in virro and in vitro growth in hexaploid wheat. 1. Behaviour of ditelocentric lines. 2. PJlanzenzucht. 80, 109-1 16. Bayliss,. M. W. (1977). The effects of 2,4-D on growth and mitosis in suspension cultures of Daucus carota. Plant Sci. Lett. 8, 99-103. Bayliss,. M. W. (1980). Chromosomal variation in plant tissues in culture. Inr. Rev. Cytol. llA, 113- 144. Behnke, M . (1979). Selection of potato callus for resistance to culture filtrates of Phytophthora infestans and regeneration of resistant plants. Theor. Appl. Genet. 55, 69-71, Benzion, G., and Phillips, R. L. (1988). Cytogenetic stability of maize tissue cultures: A cell line pedigree analysis. Genome 30, 318-325. Bhaskaran, S., and Smith, R. H. (1990). Regeneration in cereal tissue culture: A review. Crop Sci. 30, 1328-1336. Bhaskaran, S., Smith, R. H., and Newton, R. J. (1985). Physiological changes in cultured sorghum cells in response to induced water stress. I. Free proline. Plant Physiol. 79, 266-269. Bhaskaran, S., Smith, R. H., and Schertz, K. F. (1986). Progeny screening of sorghum plants regenerated from sodium chloride-selected callus for salt tolerance. J. Plant Physiol. 122,205-

210. Bhaskaran, S., Smith, R. H., Paliwal, S., and Schertz, K. F. (1987). Somaclonal variation from Sorghum bicolor (L.) Moench cell culture. Plant Cell Tissue Organ Culture 9, 189-196. Bianchi, M . W., and Viotti, A. (1988). DNA methylation and tissue-specific transcription of the storage protein genes in maize. Plant Mol. Biol. 11, 203-214. Bingham, E.T., and McCoy, J. T. (1986). Somaclonal variation in alfalfa. Plant Breed. Rev. 4, 123152. Borkind, C., Choi, J. H., Jin, Z. H.. Franz, G . , Hatzopoulos, P., Chorneau, R., Bonas, U.. Pelegri, F., and Sung, Z. R. (1988). Developmental regulation of embryogenic genes in plants. Proc. Natl. A L U ~Sci. . U.S.A. 85, 6399-6403. Brand, A. J.. and Bridgen, M. P. (1989). “UConn White”: A white flowered Toreniafournieri. HortScience 24, 7 14-7 15. Bressan, R. A., Hasegawa, P. M., and Handa, A. K. (1981). Resistance of cultured higher plant cells to polyethylene glycol-induced water stress. Plant Sci. Lett. 21, 23-30. Bressan, R. A., Handa, A. K., Handa, S., and Hasegawa, P. M. (1982). Growth and water relations of cultured tomato cells after adjustment to low external water potentials. Plant Physiol. 70, 1303-1309. Brettell, R. I. S., Dennis, E. S., Scowcroft. W. R., and Peacock, W. J. (1986). Molecular analysis of a somaclonal mutant of maize alcohol dehydrogenase. Mol. Gen. Genet. 202, 235-239. Bright. S., Jarrett, V., Nelson, R., Crek.cn, G., Karp, A., Franklin, I., Norburg, P., Kueh, J., Rognes, S., and Miflin, B. (1983). Modification of agronomic traits using in vitro technology. In “Plant Biotechnology” (S. H. Mantell and H. Smith, eds.), pp. 251-265. Cambridge Univ. Press, New York. Browers, M. A,, and Orton, T. J. (1982). Transmission of gross chromosomal variability from suspension cultures into regenerated celery plants. J . Heredity 73, 159- 162. Brown, D. C. W., and Atanassov, A. (1985). Role of genetic background in somatic embryogenesis in Medicago. P lant Cell Tissue Organ Culture 4, 1 1 I - 122. Brown, P. T. H. (1989). DNA methylation in plants and its role in tissue culture. Genome 31, 717729. Brown, P. T. H., and L o n , H. (1986). Molecular changes and possible origins of somaclonal

R. R. DUNCAN variation. In “Somaclonal Variation and Crop Improvement” (J. Semal, ed.), pp. 148-156. Nijhoff, Dordrecht. Brown, P. T. H., Gobel, E., and Lorz, H. (1991). RFLP analysis of Zea mays callus cultures and their regenerated plants. Theor. Appl. Genet. 81, 227-232. Burr, B., and Burr, F. (1981). Transposable elements and genetic instabilities in crop plants. Stadler Symp. 13, 115-128. Burr, B., and Burr, F. (1988). Activation of silent transposable elements. In “Plant Transposable Elements” (0.Nelson, ed.), pp. 317-323. Plenum, New York. Cai, Q., Shen, Y.,Gao, M.,and Liang, Z. (1994). Preliminary studies on in-vitro induced mutation technique and its application in early maturity CMS line improvement in hybrid rice. I n “Genetic Engineering & Crop Improvement” (J. Xiong and s. Lin, eds.), pp. 1-12. April 11-15, 1994. Wuhan, China. Int. Foundation for Science, Stockholm, Sweden and Huazhong Agric. Univ., Wuhan, China. Cai, T., and Butler, L. (1990). Plant regeneration from embryogenic callus initiated from immature inflorescences of several high-tannin sorghums. Plant Cell Tissue Organ Culture 20, 101110.

Cai, T., Ejeta, G., Axtell, J. D., and Butler, L. G. (1990). Somaclonal variation in high tannin sorghums. Theor. Appl. Genet. 79, 737-747. Caligari, P. D. S., Yapabandara, Y. M. H. B., Paul, E. M., Perret, J., Roger, P., and Dunwell, J. M. (1993). Field performance of derived generations of transgenic tobacco. Theor. Appl. Genet. 86, 875-879.

Cao, J., Rush, M. C., Nabors, M. W., Xie, Q. J., Croughan, T. P., and Nowick, E. (1991). Development and inheritance of somaclonal variation in rice. In “Biological Nitrogen Fixation Associated with Rice Production” (S. K. Dutta and C. Sloger, eds.), pp. 385-420. Oxford/PIBH Pub., New Delhi, India. Cassells, A. C., Minas, G., and Long, R. D. (1980). Comparative tissue culture studies of Pelargonium hybrids using meristem and explant culture: Chimera1 and beneficially infected varieties. In ‘‘Tissue Culture for Plant Pathologists” (J. P. Helgeson and D. S. Ingram, eds.), pp. 125-130. Blackwell, Oxford, UK. Cassells, A. C., Coleman, M. C., Farrell, G., Long, R. D., and Gotz, E. M. (1986). Screening for virus resistance in tissue culture adventitious regenerants and their progeny. In “Genetic Manipulation in Plant Breeding” (W. Horn, ed.), pp. 353-455. DeGryter, Berlin. Cassells, A. C., Deadman, M. L., Brown, C. A., and Griffin, E. (1991). Field resistance to late blight (Phytophthora tuberosum L.) somaclones associated with instability and pleiotropic effects. Euphyfica 56, 75-80. Catcheside, D. G. (1939). A position effect in Oenothera. J. Genet. 38, 345-352. Cedar, H., and Razin, A. (1990). DNA methylation & development. Biochim. Biophys. Acta 1049, 1-8.

Chaleff, R. S. (1983). Isolation of agronomically useful mutants from plant cell cultures. Science 219, 676-682. Chaleff, R. S . , and Ray, T. B. (1984). Herbicide-resistant mutants from tobacco cell cultures. Science 223, 1148-1151. Chandlee, J. M. (1990). The utility of transposable elements as tools for the isolation of plant genes. Physiol. Plant. 79, 105-115. Chaplin, J. F., Burk, L. G., Gooding, G. V., and Powell, N. T. (1980). Registration of NC 744 tobacco germplasm. Crop Sci. 20, 677. Charmet, G., and Bernard, S. (1984). Diallel analysis of androgenetic plant production in hexaploid Triticale (X Triticosecale, Wittmack). Theor. Appl. Genet. 69, 55-61, Chen, T. H. H., and Gusta, L. V. (1983). Abscisic acid-induced freezing resistance in cultured plant cells. Plant Physiol. 73, 71-75.

VARIATION AND CROP IMPROVEMENT

227

Chen, T. H. H., Lazar, M. D., Scoles, G. J., Gusta, L. V., and Kartha, K. K. (1987). Somaclonal variation in a population of wheat. J . Plant Physiol. 130, 27-36. Chen, Y . (1986). The inheritance of rice pollen plant and its application in crop improvement. In “Haploids of Higher Plants in Vitro” (H. Hu and H. Yang, eds.), pp. 118-136. Springer-Verlag, Berlin. Chewier, N., Qureshi, J. A., Hucl, P., and Kartha, K. K. (1990). Heritability of in vitro regeneration in wheat (Triticum aestivum L.). Can. J. Plant Sci. 70, 547-550. Churchill, G. C., Ewan, B., Reaney, M. J. T., Abrams, S. R., and Gusta, L. V. (1992). Structureactivity relationships of abscisic acid analogs based on the induction of freezing tolerance in bromegrass (Bromus inermis Leyss) cell cultures. Plant Phvsiol. 100, 2024-2029. Close, K. R . , and Gallagher-Ludeman. A. ( 1989). Structure-activity relationships of auxin-like plant growth regulators and genetic influences on the culture induction response in maize (Zea mays L.).Plant Sci. 61, 245-252. Colijn-Hooymans, C. M., Hakkert, J. C., Jansen, J., and Custers, J. B. M. (1994). Competence for regeneration of cucumber cotyledons is restricted to specific developmental stages. Plan6 Cell Tissue Organ Culture 39, 21 1-217. Conner, A. J., and Meredith, C. P. (1985). Large scale selection of aluminum-resistant mutants from plant cell culture: Expression and inheritance in seedlings. Theor. Appl. Genet. 71, 159165. Conner, A. I., and Meredith, C. P. (1989). Genetic manipulation of plant cells. In “The Biochemistry of Plants. Vol. 15. Molecular Biology” (A. Marcus, ed.), pp. 653-688. Academic Press, New York . Corriols, L., Dore, C., and Rameau, C. (1990). Commercial release in France of Andreas, the first asparagus all-male F, hybrid. Acfa Hort. 271, 249-252. Coulondre, C.. Miller, J. H., Farabaugh, P. J.. and Gilbert, W. (1978). Molecular basis of base substitution hotspots in Escherichia coli. Nature (London) 274, 775-780. Crea, F., Bellucci, M., Damiani, F., and Arcioni, S. (1995). Genetic control of somatic embryogenesis in alfalfa (Medicago sativa L. cv. Adriana). Euphytica 81, 151-155. Creissen, G. P., and Karp, A. (1985). Karyotypic changes in potato plants regenerated from protoplasts. Plant Cell Tissue Organ Culture 4, I7 I - 182. Croughan, S. S. (1985). Bermudagrass improvement through tissue culture. Annu. Rep. Rice Res. Sta. 77, 384-385. [Louisiana Agric. Exp. Sta., Louisiana State Univ., Agric. Ctr.] Croughan, S. S.. and Quisenbeny, S. S. (1989). Enhancement of fall armyworm resistance in bermudagrass through cell culture. J . Econ. Entomol. 82, 236-238. Croughan, S . S., Quisenbeny, S. S., Eichhorn, M. M., Jr., Colyer, P. D., and Brown, T. F. (1994). Registration of Brazos-R3 bermudagrass germplasm. Crop Sci. 34,542. Cullis, C. A. ( 1976). Environmentally-induced changes in ribosomal cistron number in flax. Heredity 36, 73-79. Cullis, C. A., and Cleary, W. (1986). DNA variation in flax tissue culture. Can. J. Genet. Cytol. 28, 247-25 I . Dahleen, L. S., and Eizenga. G. C. (1990). Meiotic and isozymic characterization of plants regenerated from euploid and selfed monosomic tall fescue embryos. Theor. Appl. Gener. 79, 39-44. D’Amato, F. (1952). Polyploidy in the differentiation and function of tissues and cells in plants. Caryologia 4, 31 1-358. D’Amato, F. (1985). Cytogenetics of plant cell tissue and cultures and their regenerants. Crit. Rev. Plant Sci. 3, 73- 112. D’Amato, F. (1991). Nuclear changes in cultured plant cells. Caryologia 44,217-224. Damiani, F., Pezzotti, M., and Arcioni, S. (1990). Somaclonal variation in Lotus corniculatus L. in relation to plant breeding purposes. Euphytica 46, 35-41. Daofen, H. (1986). Jinghua No. I . , a winter wheat variety derived from pollen sporophyte. In

228

R. R. DUNCAN

“Haploids of Higher Plants in Vitro” (H. Hu and H. Yang, eds.), pp. 137-138. Springer-Verlag, Berlin. Davies, L. J., and Cohen, D. (1992). Phenotypic variation in somaclones of Paspalum dilatatum and their seedling offspring. Can. J. Plant Sci. 72, 773-784. de Boucaud, M.-T., and Gaultier, J.-M. (1983). Evidence of a different nuclear population in chimerae of regenerated plants from tobacco aneuploid tissue cultures. Cytophotometric and statistical analysis. Z. Pflanzenphysiol. 110, 97- 105. de Buyser, J., Henry, Y., Lonnet, P., Hertzog, R., and Hespel, A. (1987). ‘Florin’: A doubled haploid wheat variety developed by the anther culture method. Planr Breed. 98, 56-57. de Buyser, J., Marcotte, J-L., and Henry, Y. (1992). Genetic analysis of in vitro wheat somatic embryogenesis. Euphytica 63, 265-270. Dennis, E. S., Brettell, R. I. S., and Peacock, W. J. (1987). A tissue culture induced Adhl null mutant of maize results from a single base change. Mol. Gen. Genet. 210, 181-183. Dermen, H. (1960). Nature of plant sports. Am. Hori. Mag. 39, 123-173. Dimitrov, B. (1987). Relationship between sister-chromatid exchanges and heterochromatin or DNA replication in chromosomes of Crepis capillaris. Muta?. Res. 190, 271-276. Dolezel, J., and Novak, F. J. (1986). Sister chromatid exchanges in garlic (Alliurn sativum L.) callus cells. Plant Cell Rep. 5 , 280-283. Dolezel, J . , Lucretti, S., and Novak, F. J. (1987). The influence of 2,4-dichlorophenoxyaceticacid on cell cycle kinetics and sister-chromatid exchange frequency in garlic (Allium sativa L.) meristem cells. Biol. Plant. 29, 253-257. Donn, G., Tischer, E., Smith, J. A., and Goodman, H. M. (1984). Herbicide-resistant alfalfa cells: An example of gene amplification in plants. J. Mol. Appl. Genet. 2, 621-635. Dofling, K., Dotfiling, H., and Lesselich, G. (1993). In-vitro selection and regeneration of hydroxyproline-resistant lines of winter wheat with increased proline content and increased frost tolerance. J. Plant Physiol. 142, 222-225. Douglas, G. C., Keller, W. A,, and Setterfield, G. (1981). Somatic hybridization betweenNicotiana rustica and N . tabacum. 11. Protoplast fusion and selection and regeneration of hybrid plants. Can. J. Boi. 59, 220-227. Duncan, D. R., and Widholm, J. M. (1987). Proline accumulation and its implication in cold tolerance of regenerable maize callus. Plant Physiol. 83, 703-708. Duncan, R. R., Isenhour, D. J., Waskom, R. M., Miller, D. R., Nabors, M . W., Hanning, G. E., Petersen, K. M., and Wiseman, B. R. (1991a). Registration of GATCCPIOO and GATCCPlOl fall armyworm resistant hegari regenerants. Crop Sci. 31, 242-244. Duncan, R. R., Waskom, R. M., Miller, D. R., Voigt, R. L., Hanning, G. E., Timm, D., and Nabors, M. W. (1991b). Registration of GACIO2: Acid soil tolerant hegari regenerant. Crop Sci. 31, 1396-1397. Duncan, R. R., Waskom, R. M., Miller, D. R., Hanning, G. E., Timm, D. A., and Nabors, M. W. (1992). Registration of GC103 and GC104 acid-soil tolerant Tx430 regenerants. Crop Sci. 32, 1076- 1077. Duncan, R. R., Waskom, R. M . , and Nabors, M. W. (1995). In virro screening and field evaluation of tissue-culture-regenerated sorghum [Sorghum bicolor (L.) Moench] for soil stress tolerance. Euphytica 85, 373-380. Dyer, A. F. (1976). Modification and errors of mitotic cell divisions in relation to differentiation. In “Cell Division in Higher Plants’’ (M. M . Yeoman, ed.), pp. 199-253. Academic Press, London. Eapen, S., and George, L. (1990). Somatic embryogenesis and plant regeneration in inflorescence segments of Sorghum versicolor. Maydica 35, 55-58. Elkonin, L. A., Gudova, T. N., and Ishin, A. G. (1994). Inheritance of male sterility mutations induced in haploid sorghum tissue culture. Euphytica 80, 111-1 18. Encheva, J., Ivanov, P., Tsvetkova, F., and Nikolova, V. (1993). Development of a new initial

VARIATION AND CROP IMPROVEMENT

229

breeding material in sunflower (Helianthus annuus L.) using direct organogenesis and somatic embryogenesis. Euphvtica 68, 181-185. Engler, D., and Grogan, R. (1984). Variation in lettuce plants regenerated from protoplasts. J. Heredity 75, 426-430. Evans, D. A. (1988). Applications of somaclonal variation. In “Biotechnology in Agriculture” (A. Mizrahi, ed.), pp. 203-233. A. R. Liss, New York. Evans, D. A. (1989).Somaclonal variation: Genetic basis and breeding applications. Generics 5,4650. Evans, D. A,, and Sharp, W. R. (1983). Single gene mutations in tomato plants regenerated from tissue culture. Science 221, 949-951. Evans, D. A,, and Sharp, W. R. (1986). Applications of somaclonal variations. BiolTechnology 4, 528-532. Evans, D. A,, Sharp, W. R., and Medina-Filho, H. P. (1984). Somaclonal and gametoclonal variation. Am. J . Bor. 71, 759-774. Fedoroff, N. V., and Baker, B. (1989). Plant gene vectors and genetic transformation: The structure, function, and uses of maize transposable elements. In “Cell Culture and Somatic Cell Genetics of Plants. Vol. 6. Molecular Biology of Plant Nuclear Genes” ( J . Schell and 1. K. Vasil, eds.), pp. 101-132. Academic Press, San Diego. Feher, F., Hangyel Tarczy, M., Bocsa, L., and Dudits, D. (1989). Somaclonal chromosome variation in tetraploid alfalfa. Plant Sci. 60, 91-99. Feirer, R. P., Mignon, G., and Litvay, J. D. (1984).Arginine decarboxylase and polyamines required for embryogenesis in the wild carrot. Science 223, 1433-1435. Fienberg, A. A,, Choi, J. H., Lubich, W. P., and Sung, Z. R. (1984). Developmental regulation of polyamine metabolism in growth and differentiation of carrot culture. Planta 162, 532539. Flehinghaus, T., Deimling, S., and Geiger, H. H. (1991). Methodical improvements in rye anther culture. Plant Cell Rep. 10, 397-400. Foy, C. D., Duncan, R. R., Waskom, R. M., and Miller, D. R. (1993). Tolerance of sorghum genotypes to an acid, aluminum toxic Tatum subsoil. J. Planr Nutr. 16, 97-127. Fukui, K. (1983). Sequential Occurrence of mutations in a growing rice callus. Theor. Appl. Genet. 65, 225-230. Galiba, G., and Sutka, J. (1989). Frost resistance of somaclones derived from Triricum aestivum L. winter wheat calli. Plant Breed. 102, 101-104. Galiba, G., Kovacs, G., and Sutka, J. (1986). Substitution analysis of plant regeneration from callus culture in wheat. Plunr Breed. 97, 261-263. Garcia, A,. Dalton, S. J.. and Humphreys, M. 0. (1994). Reproductive disturbances and phosphogluco isomerase instability in Festuca arundinacea (tall fescue) plants regenerated from callus and cell suspension cultures. Heredity 73, 355-562. Gavazzi, G . , Tonelli, C., Todesco, G., Arreghini, E., Raffaldi, F., Vecchio, F., Barbuzzi, G., Biasini. M., and Sala, F. (1987). Somaclonal variation versus chemically-induced mutagenesis in tomato (Lvcopersicon esculentum L.). Theor. Appl. Gener. 74, 733-738. Gengenbach. B. G., and Umbeck, P. (1982). Characteristics of T-cytoplasm revertants from tissue culture. Maize Genet. Coop. Newslet/. 56, 140- 142. Gmitter, F. G., Jr., Ling, X., Cai, C., and Grosser, J. W. (1991). Colchicine-induced polyploidy in Citrus embryogenic cultures, somatic embryos, and regenerated plantlets. Planr Sci. 74, 135141. Goldberg, R. B. (1986). Regulation of plant gene expression. Philos. Trans. R. SOC.LondonB. 314, 343-353. Could, A. R . (1982). Chromosome instability in plant tissue cultures studies with the banding technique. In “Plant Tissue Culture” (A. Fujiwara, ed.), pp. 431. Maruzen, Tokyo.

230

R. R. DUNCAN

Gould, A. R. (1984). Control of the cell cycle in cultured plant cells. CRC Crir. Rev. Plant Sci. I, 315-344. Grafstrom, R. H., Hamilton, D. L., and Yuan, R. (1984). DNA methylation: DNA replication and repair. In “DNA Methylation: Biochemistry and Biological Significance” (A. Razin, H. Cedar, and A. D. Riggs, eds.), pp. I 1 1-126. Springer-Verlag. New York. Graham, M. D., Heavner, D. L., Nickell, C. D., and Widholm, J. D. (1993). Response of soybean genotypes to boron, zinc, and manganese deficiency in tissue culture. Plant Soil 150,307-310. Graham, M. J., Stephens, P. A. Widholm, J. M., and Nickell, C. D. (1992). Soybean genotype evaluation for iron deficiency chlorosis using sodium bicarbonate and tissue culture. J. Plant Nufr. 15, 1215-1225. Grant, W. F., and McDougall, R. B. (1995). Registration of H 401-4-4-2 birdsfoot trefoil germplasm resistant to sulfonylurea. Crop Sci. 35, 286-287. Griesbach, R. J. (1989). Selection of a dwarf Hemerocallis through tissue culture. HorrScience 24, 1027-1028. Griesbach, R. J., and Semeniuk, P. (1987). Use of somaclonal variation in the improvement of Eusfoma grandgorum. J. Heredify 78, 1 14- 116. Griesbach, R. J., Semeniuk, P., Roh, M., and Lawson, R. H. (1988). Tissue culture in the improvement of Eustoma. HortScience 23, 658-791. Griga, M., Stejskal, J., and Beber, K. (1995). Analysis of tissue-culture derived variation in pea (Pisum sativum L.)-Preliminary results. Euphytica 85, 335-339. Groose, R. W., and Bingham, E. T. (1986a). An unstable anthocyanin mutation recovered from tissue culture of alfalfa (Medicago sativa). 1. High frequency of reversion upon reculture. Plant Cell Rep. 5 , 104-107. Groose, R. W., and Bingham, E. T. (1986b). An unstable anthocyanin mutation recovered from tissue culture of alfalfa (Medicago sativa). 2. Stable non-revertant derived from reculture. Plant Cell Rep. 5 , 108-1 10. Gui, Y., Hong, S., Ke, S., and Skirvin, R. M. (1993). Fruit and vegetative characteristics of endosperm-derived kiwifruit (Acfinidia chinensis F.) plants. Euphyrica 71, 57-62. Guitton, M. C., Keller, B. T., Part, D., De Gunsberg, J. Borchardt, R. T., and Veron, M. (1988). S-adenosylmethionine, S-adenosylhomocysteine, and S-adenosylhomocysteine hydrolase variations during differentiations of Dicfyosrelium discoidem. Cell. Diger. 22, 203-210. Haberlach, G. T., Budde, A. D., Sequeira, L., and Helgeson, J. P. (1978). Modifications of disease resistance to tobacco callus tissues by cytokinins. Plant Physiol. 62, 522-525. Haccius, B. (1978). Question of unicellular origin of non-zygotic embryos in callus cultures. Phytomorphology 28, 74-81. Hall, H. K . , Skirvin, R. M., and Braam, W. F. (1986). Germplasm release of ‘Lincoln Logan’, a tissue-culture-derived genetic thornless ‘Loganberry’. Fruit Var. J . 40, 134-135. Handa, A. K . , Bressan, R. A., Handa, S., and Hasegawa, P. M. (1982). Characteristics of cultured tomato cells after prolonged exposure to medium containing polyethylene glycol. Plant Physiol. 69, 514-521. Handro, W. (1981). Mutagenesis and in virro selection. In “Plant Tissue Culture: Methods and Applications in Agriculture” (T. A. Thorpe, ed.), pp. 155-180. Academic Press, New York. Harms, C. T., and DiMaio, J. I. (1991). Primisulfuron herbicide-resistant tobacco cell lines. Application of fluctuation test design to in vitro mutant selection with plant cells. J. Plant Physiol. 137, 513-5 19. Hartman, H. T., and Kester, D. E. (1983). “Plant Propagation.” Prentice-Hall International, Englewood Cliffs, NJ. Hasegawa, P. M., Bressan, R. A., and Handa, A. K. (1980). Growth characteristics of NaCI-selected and non-selected cells of Nicotianu tabacum L. Plant Cell Physiol. 21, 1347-1355.

VARIATION AND CROP IMPROVEMENT

231

Hasegawa, P. M., Bressan, R. A., Handa, S., and Handa, A. K. (1984). Cellular mechanisms of tolerance to water stress. HortScience 19, 7-18. Heath-Pagliuso, S.. and Rappaport, L. (1990). Somaclonal variant UC-T3: The expression of Fusarium wilt resistance in progeny arrays of celery, Apium graveolens L. Theor. Appl. Genet. 80, 390- 394. Heath-Pagliuso, S., Pullman, I., and Rappaport, L. (1988). Somaclonal variation in celery: Screening for resistance to Fusurium oxysporum f. sp. apii. Theor. Appl. Genet. 75, 446-45 1. Heath-Pagliuso, S., Pullman, J., and Rappaport, L. (1989). ‘UC-T3 Somaclone’: Celery germplasm resistant to Fusarium oxysporum f.sp.apii, race 2. HortScience 24, 71 1-712. Heinz, D. J., and Mee, G. W. P. (1969). Plant differentiation from callus tissue of Saccharurn species. Crop. Sci. 9, 346-348. Heinz, D. J . , and Mee, G. W. P. (1971). Morphologic, cytogenetic, and enzymatic variation in Saccharum species hybrid clones derived from callus culture. Am. J . Bot. 58, 257-262. Helgeson, J. P., Kemp, J. D., Haberlach, G . T., and Maxwell, D. P. (1972). A tissue culture system for studying disease resistance: the black shank disease in tobacco callus cultures. Phytopathology 62, 1439-1443. Henry, Y., Marcotte, J.-L., and DeBuyser, J. (l994a). Chromosomal location of genes controlling short-term and long-term somatic embryogenesis in wheat revealed by immature embryo culture of aneuploid lines. Theor. Appl. Genet. 89, 344-350. Henry, Y.. Vain, P.. and De Buyser, 1. (1994b). Genetic analysis of in vitro plant tissue culture responses and regeneration capacities. Euphvtica 79, 45-58. Hepburn, A. G., Clark, L. E., Pearson, L., and White. J. (1983). The role of cytosine methylation in the control of nopaline synthase gene expression in a plant tumor. J. Mol. Appl. Genet. 2, 315329. Hershkovitz, M., Gruenbaum, Y., Renbaum, P., Razin, A,, and Loyer, A. (1990). Effect of CpG methylation on gene expression in transfected plant protoplasts. Gene 94, 189-193. Ho, W.-J., and Vasil, I. K. (1983). Somatic embryogenesis in sugarcane (Saccharurn oficinarum L.), I. The morphology and physiology of callus formation and the ontogeny of somatic embryos. Protoplasma 118, 169- 180. Holliday, R. (1987). The inheritance of epigenetic defects. Science 238, 163-170. Hu, H., and Zeng, J. Z. (1984). Development of new varieties via anther culture. In “Handbook of Plant Cell Culture. Vol. 3. Crop Species’’ (P. V. Ammirato, D. A. Evans, W. R. Sharp, and Y. Yamada, eds.), pp. 65-90. Macmillian, New York. Humphreys, M. W., and Dalton, S. J. (1991). Stability at the phosphoglucoisomerase (PGI/2) locus in Festuca arundinacea plants regenerated from cell suspension and protoplast culture. Genome 34, 59-65. Humphreys, M. W., and Dalton, S. J. (1992). Stability at the phosphoglucoisomerase (PGU2) locus in Lolium multijorum (2n = 4X = 28) X Festuca arundinacea (2n = 6X = 42) plants regenerated from cell suspension. Genome 35, 461-467. Isenhour, D. J., and Wiseman, B. R. (1991). Fall armyworm resistance in progeny of maize plants regenerated via tissue culture. Floridu Entomol. 74, 221-228. Isenhour, D. J., Duncan, R. R., Miller, D. R., Waskom, R. M., Hanning, G . E., Wiseman, B. R., and Nabors, M. W. (1991). Resistance to leaf-feeding by the fall armyworm (Lepidoptera:Noctuidae) in tissue culture derived sorghums. J. Econ. Entomol. 84, 680-684. Jackson, J. A., and Lyndon, R. F. (1990). Habituation: Cultural curiosity or developmental determinant’?Physiol. Plant. 79, 579-583. Jacky, P. B., Beek, B., and Sutherland, G . R. (1983). Fragile sites in chromosomes: possible model for the study of spontaneous chromosome breakage. Science 220, 69-70. Jacquemin, J. M., and Dubois, J. (1985). Somaclonal variation applied to wheat breeding. In

232

R. R. DUNCAN

“Somaclonal Variation and Crop Involvement” (J. Semal, ed.), pp. 177-181. Nijhoff, Dordrecht. Jain, R. K., Jain, S., and Chowdhury, J. B. (1991). In vitro selection for salt tolerance in Brassica juncea L. using cotyledon explants, callus and cell suspension cultures. Ann. Bor. 67,517-519. Jefferson, R. A. (1990). New approaches for agricultural molecular biology: From single cells to field analysis. In “Gene Manipulation in Plant Improvement II” (J. P. Gustafson, ed.), pp. 365-400. Plenum Press, New York. Johnson, S. S . , Phillips, R. L., and Rines, H. W. (1987). Possible roles of heterochromatin in chromosome breakage induced by tissue culture in oats (Avena sativa L.). Genome 29, 439446. Jones, P. W. (1990). Disease resistance. In “Plant Cell Line Selection” (P. I. Dix, ed.), pp. 113-149. VCH, Verlagsgesellschaft, Weinheim. Jordan, M. C., and McHughen, A. (1987). Selection for chlorsulfuron resistance in flax (Linum usitatissimum) cell cultures. J . Plant Physiol. 131, 333-338. Kaleikau, E. K., Sears, R. G., and Gill, B. S. (1989). Monosomic analysis of tissue culture response in wheat (Triticurn aestivurn L.). Theor. Appl. Genet. 78, 625-632. Karp, A. (1989). Can genetic instability be controlled in plant tissue cultures? Newsletr. Inr. Assoc. Plant Tissue Culture 58, 2- I 1. Karp, A. (1991). On the current understanding of somaclonal variation. Oxjord Surv. Plant Mol. Cell Biol. 7, 1-58. Karp, A. (1995). Somaclonal variation as a tool for crop improvement. Euphytica 85, 295-302. Karp, A., and Bright, S. W. J. (1985). On the causes and origins of somaclonal variation. Oxford Surv. Plant Mol. Cell Biol. 2, 199-234. Karp, A., and Maddock, S. E. (1984). Chromosome variation in wheat plants regenerated from cultured immature embryos. Theor. Appl. Genet. 67, 249-256. Kawai, S., and Takagi, S. (1992). Application of phytosiderophore to plant cell cultures and production of phytosiderophores by iron-deficiency stressed plant cell cultures. J. Plant Nutr. 15, 16 13- 1624. Kendall, E. J., Qureshi, J. A., Kartha, K. K., Leung, N., Chewier, N., Caswell, K., and Chen, T. H. H. (1990). Regeneration of freezing-tolerant spring wheat (Triticum aesrivurn L.) plants from cryoselected callus. Plant Physiol. 94, 1756-1762. Keyes, G. J., Collins, G. B., and Taylor, N. 0. (1980). Genetic variation in tissue cultures of red clover. Theor. Appl. Genet. 58, 265-27 I . Knauf, V. C. (1991). Agricultural progress: Engineered crop species, field trial results and commercialization issues. Curr. Oppor. Biotechnol. 2, 199-202. Komamine, A,, Matsumoto, M., Tsukuhara, M., Fujiwara, A,, Kavahara, R., Ito, M., Smith, J., Nomura, K . , and Fujimura, T. (1990). Mechanisms of somatic embryogenesis in cell culturesPhysiology, biochemistry, and molecular biology. In “Progress in Plant Cellular and Molecular Biology” (H. J. J. Nijkamp and L. H. W. Vander Plas, eds.), pp. 307-313. Kluwer, Dordrecht. Komatsuda, T., Enomoto, A,, and Nakjima, K. (1989). Genetics of callus proliferation and shoot differentiation in barley. J. Herediry 80, 345-350. Kresovich, S . , McGee, R. E., Panella, L., Reilley, A. A,, and Miller, F. R. (1987). Application of cell and tissue culture techniques of sorghum, Sorghum bicolor (L.) Moench: Progress and potential. Adv. Agron. 41, 147-170. Landsmann, J., and Uhrig, H. (1985). Somaclonal variation in Solanum tuberosum detected at the molecular level. Theor. Appl. Genet. 71, 500-505. Lapitan, N. L. V., Sears, R. G., and Gill, B. S. (1984). Translocation and other karyotypic structural changes in wheat x rye hybrids generated from tissue culture. Theor. Appl. Genet. 68, 547554.

VARIATION AND CROP IMPROVEMENT

233

Lapitan, N. L. V., Sears, R. G., and Gill, B. S. (1988). Amplification of repeated DNA sequences in wheat x rye hybrids regenerated from tissue culture. Theor. Appl. Genet. 75, 381-388. Larkin, P. J. (1985). Case histories of genetic variability in vitro: Wheat and triticale. In “Cell Culture and Somatic Cell Genetics in Plants. Vol. 3. Plant Regeneration and Genetic Variability” (I. K. Vasil, ed.), pp. 367-383. Academic Press, Orlando. FL. Larkin, P. J. (1987). Somaclonal variation: History, method, and meaning. Iowa St. J. Rex 61,393434. Larkin, P. J., and Scowcroft, W. R. (1981). Somaclonal variation-A novel source of variability from cell cultures for plant improvement. Theor. Appl. Genet. 60, 197-214. Larkin, P. J., Ryan, S. A., Brettell, R. I. S . , and Scowcroft, W. R. (1984). Heritable somaclonal variation in wheat. Theor. Appl. Genet. 67, 443-455. Larkin, P. J., Banks, P. M., Bhati, R., Brettell, R. 1. S . , Davies, P. A., Ryan, S. A,, Scowcroft, W. R., Spindle, L. H., and Tanner, G . T. (1989). From somatic variant to variant plants: Mechanisms and applications. Genome 31, 705-71 I , Lee, M., and Phillips, R. L. (1987a). Genomic rearrangements in maize induced by tissue culture. Genome 29, 122- 128. Lee, M., and Phillips, R. L. (1987b). Genetic variants in progeny of regenerated maize plants. Genome 29, 834-838. Lee, M., and Phillips, R. L. (1988). The chromosomal basis of somaclonal variation. Annu. Rev. Plant Mol. Biol. 39, 413-437. Lee, M., Geadelmann, J. L., and Phillips, R. L. (1988). Agronomic evaluation of inbred lines derived from tissue culture of maize. Theor. Appl. Genet. 75, 841-849. Lentini, Z.. Earle, E. D., and Plaisted, R. L. (1990). Insect-resistant plants with improved horticultural traits from interspecific potato hybrids grown in vitro. Theor. Appl. Genet. 80, 95104. Li, M., Ni, P., Zhang, L., and Zhang, L. (1994). Anther culture breeding of rice at the CAAS. In “Genetic Engineering and Crop Improvement” (J. Xiong and S . Lin, eds.), pp. 119-126. April 11-15. 1994. Wuhan, China. Int. Foundation for Science, Stockholm, Sweden and Huazhong Agric. Univ., Wuhan, China. Lima-de-Fana, A. ( 1969). DNA replication and gene amplification in heterochromatin. In “Handbook of Molecular Cytology” (A. Lima-de-Faria, ed.), pp. 277-325. North-Holland, Amsterdam. Liscum, E., 111, and Hangarter, R. P. (1991). Manipulation of ploidy levels in cultured haploid Petunia tissue by phytohormone treatments. J. Plant Physiol. 138, 33-38. Loo, S.-W., and Xu, Z. H. (1986). Rice: Anther culture for rice improvement in China. In “Biotechnology in Agriculture and Forestry: 2. Crops I” (Y.P. S . Bajaj, ed.), pp. 139-156. SpringerVerlag, Berlin. Lorz, H., and Scowcroft, W. R. (1983). Variability among plants and their progeny regenerated from protoplasts of Su/su heterozygotes of Nicotiana tabacum. Theor. Appl. Gener. 66, 67-75. Lon, H., Gobel, E., and Brown, P. (1988). Advances in tissue culture and progress towards genetic transformation of cereals. Plant Breed. 100, 1-25. Lo Schiavo. F., Pitto, L., Giuliano, G., Torti, G., Nuti-Ronchi, V.. Marazziti, D., Vergera, R., Orselli, S., and Terzi, M. (1989). DNA methylation of embryogenic carrot cell cultures and its variations as caused by mutation, differentiation, hormones, and hypomethylating drugs. Theor. Appl. Genet. 77, 325-33 I . Ma, H., Gu, M., and Liang, G. H. (1987). Plant regeneration from cultured immature embryos of Sorghum bicolor (L.) Moench. Theor. Appl. Genet. 73, 389-394. MacLean, N. L., and Nowak, J. (1989). Plant regeneration from hypocotyl and petiole callus of Trijolium pratense L. Plant Cell Rep. 8, 395-398.

2 34

R. R. DUNCAN

Maddock, S. E. (1985). Somaclonal variation in wheat. In “Somaclonal Variation and Crop Improvement” (J. Semal, ed.), pp. 127-135. Nijhoff, Dordrecht. Maddock, S. E., and Semple, J. T. (1986). Field assessment of somaclonal variation in wheat. J . EXP.Eot. 37, 1065-1078. Maddock, S. E., Risiott, R., Parmar, S., Jones, M., and Shewry, P. (1985). Somaclonal variation in gliadin patterns of grains of regenerated wheat plants. J . Exp. Eot. 36, 1976-1984. Maliga, P. (1984). Isolation and characterization of mutants in plant cell culture. Annu. Rev. Plant Physiol. 35, 5 19-542. Maluszynski, M., Ahloowalia, B. S., and Sigurbjornsson, B. (1995). Application of in vivo and in vitro mutation techniques for crop improvement. Euphytica 85, 303-315. Marcotrigiano, M. (1990). Genetic mosaics and chimeras: implications in biotechnology. In “Biotechnology in Agriculture and Forestry. Vol. II” (Y.P. S. Bajaj, ed.), pp. 85-1 11. SpringerVerlag, Berlin. Marcotrigiano, M., and Jagannathan, K. ( I 988). Paulownia tomentosa ‘Somaclonal Snowstorn.’ HortScience 23, 226-227. Mathis, R. J . , and Fukui, K. (1986). The effect of specific chromosome and cytoplasm substitutions on the tissue culture response of wheat (Triricum aestivum) callus. Thear. Appl. Genet. 71,797800. Mathis, R . I., Higgins, P., and Atkinson, E. (1988). Genetic control of wheat (Triricum aestivum) tissue culture response. In “Proceedings of the 7th Wheat Genetics Symposium Volume 1” (T. E. Miller and R. M. D. Koebner, eds.), pp. 763-768. Bath Press, Bath, UK. McClintock, B. (1978). Mechanisms that rapidly reorganize the genome. Stadler Symp. 10, 25-47. McClintock, B. (1984). The significance of responses of the genome to challenge. Science 226,792801. McCoy, T. J. (1988). Tissue culture selection for disease resistant plants. Iowa State J. Res. 62, 503522. McCoy, T. J., Phillips, R. L., and Rines, H. W. (1982). Cytogenetic analysis of plants regenerated from oat (Avena sativa) tissue cultures: High frequency of partial chromosome loss. Can. J. Genet. Cytol. 24, 37-50. McHughen, A., and Holm, F. (1991). Herbicide resistant transgenic flax field test: Agronomic performance in normal and sulfonylurea-containing soils. Euphytica 55, 49-56. McHughen, A., and Swartz, M. (1984). A tissue-culture-derived salt-tolerant line of flax (Linum usitatissimum). J. Plant Physiul. 117, 109-1 17. McPheeters, K., and Skiwin, R. M. (1983). Histogenic layer manipulation in chimera1 ‘Thornless Evergreen’ trailing blackberry. Euphytica 32, 35 1-360. McPheeters, K . , and Skirvin, R. M. (1989). Somaclonal variation among ex vitro ‘Thornless Evergreen’ trailing blackberries. Euphyrica 42, 155-162. Meins, F. (1982). Habituation of cultured plant cells. In “Molecular Biology of Plant Tumors’’ (G. Kahl and J. S. Schrell, eds.), pp. 3-31. Academic Press, London. Meins, F., Jr. (1989). Habituation: Heritable variation in the requirement of cultured plant cells for hormones. Annu. Rev. Genet. 23, 395-408. Mengoli, M., Bagni, N., Luccarini, G., Ronchi, V. N., and Serafini Fracassini, D. (1989). Daucus carota cell cultures: Polyamines and effect of polyamine biosynthesis inhibitors in the preembryogenic phase and different embryoid stages. J . Plant Physiol. 134, 389-394. Meulemans, M., Duchene, D., and Fouarge, G. (1987). Selection of variants by dual culture of potato and Phytuphthora infestans. In “Biotechnology in Agriculture and Forestry. 3. Potato” (Y. P. S. Bajaj, ed.), pp. 318-331. Springer-Verlag, Berlin. Meuth, M., L‘Heureux-Huard, N., and Trudel, M. (1979). Characterization of a mutator gene in Chinese hamster ovary cells. Pruc. Narl. Acad. Sci. U.S.A. 76, 6505-6509. Miller, D. R., Waskom, R. M.,Duncan, R. R., Chapman, P. L., Brick, M. A,, Hanning, G . E.,

VARIATION AND CROP IMPROVEMENT

235

Timm. D. A.. and Nabors, M. W. (1992). Acid soil stress tolerance in tissue-culture-derived sorghum lines. Crop. Sci. 32, 324-327. Mitchell, M. J., Busch, R. H., and Rines. H. W. (1992). Comparison of lines derived by anther culture and single-seed descent in a spring wheat cross. Crop Sci. 32, 1446-1451. Mohmand, A. S., and Nabors, M. W. (1990). Somaclonal variant plants of wheat derived from mature embryo explants of three genotypes. Plant Cell Rep. 8, 558-560. Moore, P. P., Robbins, J. A , , and Sjulin. J. M. (1991). Field performance of ‘Olympus’ strawberry subclones. HortScience 26, 192- 194. Monish, F., Vasil, V., and Vasil, I. K. (1987). Developmental morphogenesis and genetic manipulation in tissue and cell cultures of the Gramineae. Adv. Genet. 24, 431-499. Moyer, J. W., and Collins, W. W. (1983). ‘Scarlet’ sweet potato. HortScience 18, 111-112. Muller, E.. Brown, P. T. H.. Hartke, S., and Lorz, H. (1990). DNA variation in tissue-culture derived rice plants. Theor. Appl. Genet. 80, 673-679. Munksgaard, D., Mattsson, O., and Okkels, F. T. (1995). Somatic embryo development in carrot is associated with an increase in levels of S-adenosylmethionine, S-adenosylhomocysteine and DNA methylation. Physiol. Plant. 93, 5 - 10. Murashige, T., and Nakano, R. (1965). Morphogenetic behavior of tobacco tissue cultures and implication of plant senescence. Am. J. Bor. 52, 819-827. Murashige, T.,and Nakano, R. (1966). Tissue culture as a potential tool in obtaining polyploid plants. J. Heredity 57, 115-118. Murashige, T., and Nakano, R. (1967). Chromosome complement as a determinant of the morphogenic potential of tobacco cells. Am. J. Bot. 54, 963-970. Murata, M., and Orton, T. J. (1983). Chromosomal structural changes in cultured callus cells. In Vitro 19, 83-89. Murata, M., and Orton, T. J. (1984). Chromosome fusions in cultured cells of celery. Can. J. Genet. Cytol. 26, 395-400. Nabors. M. W., Daniels, A,, Nadolny, L.. and Brown, C. (1975). Sodium chloride tolerant lines of tobacco cells. Planr Sci. Lett. 4, 155-159. Nabors, M. W., Gibbs, S. E., Bernstein, C. S., and Meis, M. E. (1980). NaCI-tolerant tobacco plants from cultured cells. Z. Pflunzenphysiol. 97, 13- 17. Nadolska-Orczyk, A,, and Malepszy, S. (1989). In-vitro culture of Cucumis saliva L. 7. Genes controlling plant regeneration. Theor. Appl. Genet. 78, 836-840. Nevers, P., Sheperd, N. S., and Saedler, H. (1986). Plant transposable elements. Adv. Bot. Res. 12, 104- 194. Newton, R. J., Bhaskaran, S., Puryear, J. D., and Smith, R. H. (1986). Physiological changes in cultured sorghum cells in response to induced water stress. 11. Soluble carbohydrates and organic acids. Plant Physiol. 81, 626-629. Nick, H., Bowen, B., Ferl, R. J., and Gilbert. W. (1986). Detection of cytosine methylation in the maize alcohol dehydrogenase gene by genome sequencing. Nature (London) 319, 243-246. Nielsen, M. T. (1989). Registration of KDH 926, KDH 959, and KDH 960 tobacco germplasm lines with different levels of trichome exudate constituents. Crop Sci. 29, 1584. Nowak, J., Matheson. S. L., MacLean, N. L., and Havard, P. (1992). Regenerative trait and cold hardiness in highly productive cultivars of alfalfa and red clover. Euphvtica 59, 189-196. Nowick, E. M., Cao, J., and Rush, M. C. (1988). Genetic basis of selected somaclonal variants in the rice cultivar ‘Lemont.” SABRA0 20, 1-10. Oberle, I., Rousseau, F., Heitz, D., Kretz, C., Devys, D., Hanauer, A,, Boue, J., Bertheas, M. F., and Mandel, J. L. (1991). Instability of a 550-base pair DNA segment and abnormal methylation in fragile->( syndrome. Science 252, 1097-1 102. Oberthur, L. E., Harrison, S. A,, Croughan, T. P.,and Long, D. L. (1993). Inheritance of improved leaf rust resistance in somaclones of wheat. Crop Sci. 33, 444-448.

236

R. R. DUNCAN

Okkels, F. T. (1988). A theory explaining formation of somatic embryogenic cells by auxin-induced inhibition of DNA methylation. Physiol. Plant. 73, 1 IA. Oono, K. (1985). Putative homozygous mutations in regenerated plants of rice. Mol. Gen. Genet. 198, 377-384. Orton, T. J. (1983). Experimental approaches to the study of somaclonal variation. Plant Mol. Biol. Rep. I, 67-76. Orton, T. J. (1984). Somaclonal variation: Theoretical and practical consideration. In “Gene Manipulation in Plant Improvement” (J. P. Gustafson, ed.), pp. 427-468. Plenum Press, New York. Orton, T. J. (1987). Genetic instability in celery tissue and cell cultures. Iowa State J. Res. 61,481498. Ozias-Akins, P., and Vasil, I. K. (1988). In vitro regeneration and genetic manipulation of grasses. Physiol. Plant. 73, 565-569. Palmgren, G., Mattsson, O., and Okkels, F. T. (1991). Specific levels of DNA methylation in various tissues cell lines, and cell types of Daucus carota. Plant Physiol. 95, 174-178. Pandeya, R. S . , Brandle, J. E., Ankersmit, J. C. D., Rogers, W. D., and Keller, W. A. (1991). Delfield flue-cured tobacco. Can. J. Plant Sci. 71, 575-577. Pechan, P. M., Bartels, D., Brown, D. C. W., and Schell, J. (1991). Messenger-RNA and protein changes associated with induction of Brassica microspore embryogenesis. Planta 184, 161165. Peschke, V. M., and Phillips, R. L. (1991). Activation of the maize transposable element Suppressormutator (Spm) in tissue culture. Theor. Appl. Genet. 81, 90-97. Peschke, V. M., and Phillips, R. L. (1992). Genetic implications of somaclonal variation in plants. Adv. Genet. 30,41-75. Peschke, V. M., Phillips, R. L., and Gengenbach, B. G. (1987). Discovery of transposable element activity among progeny of tissue-culture derived maize plants. Science 238, 804-807. Peschke, V. M., Phillips, R. L., and Gengenbach, B. G. (1991). Genetic and molecular analysis of tissue-culture derived Ac elements. Theor. Appl. Genet. 82, 121-129. Peterson, P. A. (1993). Transposable elements in maize: Their role in creating plant genetic variability. Adv. Agmn. 51, 79-124. Phillips, R. L., Kaeppler, S. M., and Peschke, V. M. (1990). Do we understand somaclonal variation? In “Proceedings of the 7th International Congress on Plant Tissue Cell Culture” (H. 1. J. Nijkamp, L. H. W. van der Plas, and J. van Aartrijk, eds.), pp. 131-141. Kluwer, Dordrechr. Phillips, R. L., Plunkett, D. J., and Kaeppler, S. M. (1992). Novel approaches in the introduction of genetic variation and plant breeding implications. In “Plant Breeding in the 1990’s” (H. T. Stalker and J. P. Murphy, eds.), pp. 389-408. CAB International, Wallingford, UK. Pickering, R. A. (1989). Plant regeneration and variants from calli derived from immature embryos of diploid barley (Hordeum vulgare L.) and H . vulgare L. X H . bulbosum L. crosses. Theor. Appl. Gener. 78, 105-112. Piqueras, A., Olmos, E., and Hellin, E. (1994). Cytological changes related to salt tolerance in embryogenic callus of Citrus limon. Plant Cell Tissue Organ Culture 39, 13-18. Potter, R., and Jones, M. G. K. (1991). An assessment of genetic stability of potato in vitro by molecular and phenotypic analysis. Plant Sci. 76, 239-248. Quemada, H., Roth, E. J., and Lark, K. G. (1987). Changes in methylation of tissue cultured soybean cells detected by digestion with the restriction enzymes Hpa I1 and Mspl. Plant Cell Rep. 6, 63-66. Quesenbeny, K. H., and Smith, R. R. (1993). Recurrent selection for plant regeneration from red clover tissue culture. Crop Sci. 33, 585-589. Raemakers, C. 1. 3. M., Jacobsen, E., and Visser, R. G. F. (1995). Secondary somatic embryogenesis and applications in plant breeding. Euphytica 81, 93-107.

VARIATION AND CROP IMPROVEMENT

237

Rains, D. W., Croughan, T. P., and Stavarek, S. J. (1980). Selection of salt-tolerant plants using tissue culture. In “Genetic Engineering of Osmoregulation: Impact on Plant Productivity for Food, Chemicals, &Energy” (D.W. Rains, R. C. Valentine, and A. Hollaender, eds.), pp. 279292. Plenum Press, New York. Rakoczy-Trojanowska, M., and Malepszy, S. (1993). Genetic factors influencing regeneration ability in rye (Secale cereale L.). I. Immature inflorescences. Theor. Appl. Genet. 86, 406-410. Rakoczy-Trojanowska, M.,and Malepszy, S. (1995). Genetic factors influencing the regeneration ability of rye (Secale cereale L.) 11. Immature embryos. Euphytica 88, 233-239. Ramulu, K . S. (1975). Mutation breeding in sorghum. Z. Pflanzenzucht 74, 1-17. Ramulu, K. S., Devreux, M., Ancora, G., and Laneri, U . (1976). Chimerism in Lycopersicum peruvianum plants regenerated from in vitro cultures of anther and stem internodes. Z. Pjlanzenzucht. 76, 299-3 19. Ramulu, K. S., Dijkhuis, P., Roest, S., Bokelmann, G. S., and de Groot, B. (1986). Variation in phenotype and chromosome number of plants regenerated from protoplasts of dihaploid and tetraploid potato. Plant Breed. 97, 119-128. Rao, I . M., Roca, W. M., Ayarza, M. A., Tabares, E., and Garcia, R. (1992). Somaclonal variation in plant adaptation to acid soil in the tropical forage legume Selosanthes guianensis. Plant Soil 146, 21-30. Reaney. M. J. T., and Gusta, L. V. (1987). Factors influencing the induction of freezing tolerance by abscisic acid in cell suspension cultures of Bromis inermis Leyss and Medicago sativa L. Plant Physiol. 83, 423-427. Reaney, M. J. T., Gusta, L. V., and Abrams, S. R . (1989). The effects of abscisic acid, kinetin and gibberellic acid on freezing tolerance in smooth bromegrass (Bromis inermis Leyss) cell suspensions. Can. J. Bot. 67, 3640-3646. Reinbothe, C., Tewes, A,, and Reinbothe, S. (1992). Altered gene expression during somatic enibryogenesis in Nicotiana plumbaginifolia and Digitalis lanata. Plant Sci. 82, 47-58. Reish, B., and Bingham, E. T. (1980). The genetic control of bud formation from callus cultures of diploid alfalfa. Plant Sci. Lett. 20, 71-77. Robertson, A. J., Gusta, L. V., Reaney, M. J. T..and Ishikawa, M. (1987). Protein synthesis in bromegrass (Bromis inermis Leyss) cultured cells during the induction of frost tolerance by abscisic acid or low temperature. Plant Phvsiol. 84, 1331- 1336. Ronchi, V. N., Bonatti, S., and Turchi, G . (1986). Preferential localization of chemically induced breaks in heterochromatic regions of Vicia faba and Allium cepa chromosomes. 1. Exogenous thymidine enhances the cytological effects of 4-epoxyethyl- I ,2-epoxy-cyclohexane. Environ. EXP.BOI. 26, 115-126. Rowland, G. G., McHughen, A,, and McOnie, C. (1988). Field evaluation on nonsaline soils of a somaclonal variant of McGregor flax selected for salt tolerance in v i m . Can. J . Plant Sci. 68, 345-349. Rowland, G. G., McHughen, A., and Bhatty, R. S. (1989). Andro flax. Can. J. Plant Sci. 69,911913. Rowland, G. G., McHughen, A., Gusta, L. V., Bhatty, R. S., MacKenzie, S. L., and Taylor, D. C. (1995). The application of chemical mutagenesis and biotechnology to the modification of linseed (Linum usitatissimum L.). Euphvtica 85, 317-321. Rufty, R. C., Wernsman, E. A,, Main. C. E., and Gooding, G. V., Jr. (1990). Registration of NCBMR 42 and NC-BMR 90 germplasm lines of tobacco. Crop Sci. 30, 241-242. Sacristan, M. D. (1971). Karyotypic changes in callus cultures from haploid and diploid plants of Crepis capillaris (L.) Walk. Chromosoma 33, 273-283. Saunders, J. W., Acquaah, G . , Renner, K . A., and Doley, W. P. (1992). Monogenic dominant sulfonylurea resistance in sugarbeet from somatic cell selection. Crop Sci. 32, 1357- 1360. Saxena, P. K., and King, J. (1988). Herbicide resistance in Datura innoxia. Cross-resistance of sulfonylurea-resistant cell lines to imidazolinones. Plant Phvsiol. 86, 863-867.

238

R. R. DUNCAN

Scowcroft, W. R. (1985). Somaclonal variation: The myth of clonal uniformity. I n “Genetic Flux in Plants” (B. Hohn and E. S. Dennis, eds.), pp. 217-245. Springer-Verlag. Wien. Sears, R. G., Cox, T. S., and Paulsen, G. M. (1992). Registration of KS89WGRC9 stress-tolerant hard red winter wheat germplasm. Crop Sci. 32, 507. Shepard, J. F., Bidney, D., and Shahin, E. (1980). Potato protoplasts in crop improvement. Science 208, 17-24. Shimada, T., and Makino, T. (1975). In vitro culture of wheat. 111. Anther culture of the A genome aneuploids in common wheat. Theor. Appl. Gener. 46, 407-410. Shoemaker, R.C., Amberger, K. A., Palmer, R. G.,Oglesby, L., and Ranch, J. P. (1991). Effectsof 2,4-dichlorophenoxyaceticacid concentration on somatic embryogenesis and heritable variation in soybean (Glycine m a L. Mer. R.), In Vitro Cell Dev. Eiol. 27, 84-88. Sibi, M. (1976). La notion de programme genetique chez les vegetaux superieurs. 11. Aspect experimental: Obtention de variants par culture de tissus in vitro sur Lacruca sativa L. Apparition de vigueur chez les croisements. Ann. I’Amelior. Plantes 26, 523-547. Skirvin, R. M. (1978). Natural and induced variation in tissue culture. Euphyrica 27, 241-266. Skirvin, R. M., and Janick, J. (1976a). Tissue culture-induced variation in scented Pelargonium spp. J . Am. Soc. Hort. Sci. 101, 281-290. Skirvin, R. M., and Janick, J. (1976b). ‘Velvet Rose’ felargonium, a scented geranium. HortScience 11, 61-62. Skirvin, R. M., McPheeters, K. D., and Norton, M. (1994). Sources and frequency of somaclonal variation. HortScience 29, 1232- 1237. Smith, D. L., and Krikorian, A. D. (1989). Release of somatic embryogenic potential from excised zygotic embryos of carrot and maintenance of proembryonic cultures in hormone-free medium. Am. J. Bot. 76, 1832-1843. Smith, M. K., and Drew, R. A. (1990). Current applications of tissue culture in plant propagation and improvement. Aust. J. Plant Physiol. 17, 267-289. Smith, R. H., and Bhaskaran, S. (1988). Sorghum cell culture: Somaclonal vanationlscreening. Iowa State J. Res. 62, 571-585. Smith, R. H., Bhaskaran, S., and Schertz, K. F. (1983). Sorghum plant regeneration from aluminum selection media. Plant Cell Rep. 2, 129-132. Smith, R. H., Duncan, R. R., and Bhaskaran, S. (1993). In vitro selection and somaclonal variation for crop improvement. In “International Crop Science Congress. I” (D. W. Buxton, ed.), Chap. 82, pp. 629-633. July 1992, Ames, IA. Crop Sci. SOC.Am., Madison, WI. Smith, R. R., and Quesenbeny, K. H. (1995). Registration of NEWRC red clover germplasm. Crop Sci. 35, 295. Spofford, I. B. (1976). Position-effect variegation in Drosophila. In “The Genetics and Biology of Drosophila” (M. Ashburner and E. Novitski, eds.), pp. 955-1018. Academic Press, New York. Springer, W. D., Green, C. E., and Kohn, K. A. (1979). A histological examination of tissue culture initiation from immature embryos of maize. frotoptasma 101, 269-281. Stange, L. (1984). Cellular interactions during early differentiation. In “Encyclopedia of Plant Physiology, New Series. Vol. 17” (H. F. Linskens and J. Heslop-Harrison, eds.), pp. 424-452. Cellular Interactions, Springer-Verlag, Berlin. Stephens, P. A., Widholm, J. M., and Nickell, C. D. (1990). Iron deficiency chlorosis evaluation of soybean tissue culture. Theor. Appl. Genet. 80, 417-420. Sun, Z. X., Zhao, C. Z., Zheng, K. L., Qi, X. F., and Fu, Y. P. (1983). Somaclonal genetics in rice, Oryza sativa L. Theor. Appl. Genet. 67, 67-73. Sunderland, N. (1977). Nuclear cytology. In “Plant Cell and Tissue Culture’’ (H. E. Street, ed.), pp. 177-206. Univ. California Press, Berkeley. Swanson, E. B., Coumans, M. P., Brown, G. L., Patel, J. D., and Beversdorf, W. D. (1988). The characterization of herbicide-tolerant plants in Erassica napus L. after in vitro selection of microspores and protoplasts. Plant Cell Rep. 7, 83-87.

VARIATION AND CROP IMPROVEMENT

239

Swartz, H. J., Galletta, G . J., and Zimmerman, R. H. (1981). Field performance and phenotypic stability of tissue-culture-propagated strawberries. J. Am. Soc. Hort. Sci. 106, 667-673. Swedlund, B., and Vasil, I. K. (1985). Cytogenetic characteristics of embryogenic callus and regenerated plants of Penniserum americanum (L.) K. Schum. Theor. Appl. Genet. 69, 575581.

Symillides, Y.. Henry, Y., and De Buyser, J. (1995). Analysis of Chinese Spring regenerants obtained from short- and long-term wheat somatic embryogenesis. Euphyrica 82, 263-268. Thomas, E., Bright, S. W. J., Franklin, J., Gibson, R. W., Lancaster, V., and MiWin, B. J. (1982). Variation amongst protoplast-derived potato plants (Solanum tuberosum cv. Maris Bard). Theor. Appl. Genet. 62, 65-68. Tomes, D. T., and Smith, 0. S. (1985). The effect of parental genotype on initiation of embryogenetic callus from elite maize (Zea mays L.) germplasm. Theor. Appl. Gener. 70, 505-509. Toncelli, F., Martini. G., Giovianazzo. G.. and Nuti Ronchi, V. (1985). Role of permanent dicentric systems in carrot somatic embryogenesis. Theor. Appl. Genet. 70, 345-348. Tsai, C. K., Chien, Y. C., Ke, S. Q., He, Z. C., Jiang, R. X., Zhou, Y. L., Ye, Y. P., Hong, S. R., and Huang, R. H. (1992). Studies on the somaclonal variation of regenerated plants from protoplasts of Actinidia deliciosa. Acta Bot. Sinica 34, 822-828. van den Bulk, R. W., Loffler, H. J. M., Lindhout, W. H., and Koornneef, M. (1990). Somaclonal variation in tomato: Effect of explant source and a comparison with chemical mutagenesis. Theor. Appl. Genet. 80, 817-825. Van Harten, A. M., Bouter, H., and Broertjes, C. (1981). In virro adventitious bud technique for vegetative propagation and mutation breeding of potato (Solanum ruberosum L.), 11. Significance of mutation breeding. Euphyrica 30, 1-8. Van Swaaij, A. C., Jacobsen, E., Kiel, J. A. K. W., and Feenstra, W. J. (1986). Selection, characterization, and regeneration of hydroxyproline-resistant cell lines of Solanum tuberosum: Tolerance to NaCl and freezing stress. Physiol. Plant. 68, 359-366. Van Swaaij, A. C., Nijdam, H., Jacobsen. E., and Feenstra, W. J. (1987). Increased frost tolerance and amino-acid content in leaves, tubers, and leaf callus of regenerated hydroxyproline-resistant potato clones. Euphytica 36, 369-380. Vasil, 1. K. (1983). Regeneration of plants from single cells of cereals and grasses. In “Genetic Engineering of Eukaryotes” (P. Lurquin and A. Kleinhoes, eds.), pp. 233-252. Plenum Press, New York. Vasil, I. K. (1987). Developing cell and tissue culture systems for the improvement of cereal and grass crops. J. Planr Phvsiol. 128, 193-218. Vasil, V., and Vasil, 1. K. (1982). The ontogeny of somatic embryos of Penniserum americanum (L.) K. Schum. I. In cultured immature embryos. Bor. Gaz. 143, 454-465. Vasil, V., Lu, C.-Y., and Vasil, 1. K. (1985). Histology of somatic embryogenesis in cultured immature embryos of maize (Zea mays L.). Protoplasma 127, 1-8. Veilleux, R. E. (1994). Development of new cultivars via anther culture. HorrScience 29, 12381241, Vodkin, L. 0. (1989). Transposable element influence on plant gene expression and variation. In “The Biochemistry of Plants. Vol. 15” (A. Marcus, ed.), pp. 83-132. Academic Press, San Diego. Waldo, G. F. (1977). ‘Thornless Evergreen’-Oregon’s leading blackberry. Fruit Var. J. 31, 26-30. Walker, G . C. (1984). Mutagenesis and inducible responses to deoxyribonucleic acid damage in Escherichia coli. Microbiol. Rev. 48, 60-93. Waskom, R. M., Miller, D. R., Hanning, G. E., Duncan, R. R., Voigt, R. L., and Nabors, M. W. (1990). Field evaluation of tissue-culture-derived sorghum for increased tolerance to acid soils and drought stress. Can. J. Plant Sci. 70, 997-1004. Weinburg, G., Ullman, B., and Martin, D. W., Jr. (1981). Mutator phenotypes in mammalian cell

R. R. DUNCAN mutants with distinct biochemical defects and abnormal deoxyribonucleoside triphosphate pools. Proc. Natl. Acad. Sci. U.S.A. 78, 2447-2451. Wen, F. S., Sorenson, E. L., Barnett, F. L., and Liang, G. H. (1991). Callus induction and plant regeneration from anther and inflorescence culture of sorghum. Euphytica 52, 177- 181. Wenzel, G., and Foroughi-Wehr, 9 . (1990). Progeny tests of barley, wheat, and potato regenerated from cell cultures after in vitro selection for disease resistance. Theor. Appl. Genet. 80, 359365. Wersuhn, G., Kalettka, T., Gienapp, R., Reinike, G., and Schulz, D. (1994). Problems posed by in vitro selection for aluminum-tolerance when using cultivated plant cells. J . Plant Physiol. 143, 92-95. Wu, J. (1986). Breeding haploid corn by anther culture. In “Haploids of Higher Plants in Vitro” (H. Hu and H. Yang, eds.), pp. 149-161. Springer-Verlag, Berlin. Xie, Q. J. (1990). Rice improvement through tissue culture: Inheritance of somaclonal variation and improvement of somaculture techniques. Ph.D. Dissertation, Louisiana State Univ., Baton Rouge, LA. [Dissert. Abstr. 91232471 Xie, Q . J., Rush, M. C., and Cao, J. (1990). Somaclonal variation for disease resistance in rice (Oryzu sativu L.). In “Pest Management in Rice” (B. T. Grayson, M. 9. Green, and L. G. Copping, eds.), pp. 491-519. Elsevier, New York. Xie, Q. J., Rush, M. C., and Oard, J. H. (1995). Homozygous variation in rice somaclones: Nonrandom variation instead of mitotic recombination. Crop Sci. 35, 954-957. Xin, Z., and Li, P. H. (1992). Abscisic acid-induced chilling tolerance in maize suspension cultured cells. Planr Physiol. 99, 707-71 1 . Xin, Z., and Li, P. H. (1993). Alteration of gene expression associated with abscisic acid-induced chilling tolerance in maize suspension-cultured cells. Plant Physiol. 101, 277-284. Yang, X., and Fu, H. (1989). Hua-03: A high protein indica rice. Inr. RiceRes. Newslett. 14, 14-15. Yang, Y.W., Newton, R. J., and Miller, F. R. (1990). Salinity tolerance in Sorghum. 11. Cell culture response to sodium chloride in S. bicolor and S. halepense. Crop Sci. 30, 781-785. Yarlett, N., and Bacchi, C. J. (1988). Effect of DLa-difluoromethylomithine on methionine cycle intermediates in Trypanosoma brucei brucei. Mol. Biochem. Purasitol. 27, 1- 10. Ye, I. M., Kao, K. N., Harvey, 9. I,., and Rossnagel, 9. G. (1987). Screening salt-tolerant barley genotypes via F, anther culture in salt stress media. Theor. Appl. Genet. 74, 426-429. Yisraeli, J., and Szyf, M. (1984). Gene methylation patterns and expression. In “DNA Methylation: Biochemistry and Biological Significance” (A. Razin, H. Cedar, and A. D. Riggs, eds.), pp. 353-378. Springer-Verlag, New York. Zapata, F. J., Alejar, M. S., Tomzo, L. 9. Novero, A. U.,Singh, V. P., and Senadhira, D. (1991). Field performance of anther-culture-derived lines from F, crosses of Indica rices under saline and nonsaline conditions. Theor. Appl. Genet. 83, 6- 1 1 . Zehr, B . E., Williams, M. E., Duncan, D. R., and Widholm, J. M. (1987). Somaclonal variation in the progeny of plants regenerated from callus cultures of seven inbred lines of maize. Can. J. Bot. 65, 491-499. Zhao, Y., He, X., Wang, J., and Liu, W. (1990). Anther culture 28: A new disease-resistant and high-yielding variety of winter wheat. In “Biotechnology in Agriculture and Forestry 13: Wheat” (Y. P. S. Bajaj, ed.), pp. 353-362. Springer-Verlag, Berlin. Zhu, D., and Pan, X. (1990). Rice (Oryzu sativa L.): Guan 18: An improved variety through anther culture. in “Biotechnology in Agriculture and Forestry 2. Haploids in Crop Improvement I.” (Y.P. S. Bajaj, ed.), pp. 204-211. Springer-Verlag, Berlin.

GEOSTATISTICAL ANALYSIS OF A SOIL SAL~VITY DATASET G. Bourgault,' A. G. Journe1,lJ. D. Rhoades,z D. L. Convin,z and S. M. Lesch* 'Geological and Environmental Sciences Department, Stanford University, Stanford, California 94305 WSDA-ARS, U.S. Salinity Laboratory, 450 Big Springs Road, Riverside, California 92507

1. Introduction A. Broadview Salinity Data Set 11. Exploratory Data Analysis A. Electrical Conductivity Data B. Electromagnetic Data C. Soil Type Differentiation D. Spatial and Variogram Analysis 111. Mapping the EC, Distribution A. Comments on Results B. Influence of E M , Data IV.Filtering Structures V. Spatial Cluster Analysis VI. Stochastic Imaging A. Simulation Algorithm VII. Assessment of Spatial Uncertainty VIII. Ranking of Stochastic Images IX.Conclusions References

I. INTRODUCTION Rather than relating a specific case study with specific goals, this study aims at presenting a range of potential applications of modem geostatistics to soil survey problems using a real data set. Typical of the development of geostatistics, a discipline led by engineers, many new algorithms, although well published in 241 Advmrer in Agronmy, lbltime 58 Copyright 6 I997 by Academic Press, Inc. All rights of reproduction in any form reserved

2 42

G. BOURGAULT ET AL.

their field of inception (mining and petroleum), have not yet found their way into mainline statistical books; hence, they may not be readily accessible to professionals outside the extractive industry (Deutsch and Journel, 1992; Dimitrakopoulos, 1993; Isaaks and Srivastava, 1989; Soares, 1993). It is hoped that this presentation may raise enough interest among soil scientists so that they find it worth their time to learn more about modem geostatistical concepts of data analysis, estimation, uncertainty assessment, and stochastic imaging. All results presented in this study were obtained using standard mapping routines and the public-domain GSLIB software (Deutsch and Joumel, 1992). FORTRAN source code of the latter is public domain; hence, it is available to whomever wishes to understand the details of any particular algorithm and/or to modify it to fit any particular problem at hand. A common denominator of many soil sciences data sets is the sparsity of “hard” or direct measurements of the primary variable of interest, usually balanced by the prevalence of “soft” or indirect information related to the primary variable. Examples of hard data are core measurements and more generally expensive field-based data as opposed to soft data obtained, e.g., from remote sensors. Geographical information systems (GIS) and geostatistics pursue a similar objective-that of providing tools for the integration of different objectives, and that of providing tools for the integration of different information sources with varying relevanceheliability to build maps that summarize and expand the original hard data set. Geostatistics propose to add to the GIS toolbox various spatial data analysis tools to explore and model patterns of spacehime dependence between the data available. The resulting numerical models, e.g., variograms or conditional distributions, can then be put to use for various mapping purposes and an assessment of the reliability of such maps. Just as there is no unique or optimal sequence in using GIS tools, such as concatenation, intersection, or interpolation, there is no unique geostatistical approach to spatially distributed data. Many alternative covariance/variogram models can be fitted to the same data set depending on ancillary information available to the operator (including his or her own prior experience); there are many different kriging algorithms (generalized least squares regression) that can be used toward the same mapping goal depending on which particular aspect of the data one wishes to capitalize on. What may be lost to statistical objectivity (an extremely debatable concept) is gained in flexibility and ability to handle soft, yet essential information of various types. It is better to have a somewhat subjective but accurate assessment that accounts for all relevant information than a supposedly objective assessment that misses critical aspects of the problem. This chapter will illustrate the toolbox aspect of geostatistics, presenting several alternative ways to reach the same goal and proposing cross-validation exercises to help the operator in his or her decision.

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 243

A. BROADVIEW SALINITYDATASET As mentioned previously, the data set used in the following study is more a support for demonstration of geostatistical algorithms than the data base of an actual case study. For such demonstrative purposes, the name of the location involved and even the measurement units of the data could have been omitted, and coordinate values could have been changed by any one-to-one monotonic transform leaving unchanged the relative patterns of spatial variability of the various attributes. The actual and complete Broadview salinity data base and its statistical analysis are presented in various papers and reports of the U.S. Salinity Laboratory (Lesch et al., 1995a,b). The reader is referred to these papers for any question related to sampling and salinity assessment in the Broadview water district. The results of the present study are based on a limited data set ignoring, in particular, such critical variable as soil water saturation: they should only be used to assess the worth of adding geostatistical tools to GIS and other toolboxes available to the soil scientist. The Broadview data set covers approximately 6000 acres and comprises the following: a soil map digitized into 7 soil types (see Fig. la). 315 soil core measurements of electrical conductivity (EC,), taken at four depths (0-1, 1-2, 2-3, and 3-4 ft) (see Fig. lb). Unit is dSlm. 2385 measurements of soil vertical (EM,) and horizontal (EM,) electromagnetic response (see Fig. lc). Unit is dSlm. Each measurement is deemed representative of the vadose zone (upper 4 ft of soil). The extent of that electromagnetic information delineates the study area, as shown in Figs. lalc. For the purpose of this study, core EC, values are considered hard data directly related to soil salinity and represent the primary variable to be evaluated throughout the vadose zone. The electromagnetic induction readings represent a secondary variable less directly related to soil salinity; they are considered soft data used to complement the hard EC, data. With little loss of location accuracy, electromagnetic data have been relocated to the nodes of a 2D regular grid 100 x 100 m. The few grid nodes with no electromagnetic sample within a radius of 50 m were left uninformed. Another option could have been to interpolate the few missing nodal values. The grid includes 2385 E M , measurements. To allow standardization of the grayscales, the E M , data plotted in Fig. lc have been rescaled by a factor equal to the ratio of the standard deviations of original EC, and E M , data. Although measured in the same unit (dSlm),E C , and E M , data have different

G. BOURGAULT ET AL.

244 9400.

a: Soil categories

7 6 5 4 3 2 1

8400. 7400. 6400. 5400. 4400.

b: Hard ECe

9400.

25.0 22.6

8400.

20.2 17.8

7400.

15.4 13.0 10.6

6400.

8.2

5.8

5400. 4400. 16105

16io5.

3.4 1.oo 18105.

20105.

22105.

24105.

i8io5.

2oio5.

22105.

24105.

Figure 1 Broadview salinity data sets. (a) Seven soil types, (b) 315 core measurements of electrical conductivity (hard EC, data), and (c) gridded electromagnetic data (soft E M , data). To allow easier comparison of spatial structures, the E M , variance has been rescaled to that of the hard E C , data.

mean values (5.24 and 1.46, respectively). Such differences in mean values between hard and soft data are not uncommon in the earth sciences and are filtered out by various unbiasedness constraints in the algorithms used. The soil map in Fig. la was digitized to allow marking each EC, and E M , sample location with a specific soil type denoted 1-7.

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 245

11. EXPLORATORY DATA ANALYSIS The aim of an exploratory data analysis (EDA) is to acquire an overall familiarity with the data, their interrelations, statistical grouping, spatial distribution, clustering, etc. At this stage, the operator should not be constrained by any specific goal but rather he or she should be attentive to any clue the data may give that may prove useful in later interpretations. Because geostatistics deals with spatial data, extensive use should be made of isopleth maps and GIs-related routines depicting the relations between data values and their space/time coordinates. Beware that random sampling (random drawing of sample coordinates) does not make the data values independent inasmuch as it is the physical generating process that makes the data dependent and not the human decision about where samples are taken.

A. ELECTRICAL CONDUCTMTY DATA Figure 2 shows the succession of four grayscale EC, maps corresponding to the four measurement depths. The vertically averaged map is that shown in Fig. lb. There appears to be a gradual increase in soil salinity with depth, corroborated by the histograms shown on the right in Fig. 2. A diagonal transect N120"E crosscutting the N3O"E elongation of the seven soil categories shown in Fig. la was defined, then EC, data values were plotted against their coordinate value along that transect (see Fig. 3). At each depth level, the n = 315 EC, data were ranked from r ( l ) = 1 to fin) = 315 and their standardized ranks vCi) = f i i ) / n ,or uniform scores distributed in [0,I], are grayscale plotted in Fig. 3. This uniform score transform allows identifying each level-specific EC, data set to the same uniform [0,1] distribution. This transform thus filters out the vertical trend previously observed and allows comparison of the strictly horizontal structures. The four N12O"E grayscale transects of uniform scores shown on the left in Fig. 3 show similarity of the horizontal variability of EC, data over the four depth levels. This is confirmed by the EC, uniform score (uscore) semivariograms calculated along the N120"E direction and given on the right in Fig. 3. Therefore, these uscores-standardized variograms can be pooled together into a single model valid for all four depth levels. The rank (or uscores) correlations between two vertically consecutive EC, data (thus with the same horizontal coordinates) are 0.64 for 1-2 ft, 0.80 for 2-3 ft, and 0.89 for 3-4 ft. Therefore, except for the first transition from 1 to 2 ft, the EC, data are quite redundant from one level to the next one: there is little gain to be expected from a 3D interpolation versus a much simpler 2D exercise using only data from the level being estimated.

G. BOURGAULT ET AL.

2 46

o.25,ECe:Soll depfh 0-1 ff.

25 0 226 20 2 17 8 154 130 106 82 58 34 1 00

Number of Data 315 mean 2.96 std. dev. 1.46 ml.of var 0.49 maximum 10.19 upper quanile 3.51 median 2.55 lower quattile 1.95 minimum 0.73

0 20 0 15

0.10 0 05 OW 00

50

15.0 Variable

10.0

20.0

25.0

ECe: Soll depfh 1-2 ft, -

9400

4

8400 7400 6400 5400 4400 Variable

.ECe: Soil depth 2-3 ft.

n,~i n

9400

Number of Data :315 mean 6 29 n d dev 474 cod of var 0 75 maximum 31 62 upper quartile 8 63 median 4 47 lower quartile 2 71 minimum 1 3 3

-1 E j q

8400

7400

0.08

0.04

6400

5400

i

0.M)

4400

O.( I

ECe: Soil depth 3-4ft.

9400

50

100 150 Vanable

200

250

ECe: Soil depth 3-4 ft.

$,

Number of 01 Data 315 mean 7 43 std Std dev 5 58 coef of var 0 75 maximum 40 10 u p p r quartile 9 82 median 6 07 lower quamle 3 47 minimum 0 82

008

8400

7400

p s

6400

U

g

006 004 0 02

5400

...

4400 16105

..

18105

OW

20105

I

.

22105

.

.

,

.

24105

.

00

50

100

150

200

250

Variable

Figure 2 Grayscale plots of EC, data and corresponding histograms showing increase with depth of soil salinity.

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 247

;i0

012:

0 80 0 70 060 0 50 0 40 030

Oo8-

004-

0 20 0 10

O0

0 0 0 - , , , , , , , . . ., ,.. ,.

..,...,

, , ,

X (N120E)

ECe 1-2 ft. direction N l Z O E

1

0.121

- 0 0 . 0 0 14500

16500

18500 20500 X (N120E)

500.

1000.

1500.

2000

2500.

Distance

22500

ECe 2-3ft. direction N12OE

0o

o 0

500

l

1000

1500

, 2000.

2500

Distance

ECe 3-4 ft. direction N120E

oil----0 00 0

500.

1000

1500.

2000.

2500

Distance

Figure 3 Grayscale plots of standardized ranks (uniform scores) of level-specific EC, data projected along a N120"E transect. The correspondingN120"E semivariograms are given to the right.

2 48

G. BOURGAULT ET AL.

For the remainder of that study and for reason of conciseness, only the vertically averaged EC, data (see Fig. lb) were considered together with the corresponding 2D-distributed electromagnetic data (see Fig. lc). is the cumulaNote that the uniform score transform x + F,(x), where Fx(*) tive distribution function (cd8 of random variable X , is the first step of a normal score transform (Deutsch and Journel, 1992, p. 138). Unless properties specific to the Gaussian distribution are to be called for, there is no need for going beyond the standardized rank transform F,(.). This rank transform, by definition, preserves the rank of the data as does the commonly used, albeit somewhat arbitrary, log transform. From the histograms of Fig. 2, the EC, data appear neither normal nor lognormal distributed; this was confirmed by probability graph plots (Deutsch and Journel, 1992, p. 201) not shown here.

B. ELECTROMAGNETIC DATA Figure 4a shows an extreme redundancy between the two secondary data, vertical (EM,) and horizontal (EM,) electromagnetic measurements. This redundancy was confirmed by maps and variograms analysis (not shown here). Because E M , has slightly better correlation with colocated vertically averaged EC, (see Figs. 4b and 4c), only E M , was retained as a source of secondary data for the rest of the study. The grayscale map of this E M , data was shown in Fig. lc. Observe on Fig. 4b the nonlinear relation EC,-EM, To linearize that relation and capitalize on linear regression tools (such as kriging), a transform of the variables is necessary. If the two variables were to be made Gaussian distributed, a normal score transform (Isaaks and Srivastava, 1989, p. 138) would be necessary. Because the histograms of the original EC, and E M , values are not lognormal, the log transform does not identify the normal score transform. In any case, there is currently no need for any Gaussian assumption; hence, the rank transform (uniform scores) is enough. Figure 4d shows the scattergram of the uniform score transforms of E C , and E M , data. Note how the rank transform has succeeded in linearizing the original regression between E C , and E M , data (see Fig. 4b). The linear rank regression remains though heteroscedastic, in the sense that higher ranks of EC, are better predicted by corresponding high E M , ranks than are lower ranks. These rank regressions will be fine-tuned later using soil type information.

C. SOILTYPE DIFFERENTIATION The 3 15 vertically averaged EC, data values were plotted against their coordinates along the N120"E transect and grayscale coded for soil type (see Fig. 5a). Figure 5b provides a similar profile for the 2385 E M , data.

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 249 3.0

-

25.0

a: EMh EMv

b: ECe-EMv 6

2.51

.

2".

EMv

EMv

,,,,~ c: ECe - EMh

Number of data 315 XVariaMe: m n 1.47 std. 0.52 Y VariaMe: mean 5.24 sld. dav. 3.32

'

Numberoldala315 X Variable: m an 1.01 ad. f v . 0.40

0.2. ^

EMh

^

f' .: _ I'

uscEMv

Figure 4 Scattergrams. (a) Soft E M , versus soft E M , data; (b) hard E C , versus soft E M , data; ( c ) hard EC, versus soft E M , data; and (d) uniform scores of EC, versus E M , data.

Except for soil types I and 6 (the latter being nonrepresentative because of lack of data), the ranges of EC, values appear homogeneous across soil types. This is confirmed by the EC, histograms per soil type (not shown). The histograms of E M , data per soil type (Fig. 6 ) would lead one to differentiate the following two groups based on mean E M , value: A first group including soil types 1, 4, 5 , and 7 with a mean E M , value around 1.6 A second group including soil types 2, 3, and 6 with a lower mean E M , value around 1.2

Note that these two groups are intermingled in space. E M , data considered to be exhaustively sampled, are used to inform unsampled primary EC, values. To investigate how soil type influences the relation, EC,-EM,, the seven soil-specific rank scattergrams of uscores of colocated E C , and E M , data are shown in Fig. 7. It appears that the linear rank regression

G. BOURGAULT ET AL.

2 50

b: EMv

a: ECe

25.0

;oil type

3

,

0

0

7

7 6 5 4

3 2 1 0.0 I

, ,

,

14dob. i6doo.

. isdoo. . ,

*

,

,

,

, ,

,

zodoo. zzdoo. zr

I.

0.00

, , 14d00.

X (N120E)

.

,

isdoo. isdoo.

20600. 22d00. zr X (N120E)

I

Figure 5 Profiles of EC, and E M , data along a N120"E transect. The grayscale code indicates soil type.

observed when data are pooled across all soil types (Fig. 4d) is in fact constituted by several different regressions better fitted by power models of the type, vECe

=

[vEM,]wsI

(1)

where vEC,and vEMv are the uniform scores valued in [0,1], and cost is a power exponent specific to each soil type st. The regression fits shown in Fig. 7 correspond to cost = 1 (linear rank regression) for soil cost > 1 for soil types 1, 4, 5, and 7.

types overall and type 6.

qf< 1 soil types 2 and 3.

These power regression models are used later to incorporate the secondary E M , information while accounting for soil type. Were the electromagnetic data EM, a variable of primary interest, further considerations would be given to splitting its spatial distribution into soil type groups. However, because E M , represents only secondary information destined to supplement the hard EC, data, it was decided to model its spatial distribution across all soil types. The only soil type differentiation kept is that of the previous rank regression power models.

D. SPATIALAND VAR~OGRAM ANALYSIS In preparation for spatial interpolation of the EC, values, joint variogram analysis of the EC,-EM, data was performed across all soil types. Figure 8 shows the sample (cross)semivariogramsfor the original EC, and

€Mv (all)

EMV (vpe 1)

01 DBIB 2%

Number 01 Dam 755 mean 1 6 4 sld dev 0 4 5

coef 01 var 0 27 maximum 3 09

maxlmm 309

94

Vanable

Varrable

0.w m. 0.54

,

,

.

1 .54

,

,

.

,n, ,

,

2.54

,

,

3.54

Variable

EMV (type 4)

-EMV

Numbrof Data 09

Variable

EMv (type 6)

5)

Numbor 01 Data 187

Variable

NWM 01 Datn 69

mean 1.17 std. dw. 0.20 d u a l 0.17 ms*imurn 1.57 upperquame 1.29 m6dm I . % , -1.

1-r

0.12

.w mmimum 0 . 4 quam1e

1

maxlmum 2.44 Upper quaoil0 1 . y median 1. 7 lower quanlle 1.63 mlnlmum 1.15

E

E

U

U

0.04

0.W

0.00

2.54

VariaMe

3.34

0.00

3.54

Variable

Figure 6 Histograms of E M , data across all soil types and per soil type. The white histograms correspond to soil type groups 2 . 3. and 6 with lower mean, the dark histograms correspond to groups 1. 4, 5 , and 7 with higher mean.

G. BOURGAULT ET AL.

2.52

Uscores ECeEMv, soll type 1

Uscores ECe-EMv, all roll types

uscEMv

uscEMv

Uscores ECe-EMv, soll type 3

Uscores ECe-Ehlv, soil type 2

Number of data 83

1%

XVarlabkm n 24 sld. YVarl.Ma:mean nd. dev. 0.27 rank c o d a t ~ o n 0.88

Xv. .

.

w

8 0.4:

Number of data 38

02: 0.0 0.0'

std. dav. 0.21 rank corralstion 0.49 '

uscEMv

'00:2'' '014.

'

'OX

'

'0:8'

'

uscEMv

Uscores ECe-EMv, soll tyw 4

1

..

./

.

..

Uscores ECe-EMv, soil type 5

)

.

0.81

Number of data 60

.' *

YVadable: m std. 0.30 rank cornlatlon 0.66

uscEMv

uscEMv

Uscores ECe-EMv, soil rype 6

Uscorns ECe-Ehlv, soil type 7 I

I:::! -...I/. 0.01 0.0.

.* '

0:2.

* '

.00:4' '

Number of data 11

D:e'

uscEMv

'

rank cornlaurn 0.13 .0!8' . '

Number of data 15

I:# .

0.0 0.0'

'

'0:s ' '0:C

'

X V a ~ W b k man 72 Id. &v. YvamL:mean std. dav. 0.27 rank cormlaUon 0.92 ' O S . '0%' ' '

uscEMv

Flgure 7 Rank scattergrams of EC,-EM, data pooled across all soil types and per soil type. The regression curve shown is fitted using a power model.

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 253 ECe, N030E

ECe, N 12Of

14.0 11.2

14.0I

.

-

*

.. ... .... .. -

11.2

'

.'

8.4 5.6

2.8

-

2.8

0.0 . . . . . . . . . . . . . . . . . . . . . . . 0.0 600.0 1200.0 1800.0 2400.0 3 W . O

0.0

1

u 600.0 1200.0 1800.0 2400.0 3ooo.O

0.0

DiStam h

14.0 t 11.2 E

ECe-EMv, N030E

ECe-EMv, N120E

I4.O

:

e

.

.""~'~"."'..'"~~'' 600.0

1200.0

1800.0 2400.0

30000

EMv lrescaled). NO3OE

gj

-

8.4

-

I

. .

0.0 0.0

11.2

7

0.0

600.0

1200.0

1800.0 2400.0

L

0.0 0.0

800.0

1200.0

1800.0 2400.0

3000.0

EMv (rescaled). N12OE

14.01

3000.0

0.0 ' 00

~

"

1

600.0

"

~

12W.0

"

~

"

~

"

~

1800.0 2400.0

'

"

'

WW.0

Rgure 8 Sample (cross)semivariograms of EC,-EM, data and the model of coregionalization fitted.

EM, data in the two main directions, N30"E along soil types elongation and NI20"E across it. The E C , and E M , data used are those given in Figs. Ib and Ic. For reason of stability of future cokriging matrices, the secondary EM, data have been rescaled by a factor allowing identification of their variance to that of the primary data EC,; equivalently, one could have worked on (cross)correlograms instead of variograms.

~

~

~

2 54

G. BOURGAULT ET AL.

The solid lines in Fig. 8 show the fit by a model of coregionalization (Journel and Huijbregts, 1978, p. 172; Isaaks and Srivastava, 1989, p. 390). That model features An isotropic nugget effect accounting for about one-third of the total spatial variance of the EC, data A first isotropic structure of range 700 m accounting for another third of the EC, variance A second anisotropic structure of range 3000 m in the N120"E direction across soil types and 16,000 m in the N3O"E direction along soil continuity The model for E M , is similar although with lesser nugget effect due to the larger definition volume (averaging effect) of the electromagnetic data. The original EC, and EM, data were then normal score transformed (Deutsch and Journel, 1992, p. 138) so that both histograms identify a standard Gaussian distribution, and the corresponding sample (cross)semivariograms were calculated and modeled (see Fig. 9). The coregionalization model features the same characteristics as those fitted to the original data. Note that sampling fluctuations have not been significantly reduced by the normal score transform; this would have also been true had a log-transform been used. The uniform scores of the E M , and EC, data used are shown in Figs. 1Oa and lob. Compare these scores to the datq in Figs. Ic and lb, respectively: except for the different grayscales, they are essentially the same. Again, we prefer comparing data through the uniform standardization in [O,l] provided by the standardized ranks (uniform scores). Figure 1Oc shows the location of 26 EC, random samples taken from the 3 15 original EC, data; this subsample is used later in the cross-validation exercises.

111. MAPPING THE EC, DISTRIBUTION To demonstrate the various geostatistical mapping algorithms, four different approaches and two sampling cases are considered. EC, estimation is performed at each node of the 100 X 100-m grid covering the study area as defined by the template of electromagnetic data (see Fig. lc). The three different approaches are 1. Simple kriging (SK) (Deutsch and Journel, 1992, p. 62): EC, is estimated by a linear combination of the neighboring EC, data plus the overall EC, sample mean, m = 5.24, No secondary information is used; thus, this approach represents a base case. 2. Simple cokriging (coSK) (Deutsch and Joumel, 1992, p. 71): EC, is esti-

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 255 Nscores ECe, NWOE

Nscores ECe, N120E

#

''_

..

1.1

- 0. 1

0.3

0.0 0.0

600.0 1200.0 1800.0 2400.0 3000.0 Distance h

0.0 0.0

Nscores ECe-EMv. N030E

600.0

Nscores €0-EMv, N120E I

.L

800.0

1200.0 1800.0 2400.0 3 w o . O

Nscores EMv, NWOE

0.0

600.0

1200.0 1800.0 2400.0 3ooo.0

....... ~

1.0 L

0.0

1200.0 1WO.O 2400.0 3000.0

0.0;. 0.0

.. *

'

~~~

.

. ' ,... .. , ' , . . '

800.0

'. ..

*

.'

1200.0 1800.0 2400.0 3000.0

Nscores EMv, N120E

1-41

0.0

600.0

1200.0 1800.0 2400.0 3000.0

Figure 9 Sample (cross)sernivariograms of normal score EC,-EM, data and the model of coregionalization fitted.

mated by a linear combination of neighboring EC, data and E M , data rescaled to the EC, variance, u2 = 11.O. The coregionalization model shown in Fig. 8 is used. This is the most straightforward cokriging; it does not correct for the nonlinear relation observed between EC, and E M , data (see Fig. 4b). 3. Probability-field estimate (P-field): This approach is a variant of the p-field simulation algorithm introduced by Srivastava (1992); it accounts for the soil

256

G. BOURGAULT ET AL.

Figure 10 Grayscale maps o f uniforms scores o f E M , and EC, data

type information available at each location being estimated. A simple kriging of the EC, normal score is performed using only EC, normal score data. Adopting a Gaussian random function model for these normal scores (Deutsch and Journel, 1992, p. 136), the conditional distribution of any unsampled EC, normal score

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 257

value is Gaussian with mean and variance identified to the simple kriging mean and variance. The Gaussian conditional cumulative distribution function (cdf) can be denoted by Prob {Y(u) 5 y ( neighboring y(u,) data} = G ( Y - Y;K(ll)) u,,(u)

(2)

where Y(u) is the normal score transform of EC, at grid location u, y&(u) and u&(u) are the simple kriging estimate and variance using neighboring normal score EC, data (y(ua));and the function G(.) is the standard normal cdf Let q(u;p) be the corresponding conditional quantile function or inverse of the previous conditional cdf:

Srivastava’s p-field approach (Srivastava, 1992) consists of simulating a p-field, that is, a set of spatially correlated uniformly distributed p,(u) values, then transforming them through the previous quantile function into simulated normal score values y,(u) for EC,: Y&U) =

q(u;p,(u))

(4)

Because there may be several realizations (outcomes) for ~ , ~ (atu any ) location u, there may be several “simulated” realizations Y,~(U),hence the subscript notation s for simulation. The variant proposed here consists of determining, at each location u, a single value p(u) resulting in a single estimate y*(u) for the EC, normal score value: Y*W = q(u;p(u))

(5)

with p(u) = [vEM,(u)]”,,, as given by the power model in Eq. [ I ] , and sf is the soil type prevailing at location u. In words, the p-field value to be plugged into the conditional quantile function q(u;p) is the p value obtained by the regression model [Eq. (1) and Fig. (7)] specific to the soil type prevailing at u and to the uniform score vEMv(u) of the electromagnetic datum at u. This p-field approach requires the Gaussian random function model to determine the conditional cdf Eq. (2) from the only two parameters, mean and variance, provided by simple kriging. A final step back-transforms the normal score estimate y*(u) into EC, estimates expressed in the original EC, units. Note: The three different approaches proposed here to interpolate EC, values do not cover the range of different geostatistical algorithms that could be used for

258

G. BOURGAULT ET AL.

this purpose. The first two approaches proposed are the most straightforward and are likely to be familiar to many readers. The latter approach is a bit more involved; it is intended to give the reader a glimpse of the forefront of applied geostatistics in which new variants are constantly proposed to better match the problem at hand and the specific data available. For each of the previous three approaches, two sampling cases are considered: I . All 3 15 hard EC, data are used together with the (exhaustive)EM, and soil type information present at all nodes being estimated 2. A subsample of only 26 hard EC, data is used (see Fig. 10c) in addition to the previous E M , and soil type information

This latter sampling case allows a model-validation exercise using the remainder 289 hard EC, data. The problem with using such a small sample size (26) is the difficulty of doing any reliable statistical inference. We have decided to set apart the two problems of statistical inference and model validation of the estimation approaches proposed. More precisely, for the latter sampling case, although only 26 EC, data were retained for the various krigings, the statistics needed (histogram, variograms, and regression) are those established using all 3 15 hard data, i.e., the same statistics used for the first sampling case. This decision corresponds to the extractive industry practice of borrowing statistics from a similar and better sampled field but using only field-specific data for local estimation. The objective of this specific model validation exercise is twofold: 1. Observe the performance of each model or algorithm under data sparsity 2. Evaluate the worth of the secondary information (EM,) under the same conditions of data sparsity Three approaches times two sampling cases result in six sets of results. Each set of results given hereafter includes An estimated EC, map in the original EC, unit. The corresponding estimated EC, uniform score map, unit free and valued in [O, I]. These uniform scores are the standardized ranks of the previous EC, estimates. Again, this standardization allows a visual comparison that is unit free and free of color or grayscale effect. The semivariograms of the EC, estimated values plotted against the model fitted to the (315) sample semivariograms. That model is the one depicted by the continuous curves in the two top graphs in Fig. 8. This comparison allows for the evaluation of the smoothing effect (Deutsch and Journel, 1992, pp. 17, 61) of the estimation algorithm considered. For the three sets of results corresponding to the second sampling case, the

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 2 S9 Table I Summary of Resultsa Full sample (315)

Reference Simple kriging Simple cokriging P-field

Crossv. sample (289)

m

U2

m

(TI

5.24 5.15 5.21 5.31

11.00 4.54 8.29 13.32

5.30 5.34 (p = 0.21) 5.43 (p = 0.76) 5.57 (p = 0.75)

11.02 0.49 8.45 13.03

a The first two columns give the mean and variance of EC, estimates to be compared to the reference EC, sample used (size 315). The third and fourth columns give the mean and variance of 289 reestimated EC, values and their linear correlation with the 289 actual values. For the latter, the EC, subsample size is 26.

cross-validation scattergram of the 289 "true" EC, values versus the corresponding estimated values Table I provides a summary of the major results.

A. COMMENTS ON RESULTS I. Simple Kriging The results of the base case, simple kriging using only the full sample of hard EC, data (3 15), are shown in Fig. 1 1. The grayscale map of the EC, estimates (top map) reveals a severe smoothing effect: the variance of the estimates is only 4.54 versus the 315 hard data variance of 1 1 .OO. This smoothing effect is a wellknown shortcoming of all linear weighted average-type estimators including kriging (Journel and Huijbregts, 1978, p. 450): typically, the distribution of estimates understates the actual proportions of extreme values, whether high or low values. If detection of spatial patterns of extreme values is the goal of the study, then kriging is not an appropriate mapping algorithm (Journel and Alabert, 1988). Instead, one should consider one of the stochastic imaging algorithms, also known as conditional simulations (Deutsch and Journel, 1992, p. 117), which aim to reproduce the patterns of spatial variability seen from the sample and modeled through the variogram. Conditional simulations are the topic of the latter part of this chapter (see Section VII). The uniform score transform (middle map in Fig. 11) filters the effect of smoothing on the global variance and reveals N3O"E structures clearly associated

2 60

G. BOURGAULT ET AL.

Figure 11 Simple kriging results; full EC, sample (315). The semivariograms (dots) of the estimated values versus the model fitted from the full sample (continuous curve). Note the severe smoothing effect (lesser variance of the estimates).

to the soil type direction of continuity. These structures are somewhat similar to those seen on the rescaled EM, (soft data) map in Fig. lc. The lower graphs in Fig. 1 1 confirm the variance deficiency of the simple

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 261 kriging estimates. Note that the shape of the variogram models is reasonably well reproduced. The smoothing effect of kriging is particularly dramatic when only 26 EC, data are available (see Fig. 12): the variance drops to 0.49 versus 1 1 .O for the original

Figure 12 Simple kriging cross-validation results; EC, subsample (26). The lower right graph gives the scattergram of actual EC, values versus the reestimated values.

2 62

G. BOURGAULT ET AL.

sample data. Although globally unbiased, the distribution of the 289 reestimated values fails badly in reflecting the actual proportions of nonmedian EC, data values outside the interval [4.0, 8.01 (see scattergram at the lower right in Fig. 12). This is known, in geostatistical jargon, as conditional bias (Journel and Huijbregts, 1978, p. 458).

2. Simple Cokriging When using the densely sampled secondary (EM,) information, the smoothing of the EC, estimates is partially corrected to 8.29, a value still less than the original sample variance of 11.OO. The variograms of the cokriging estimates approach those of the model much better. The spatial structures of the cokriging estimates closely reproduce those of the EM, data (compare the uscores maps of Figs. 13 and 10a); this is as expected given the large EM, sample size and the relatively strong correlation (0.73) between colocated EC, and E M , data (see Fig. 4b). The contribution of the dense EM, secondary information is more dramatic when only 26 hard EC, data are available (compare Figs. 12 and 14). The scattergram of true versus reestimated EC, values (lower right graph in Fig. 14) indicates a substantial correction of the smoothing effect and related conditional bias of the simple kriging estimates. In this cross-validation exercise, cokriging (using the secondary information) has raised the true-versus-estimatecorrelation from a low 0.21 to a reasonable 0.76.

3. P-Field Estimates In addition to the secondary E M , data, the p-field approach implemented here accounts for the soil type information. From the results of Fig. 15, it appears that the smoothing effect seen on the simple kriging and cokriging estimates in Figs. 11 and 13 has been overcorrected. The variance of the p-field estimates, 13.30, is now larger than that of the original EC, sample values, 1 1 .OO; the overcorrection takes place in the N120"E direction across soil continuity (see the lower right variograms in Fig. 15). The soil type information appears to have imposed too much of the soil discontinuity along that direction. The cross-validationresults of Fig. 16 confirm the correction of the smoothing effect: the variogram model is well reproduced in both N3O"E and N120"E directions. The correlation true-versus-reestimated values is not significantly improved from the results of cokriging (Fig. 14). Note that the two dots departing most from the 45" line of the scattergram in Fig. 16 are the same as those in Figs, 12 and 14: no estimation algorithm can improve the estimation of outlier values, i.e., values departing significantly from the statistics of the data used. At best,

GEOSTATISTICALANALYSIS OF A SOIL SALINITY DATA SET 263

Figure 13 Simple cokriging results; full EC, sample (315).

the occurrence of such outlier values can be simulated, as is done in stochastic simulation. In summary, from the results in Figs. 11-16, it appears that I . Accounting for a dense secondary information, such as E M , data, does improve the resolution of the estimated EC, map more so if the primary

2 64

G. BOURGAULT ET A L

Figure 14 Simple cokriging cross-validation results; EC, subsample (26).

EC, data set is sparse. However, one should question whether the additional resolution borrowed from the secondary information (EM,) reflects actual patterns of variability of the primary variable (this is more thoroughly discussed in the following sections).

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 265

Figure IS P-field results; full EC, sample (315).

2. Utilization of a second soft information-in this case, soil type-does not bring significant improvement if that information is partially redundant with the first soft information used (in this case, exhaustive EM, data). The p-field approach utilizing soil type data only performs marginally better

2 66

G. BOURGAULT ET AL.

Figure 16 P-field cross-validation results; EC, subsample (26).

than the cokriging approach that accounts only for E M , as soft information (cf. Fig. 16 to Fig. 14). Nevertheless, for the subsample case, the p-field approach provides a better variogram reproduction.

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 267

B. INFLUENCE OF EM, DATA In the two latter estimation algorithms, the sheer density of soft information (one EM, datum at each location being estimated) coupled with the good EC,EM, correlation (0.73) tend to overwhelm the few hard data available. One should then wonder how much of the structures seen on the EC, estimated maps in Figs. 13 and 15 pertain to EC, and how much is mere EM, import. Figure 17 recalls the sample EC,-EM, scattergram as given in Fig. 4b, then gives the four scattergrams of EC, estimates versus colocated EM, values. The correlation EC,-EM, is lowest (0.59) for simple knging estimates, as expected. Accounting for EM, data increases that correlation to a level (0.86,0.90) higher than that of the original sample (0.73). This higher correlation indicates that indeed there may be too much import of the EM, structures into the EC, mapping exercise. Note that the cokriging estimates show a linear relationship when plotted against the secondary EM, data (Fig. 17c). The p-field estimates (Fig. 17d) reproduce better the nonlinear relationship seen in the sample EC,-EM, scattergram (Fig. 17a). At the limit, one may think of forfeiting altogether the EC, data and use the EM, map after proper rescaling to identify the sample EC, histogram (Journel and Xu, 1994). The geostatistical toolbox offers one such algorithm that allows transforming any data set, e.g., the grid of EM, values, with any given histogram H I into another set of values identifying a target histogram H,, with H, possibly quite different from H I . In addition to approximating the target histogram, this algorithm allows reproducing (freezing) a few original data values at their specific locations. This algorithm is a generalization of the well-known normal score transform whose target histogram is the standard Gaussian distribution (Deutsch and Journel, 1992, p. 138). Figure 18a shows the histogram of the 3 15 EC, data (the target histogram). Figure 1% shows the histogram of the 2385 EM, data transformed to match the target histogram: note the excellent histogram reproduction. These transformed EM, data, expressed in EC, units, are taken as estimates of EC, with their map shown in Fig. 18c. Per definition of the transformation algorithm, the uscores of these E C , estimates identify exactly the EM, uscores (Fig. 10a). Figure 18e shows the scattergram of EC, estimates (actually transformed EM, values) versus the original EM, values: this scattergram has a rank correlation 1.OO reflecting the rank-preserving algorithm underlying the transform used. Recalling the scattergrams of Fig. 17, we have, indeed, gone all the way into importing all EM, structures into the EC, mapping exercise. This time, although there is a good linear correlation coefficient of 0.89 between the EC, estimates (EM, transformed) and the EM, data, the nonlinear relationship EC,-EM, is overreproduced (cf. Figs. 18e and 17a). The EC, variogram model shape is not

G. BOURGAULT ET AL.

2 68 25.0

a: Sample

Number 01 data 315

-I

X Variable: m an 9 7 sld. &v. 3% Y Variable: mean 5.24 sld. dev. 3.32

20.0/

EMv (rescaled)

Y Variable: mean 5.15

0.0

5.0

10.0

15.0

20.0

25.0

5.0

0.0

10.0

EMv (rescaled)

15.0

20.0

25.0

EMv (rescaled)

-

20.0:

..

'* Number of data 2385 *d' x varlabte:

i(

m15.01

8 m T 2 BW

Y variable: mean 5.32 std. dav. 3.32

*ank

Z I

.&o.or

%%C 8%

5.0: 0.0'

, ,

,

,

, ,

, ,

, ,

,

,

,,, ,

,

, , ,

,

Agure 17 Scattergrams EC,-EM, (original sample and EC, estimated values).

perfectly reproduced. This indicates that EC, and EM, spatial structures are not self-similar and that perfect identification of the two histograms is not enough to identify variograms (see Fig. 18d).

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 269

Figure 18 Transforming EM, data to identify the E C , sample histogram.

W . FILTERING STRUCTURES Geostatistics provides kriging algorithms for filtering from a spatial distribution any structure present in its variogram model. Consider, for example, the variogram model fitted from the 3 15 EC, data and shown in the top row of Fig. 8.

270

G. BOURGAULT ET AL.

This model includes a large nugget effect, a first isotropic structure with range of 700 m, and a second anisotropic structure with long range of 16,000 m in the N30"E direction of soil continuity and short range of 3000 m in the N120"E direction across continuity. The factorial knging algorithm (Deutsch and Journel, 1992, p. 68) allows filtering out from the simple kriging estimated map at the top of Fig. 11 (reproduced in Fig. 19a) the influence of both nugget effect and shortscale (700 m) structure, leaving the large-scale anisotropic structures (see Fig. 19b). Alternatively, one can filter out the influence of the large-scale variogram component leaving the short-scale structure (see Fig. 19c). The "sum" of Figs. 19b and 19c plus the nugget effect values at sample locations (not shown) add up to the original simple kriging map of Fig. 19a. Because the nugget effect and short-scale structure account for such a large proportion of the E C , spatial variance, the impact of the previous filtering is better seen on the corresponding uscore maps (see Fig. 20). Recall that the uniform score transform standardizes all distributions (hence variances) to a uniform distribution in [0, I]. Figures 19b and 20b depict the clear anisotropy of the large-scale structure associated to the soil type distribution (cf. Fig. la). Conversely, Figs. 19c and 20c zoom on shorter scale patterns of soil salinity possibly related to human activities: note the appearance of 1 X 1-km quadrats delineating different soil usage (Lesch et al., 1995b). Recall that these maps are based on simple kriging-that is, ignoring the E M , information. Instead of being integrated in the simple kriging system (Deutsch and Journel, 1992, p. 68), the factorial kriging algorithm can be used on any already available map such as the cokriging EC, map shown at the top of Fig. 13. Figures 21 and 22 give for simple cokriging the same series of maps as given in Figs. 19 and 20 for simple kriging, the difference being utilization of the secondary E M , information. As noted in the previous section, accounting for the dense E M , information adds considerable local resolution to the estimated EC, maps (cf. Figs. 21a and 19a). After filtering, the EC, cokriging maps (Figs. 21b and 22b) depict the same large-scale structure (related to soil type) seen on the filtered simple kriging maps (Figs. 19b and 20b). However, the short-scales structures seen on the cokriging maps (Figs. 21c and 22c) differ markedly from those seen on the simple kriging maps (Figs. 19c and 20c). To further investigate that difference in short-scale structures, the E M , data rescaled to the EC, variance 11 .O were directly filtered (see Figs. 23 and 24, whose layouts are the same as those in Figs. 21 and 22). It appears that the large-scale structure of the E M , data is indeed similar to that observed on both the simple kriging and the cokriging EC, estimates (cf. Figs. 23b and 24b to Figs. 19b and 20b and to Figs. 21b and 22b). However, the shortscale structures of the E M , data (Figs. 23c and 24c) are clearly different from those of the simple kriging EC, estimates (Figs. 19c and 20c). The short-scale E M , structures reveal a NS-EW lattice possibly linked to the layout of the

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 2 7 1

Figure 19 Filtering applied to the simple kriging map. (a) Simple knging EC, estimates, as in Fig. 1 la, (b) large-scale EC, patterns after filtering, and (c) short-scale EC, patterns after filtering.

electromagnetic measures, I whereas the short-scale EC, structures (Figs. I9c and 20c) are more curvilinear, possibly related to human activity. Figure 25a shows the scattergram of the regional components of EM, and EC, IPosterior to this comment, discussion with Scott Lesch (Lesch er al., 1995a,b) pointed to an interpretation of the mostly EW short-scale lattice sturcute seen on the EM,, map (Fig. 24c). This EW

272

G. BOURGAULT ET AL.

Figure 20 Uscore transform of simple kriging estimated maps. (a) Prior to filtering, (b) largescale EC, patterns after filtering, and (c) short-scale EC, patterns after filtering.

structure may be the result of repeated directional flood irrigation causing, over the years, a certain amount of soil salinity to redistribute along furrows from head to tail in each field. There are not enough EC, hard data within each field to reveal that structure.

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA S E T 273

figure 21 Filtering applied to the cokriging map. (a) Simple cokriging EC, estimates, (b) largescale EC,. patterns after filtering, and (c) short-scale EC, patterns after filtering.

as read from Figs. 19b and 23b, respectively. Figure 25b shows the scattergram of the corresponding short-scale components as read from Figs. 19c and 23c. Note the poor correlation of the short-scale structures (p = 0.30) as opposed to

274

G. BOURGAULT ET AL.

Figure 22 Uscore transform of simple cokriging estimated maps. (a) Prior to filtering, (b) largescale EC, patterns after filtering, and (c) short-scale EC, patterns after filtering.

the better correlation of the regional, large-scale, components (p = 0.63). On the regional scale scattergram, there appears to be two populations most likely related to the two groups of soil types previously distinguished from EM, histo-

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 275

Figure 23 Filtering applied to the EM, data. (a) EM, data rescaled to EC, variance, prior to filtering; (b) large-scale EM, patterns after filtering; and (c) short-scale EM, patterns after filtering.

grams per soil type (see Fig. 6). Were these two groups separated, the regional structures correlation observed on Fig. 25a would be even higher. This analysis would suggest filtering the short-scale structures from the E M , data before using them in the cokriging of EC, values.

276

G. BOURGAULT ET AL.

Figure 24 Uscore transform of EM, data. (a) Prior to filtering, (b) large-scale EM,pattems after filtering, and (c) short-scale EM, patterns after filtering.

V. SPATIAL,CLUSTER ANALYSIS Spatial clustering allows to delineate in space relatively contiguous zones with similar attribute values. When dealing with multiple attribute values, e.g., EM,,

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 277 12.9

3

10.9

Regional

. Local

.. & -2 I

-1

a

.a:

L

.*-

. . .. :-.v .

Number of data 2385

2.9

3.

10.19 2.46 0 ) 1.9 5.15 1.50 3 0.9 correlation 0.63 0.57 -I -0.1

Number of data12385 X Variable: mean 0.00 std. dev. 1.28 Y Variable: mean,o.oo std. 0.78 - dev. ~. mlation 0.30 rrelation0.29

8

L.

.-.

-2.1 - l . .4 l L d.4

-

-11.4

-

-

0!6

216

416

Local EMv

Regional EMv

Rgure 25 Scattergrams of EM,-EC, scale-specific components of spatial variability. (a) Largescale (regional) and (b) short-scale (local).

EM,, and EC, values, definition of such zones from multiple contour maps would be fastidious and highly subjective. Standard clustering techniques (Sneath and of colocated Sokal, 1973) typically consider only the correlation matrix [pkkp(0)] attribute values, thus ignoring the spatial information provided by the full correlation matrix [pll,(h)], V lag distance h between samples. Such standard clustering often results in groups homogeneous in terms of attribute values but dispersed in space; more precisely, any particular group may comprise several disjoint areaslvolumes in space (see Fig. 26b). Bourgault et al. (1992) have proposed to weigh the traditional similarity measure (s,,) between any two samples i,j by a function r(h,) of their separation vector h,,:

s, = s,,

x r(hJ

(6)

where sv = ZT Z-1 * Z, is the traditional similarity measure, Z, is the K vector of standardized attribute values at location u,, E is the K X K correlation matrix of colocated attribute values, and 2-1 is the traditional Mahalanobis distance, e.g., for K = 2 attribute values (EM,, EM,), EMV(u,) - m,

zi =

EMH(u;) - mH UH

1

where m,, a$ is the mean and variance of EM, data, m H ,a:, is the mean and variance of EM, data,

X

=

( piH'TH)

with pVHbeing the linear correla-

tion coefficient between colocated EM, and EM, data, and r(h) = E.kK=]E:== wkk,pkk,(h)is the multivariate covariance defined as a linear combination of the

278

G. BOURGAULT E T AL.

Figure 26 Comparison of standard versus spatial clustering of electromagnetic data in reproducing soil type groups.

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA S E T 279

K X K standardized (cross)covariance functions pkk, (h) between any two RV’s Zk(u) and Zk,(u + h) separated by vector h. h, = u, - ui is the vector separating the two samples i, j at locations ui,uj. Typically, the ( K X K) weight matrix [wkkt]is identified to the Mahalanobis distance: [wkk,]= Z-1. Spatial continuity, anisotropy, and cross-correlation are accounted for through the term r(h,) in the definition [Eq. (611 of the similarity measure S,. Cluster analysis using this measure will result in grouping of samples with similar attribute values (term so) but also spatially close together [term r(h,)]. The spatial cluster analysis algorithm is demonstrated using only the two densely sampled electromagnetic attributes, EM, and EMH.No soil data were used in order to check that cluster analysis using only electromagnetic data does result in groups consistent with soil type differentiation. The algorithm progresses as follows: 1. An initial number of groups is chosen arbitrarily, not too large to allow statistical characterization of each group. Here, seven groups were retained according to the actual number of soil types. 2. All 2385 electromagnetic (EM,and EM,) samples are randomly assigned, with equal probability +,to one of the seven groups. 3. For each sample i, with i = 1;-.,2385,

Calculate its average similarity with group (g) defined as

where I(g)l is the number of samples j currently classified in group (g). Assign sample i to the group with which it has the highest similarity, then update the constitution of all groups. 4. Step 3 is repeated until no change is observed in the constitution of the G groups. 5. The targeted number Go with Go 5 G, of groups is obtained by concatenating groups having similar characteristics; for example, in the case of Fig. 29, Go = 2 corresponding to the two super groups having mean EM, values greater (lesser) than the overall mean EM, = 1.47. Figures 26b and 26c show the Go = 2 super groups resulting from a standard clustering algorithm using similarity measure so and the proposed spatial clustering algorithm using measure s,.Figure 26a shows the reference binary soil type map regrouping the seven original soil types into only two super groups depending on whether the mean EM, values exceed the overall mean 1.47. Utilization

280

G. BOURGAULT ET AL.

of spatial information results in a much cleaner image closer to the reference image obtained from soil type data.

VI. STOCHASTIC IMAGING In the previous sections, various estimated EC, maps have been presented but their accuracy was not assessed. As opposed to mere interpolation algorithms, the main contribution of a geostatistical approach is to provide an assessment of the reliability of any given estimated value. What is the reliability of the EC, estimated value at any specific location u? What is the reliability of any cluster of, e.g., high EC, estimated values as seen on the estimated map of Fig. 13? Can there be alternative estimated maps using the same information? Besides the kriging estimated value, the solution of any kriging system yields a kriging variance-that is, the minimized error variance (Isaaks and Srivastava, 1989, p. 286). Unfortunately, because this kriging variance is data values independent, it is a poor measure of estimation accuracy; instead, it is only a ranking index of data configuration-the data configuration corresponding to a lesser kriging variance would yield on average (over all possible data values for that configuration) a more accurate estimate. Even if the kriging variance u$(u) was a measure of accuracy of the estimated EC, value at location u, the two kriging variances u$(u) and u$(u’) would not provide assessment of joint accuracy at the two locations u and u’. For example, these two kriging variances would not allow assessing the probability that the two unsampled values Z(u), Z(u’) be jointly above a given threshold zo. The concept of stochastic simulation (stochastic imaging) was developed to answer this need for a joint spatial measure of uncertainty (Deutsch and Journel, 1992, p. 17). As opposed to kriging or any other interpolation algorithm, stochastic simulation yields not one but many alternative equiprobable2 images of the distribution in space of the attribute under study (in this case, EC,; see Fig. 29). The difference between these alternative realizations, or stochastic images, provides a visual and numerical measure of uncertainty, whether involving a single location u or many locations jointly. Similar to kriging , there are many stochastic simulation algorithms (Deutsch and Journel, 1992, p. 117) depending on which particular feature (statistics) of the data ought to be reproduced. The first goal of geostatistical simulations is to correct for the smoothing effect observed in any (co)kriging estimated map. *These stochastic images are equiprobable in the sense that, for a given simulation algorithm with its specific computer code and choice of statistics, each image is uniquely indexed by a seed number that starts the algorithm. The seed numbers are drawn from a probability distribution uniform in [0,1]; hence, each image is equal likely to be drawn.

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 281 Hence, the simulated values have a similar spatial continuity (variogram) to that of the sample data set used. In the following section, we present an indicator simulation algorithm modified to account for the soft information provided by electromagnetic (EM,) data. The indicator simulation algorithm (Deutsch and Journel, 1992, p. 146) allows the simulated values to display different spatial continuities (variograms) for different classes of values.

A. SJMULATION ALGORITHM Any unsampled EC, value at location u is interpreted as a random variable Z(u). This random variable (RV) can be seen as a set of possible outcome values or realizations, z(‘)(u), 1 = 1, 2;*.*, characterized by a probability distribution, denoted Prob {Z(u) 5 zl(n)}, where the notation ((n)is read as “conditional to the information set (n).” In the approach adopted here, this probability distribution is modeled by a weighted linear combination of neighboring indicator data i(ua;z), which is set to I if the EC, datum value z(u,) at sample location u, does not exceed threshold z and set to zero otherwise: n

Prob {Z(u) 5 zl(n)} =

A,(z)

i(ua; z)

(8)

a= 1

The weights h,(z) are given by an indicator knging system specific to each threshold value z (Deutsch and Journel, 1992, p. 150). The n indicators retained correspond to the hard EC, sample values found in the neighborhood of location u. Nine threshold values z corresponding to the nine deciles of the EC, sample histogram (sample size is 3 15) were retained to discretize the range of variability of Z. In this case, the indicator simulation algorithm accounts for the spatial continuity specific to each decile of the EC, data values. The model [Eq. (8)] accounts only for the hard data. Introduction of the soft E M , information was done through the “external drift” concept (Deutsch and Journel, 1992, p. 67) whereby the set of n weights AJz) is constrained such as to ensure that the expected value of the estimator [Eq. (S)] identifies a prior probability deduced from the E M , information. More precisely, the constraint is n a= I

where p(u;z) = Prob {Z(u) 5 zlY(u) = y(u)} is the prior probability of EC, value Z(u) given the colocated E M , sample value y(u). The qualifier, “prior,” indicates that this probability value is obtained prior to using the neighboring values.

G. BOURGAULT ET AL.

282

These prior probabilities [Eq. (8)] are read from the scattergram of colocated EC, and E M , values [see Figs. 27b-27d for three such prior distributionscorresponding to three specific conditioning y(u) value (EM,)]. Following the suggestion made at the end of Section IV, only the regional (large-scale) component of the E M , data was retained for the calibration scattergram of Fig. 27a. The map of this EM, regional component was shown in Fig. 23b. Figures 28a and 28b show the greyscale maps of the nine sets of prior probability values p(u;z,), with one probability value per location u, and one map for each of the nine decile threshold values zk, k = l;.., 9. Figs. 28a and 28b actually map the residual value p(u;z,) - p(z,), with p(zk) = kllO being the marginal probability values. Dark areas are areas where the prior probability

d: conditionalpdf ECe I EMvtl5

c: conditional pdf ECe I E M v d 0

0.004: 0.0031

$

1

o.O 0.002:

0.0000.0

ECe

ECe

Figure 27 Scattergram of colocated EC, and E M , (regional component) values. The scattergram is intersected by three vertical lines; EM, = constant providing three EC, conditional probability distributions.

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 283

Figure 28 Maps of centered prior probability values for EC, as obtained from the colocated EM, data values.

p(u;z,) exceeds its marginal or mean value p(z,); conversely, light areas are where that prior probability is lesser than the marginal value. The sequential simulation paradigm calls for visiting along a random sequence all nodes u of the simulation grid. At any such node u, the prior probability values p(u;z,), k = l;.., 9 are updated through the indicator kriging process [Eqs. (8) and (9)] to account for neighboring original values and EC, values

2 84

G. BOURGAULT ET AL.

Figure 28 (continued)

simulated at previously visited nodes (Deutsch and Journel, 1992, p. 123). A simulated EC, value, z(l)(u), is then drawn from that updated probability distribution. A stochastic image, e.g., the lth, is completed when all nodes u of the simulation grid have been visited and filled in with a simulated value. iteration of the entire process starting from another random seed number provides another equiprobable stochastic image. Fifty such stochastic images were generated; the three first realizations are shown in Fig. 29. The pointwise average of the 50 realizations is shown in Fig. 30a. This average, also called E-type map for expected value map, is similar although not identical to the direct simple cokriging and p-field maps respectively shown at the top of Figs. 13 and 15.

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 285

Qure 29 Three stochastic images of the distribution in space of EC, values

286

G. BOURGAULT ET AL.

Figure 30 Pointwise average and standard deviation of the 50 EC, stochastic images. The bottom graph gives their scattergram showing that high E-type estimated values are also the most uncertain.

GEOSTATISTICALANALYSIS OF A SOIL SALINITY DATA SET 287

VII. ASSESSMENT OF SPATIAL UNCERTAZNTY The availability of the 50 equiprobable stochastic images of the distribution in space of EC, values allows derivation of multiple measures of uncertainty beyond a mere visual inspection of these images. Local uncertainty: The uncertainty about EC, at any location u can be assessed by any measure of spread of the 50 simulated EC, values at that location, z(‘)(u), I = 1;**, 50. For example, one could consider the standard deviation of these 50 values. The corresponding grayscale map is shown in Fig. 30b. Note that as opposed to the kriging variance, a variance of simulated values is an estimation variance conditional to the data values retained to simulate these values. At EC, sample locations, that conditional estimation variance is zero (white pixels in Fig. 30b). Elsewhere, that estimation variance depends on the data values and not only on the data configuration; this property is known in statistics as heteroscedasticity. Here, the estimated high E-type values are also the most uncertain in that the corresponding variance between simulated values is larger (see the scattergram in Fig. 30c). Probability maps: At each location u one can count the proportion of simulated values z([)(u) lesser (or greater) than any given threshold value 2, then map these proportions. Figure 31 shows two such probability maps: (i) the probability that EC, is no greater than the first decile z, = 2.18 of the sample EC, histogram. Dark areas (high probability) on this map are areas where EC, is surely low valued; and (ii) the probability that EC, exceeds the ninth decile zg = 10.0 of the sample E C , histogram. Dark areas (high probability) on this map point to areas where EC, is surely high valued. Note that probability maps are unit free, valued in l0,lI. Quantile maps: For some applications it is convenient to merge in a single map an “estimate” of the attribute value and the assessment of the accuracy of that estimate. Quantile maps provide such joint assessment. Figure 32a provides a low (0.1) quantile map; more precisely, the map of the EC, value that is exceeded by 90% of the simulated values at the same location u. Therefore, a location appearing high (dark) in Fig. 32a has a high probability (90%)to be actually higher. Dark areas on a low-quantile map are areas that are surely high valued. Conversely, Fig. 32b shows a high (0.9) quantile map; more precisely, the map of the EC, value that is higher than 90% of the simulated values. Therefore, a location appearing low (light gray) in Fig. 32b has a high probability (90%)to be actually even lower. Light gray spots on a high quantile map point to areas that are surely low valued.

288

G. BOURGAULT ET AL.

Figure 31 Probability maps. (a) EC, does not exceed z, = 2.18, and (b) EC, exceeds zg = 10.0.

Note that quantile maps are in the unit of the attribute itself. Figure 32c shows the median or 0.5 quantile map. The median is the value that has equal probability to be exceeded or not exceeded by the simulated values. It can be used in lieu of the E-type (average) map in Fig. 30a. As opposed to the E-type map, the median map carries a measure of its uncertainty. Probability and quantile maps are numerical and visual aids to decision making. They allow spotting areas that are, e.g.,

I . Critical and surely so; hence, where remediation work can start without delay 2. Potentially critical (e.g., high estimated pollution but with large uncertainty); such areas should be earmarked for additional data 3. Safe and surely so; hence, where no further action is warranted

GEOSTATlSTICAL ANALYSIS OF A SOIL SALINITY DATA SET 289

Figure 32 Three quantile maps of the EC,. spatial distribution.

As more or upgraded information becomes available, the probability distributions of Eq. (8) should be updated again and corresponding stochastic images should be generated leading to updated probability and quantile maps.

G. BOURGAULT ET AL.

2 90 a: Mean

Number of Data 50

Variable

b: Lowest mean, realization #47 25 0 22.5 20 0

17.5 15 0 12 5 10.0

7.5 5.0 25

00

25.0 22.5 20.0

17.5 15 0 12.5 10.0 75 50 25

0.0

Figure 33 Ranking of stochastic images. (a) Histogram of simulated overall mean values, (b) realization with lowest mean, and (c) realization with highest mean.

GEOSTATISTICAL ANALYSIS OF A SOIL SALINITY DATA SET 291

VIII. RANKING OF STOCHASTIC IMAGES The various stochastic images can be ranked according to a criterion relevant to their usage. If the attribute value is a pollutant concentration, one may want to rank the stochastic images according to their global mean concentration. If the concentration z(u) is weighted by a “criticality” factor c(u), with c(u) high in critical zones such as playgrounds and c(u) low in less critical zones such as fenced industrial yards, one may consider the stochastic images of the new variable c(u) X z(u) and rank them according to their global mean. Typically, such operation can be achieved with the help of CIS tools. Figure 33a shows the histogram of the 50 simulated global mean EC, values. Figs. 33b and 33c show the corresponding two realizations with, respectively, the lowest and highest global E C , mean value.

M.CONCLUSIONS The aim of this study is not so much assessment of soil salinity but rather to present a typical geostatistical analysis of a data set representative of the diversity and complexity of data sets handled through GIs. There is much more to geographical (spatial) data analysis than performing elementary operations of overlay, merge, and split and then merely mapping data with somewhat arbitrary, eye-pleasing, spline algorithms. The data talk when their geographic interdependence is revealed; there is an essential third component to any two data values taken at two different locations in space or time-their relation is seen as a function of the separation vector linking these two locations. Pictorial and numerical models of patterns of spaceltime dependence allow us to go far beyond data locations into alternative (stochastic equiprobable) maps that depict the true complexity of the data while always preserving an assessment of uncertainty. Present GIS essentially fail to read between the lines of data. When statistics is used, it is elementary statistics, which ignores data locations and the relation of data with space and/or time. It is suggested that the most robust geostatistical tools, as presented in this study, be made available to soil scientists and users of geographical information systems. There cannot be efficient data utilization without data interpretation and modeling. When data are distributed in space, such interpretation and modeling necessarily call for geostatistics.

292

G. BOURGAULT ET AL.

REFERENCES Bourgault, G., Marcotte, D., and Legendre, P. (1992). The multivariate (co)variogram as a spatial weighting function in classification methods. Marh. Geol. 24, 463-478. Deutsch, C. V., and Journel, A. G. (1992). “GSLIB: Geostatistical Software Library and User’s Guide,” pp. 340. Oxford Univ. Press, London. Dimitrakopoulos, R. (ed.) (1993). “Geostatistics for the Next Century,” pp. 497. Kluwer, Academic, Dordrecht . Isaaks, E. H., and Srivastava, R. M. (1989). “Introduction to Applied Geostatistics,” pp. 561. Oxford Univ. Press, London. Joumel, A. G . (1989). “Fundamentals of Geostatistics in Five Lessons. Short Course in Geology,” Vol. 8, pp. 40. American Geophys. Union Press, Washington DC. Journel, A. G., and Alabert, F. (1988). “Focusing on Spatial Connectivity of Extreme-Valued Attributes: Stochastic Indicator Models of Reservoir Heterogeneities,” SPE paper No. 18324. SOC.of Pet. Eng. Journel, A. G . , and Huijbregts, Ch. J. (1978). “Mining Geostatistics,” pp. 600. Academic Press, San Diego. Joumel, A. G . , and Xu. W. (1994). Posterior identification of histograms conditional to local data. Marh. Geol. 26, 323-360. Lesch, S. M., Strauss, D. J . , and Rhoades, J. D. (1995a). Spatial prediction of soil salinity using electromagnetic induction techniques. I . Statistical prediction models: A Comparison of multiple linear regression and cokriging. Water Resour. Res. 31, 373-386. Lesch, S. M.. Strauss, D. J.. and Rhoades, J. D. (1995b). Spatial prediction of soil salinity using electromagnetic induction techniques. 2. An efficient spatial sampling algorithm suitable for multiple linear regression model identification and estimation. WaterResour. Res. 31, 387-398. Sneath, P. H. A , , and Sokal, R. R. (1973). “Numerical Taxonomy-The Principles and Practice of Numerical Classification,” pp. 573. Freeman, San Francisco. Soares, A. (ed.) (1993). Geostat Troia 1992. In “Proceedings of the 4th Geostatistical Congress,” Vols. 1-2, pp, 1088. Kluwer, Academic, Dordrecht. Srivastava, R. M. (1992). “Reservoir Characterization with Probability Field Simulation,” SPE paper No. 24753. SOC.of Pet. Eng.

FURTHER PROGRESS IN CROP WATERRELATIONS Neil C. Turner CSIRO Division of Plant Industry, Centre for Mediterranean Agricultural Research, Private Bag, P.O., Wembley, Western Australia 6014, Australia

I. Introduction 11. Measurement of Water Deficits 111. “Sensing” Water Deficits A. Abscisic Acid as a Stress Signal B. Abscisic Acid and Growth C. Turgor and Growth IV.Water Deficits and Yield V. Use of Reserves to Maintain Yields under Water Deficits VI. Water Use Efficiency VII. Drought Resistance A. Breeding for Improved Drought Resistance B. Use of Molecular Techniques in Breeding for Improved Drought Resistance VIII. Concluding Remarks References

I. INTRODUCTION Two decades ago Begg and Turner (1976) published their review titled “Crop Water Deficits.” They emphasized the importance of plant water deficits in contrast to the previous review in Advances in Agronomy by Russell (1959), which had emphasized soil water relations. Ten years ago Turner ( 1986a) updated Begg and Turner’s review. At that time, the role of phytohormones in signaling soil drying to the shoot was being elucidated and new methods for identifying genotypic variation in transpiration efficiency had been discovered. Inevitably, these new approaches to crop water relations have generated a flurry of studies and an evaluation of their importance is now warranted. By 1986, considerable 293 .,Igronmny, Valrrine 58 Copyright 0 1997 by Academic Press, Inc. All rights of reproduction in any hmn reservrd. Adrunres

171

2 94

NEIL C . TURNER

progress in identifying drought-resistance characteristics had been made and physiologists, breeders, and agronomists were combining to develop breeding strategies for improved drought resistance. However, the incorporation of these characters in mainstream breeding programs was controversial. Review of progress in breeding for drought resistance using morphophysiological characteristics is worthy of examination. Finally, the past decade has seen considerable progress in the use of molecular techniques in plant breeding. Its impact on improved crop performance under water-limited conditions warrants evaluation. In the decade to 1986, approximately 1500 papers on plant water relations were published annually according to the “Water-in-Plants Bibliography” (Turner, 1986a). The turbulent changes in eastern Europe in the past decade has resulted in the “Water-in-Plants Bibliography” being discontinued and thus the number of papers published on plant water relations in the past decade is not as easily determined. Nevertheless, the number of publications in crop and plant water relations continues to be prodigious, just as droughts continue to impact on global food production and global food shortages. As with the two previous reviews, a comprehensive review of the large number of published papers on crop water relations in the past decdde will not be attempted. Rather, the development of ideas and methodologies that have emerged during the past decade will be reviewed and evaluated.

II. MEASUREMENT OF WATER DEFICITS A state-of-the-art conference on measurement of soil and plant water status was held in 1987 (Hanks and Brown, 1987a,b,c) that highlighted progress and problems with measurement techniques. The conference demonstrated that most of the techniques for measurement of crop water relations reviewed in 1986 (Turner, 1986a) are robust and provide reliable estimates of plant water status. Despite some concerns expressed by Balling and Zimmermann (1990), the pressure chamber continues to provide reliable measurements of leaf water potential and tissue water relations parameters (Turner, 1988). However, the adoption of infrared thermometry to measure canopy temperatures and crop water stress index (Jackson, 1982) has not been as widely adopted as first expected. For reliable results, infrared thermometry relies on complete canopy cover, clear sunny conditions, and a relatively even windspeed (Turner and Nicolas, 1987). Jackson (1987), in a reevaluation of infrared thermometry, showed that canopy temperature can only be reliably used to estimate the crop water stress index and irrigation requirement when simultaneous measurements of net radiation and windspeed are made and the aerodynamic resistance of the canopy is estimated. The requirement for these further measurements in addition to canopy and air

FURTHER PROGRESS IN CROP WATER RELATIONS

295

temperatures, and the uncertainties in the estimation of the aerodynamic resistance of the canopy, particularly those arising from vertical air currents, has resulted in the limited use of infrared thermometry for estimating the development of water deficits. The pressure probe technique for the direct measurement of turgor pressure of individual cells (Zimmermann et a l . , 1969) and its miniaturization (Husken et a/. , 1978) was mentioned in passing in Turner (1986a) because at that time it was considered suitable only for laboratory measurements. However, the technique has proved invaluable in understanding the role of turgor in plant water relations. The first instrument (Zimmermann et a/., 1969) was used only on large cells to measure turgor pressure over relatively long periods as the external environment of the cells was changed (Steudle and Zimmermann, 1974). Miniaturization of the pressure probe (Husken el al., 1978) enabled the measurement of higher plant cells and was designed so that volume changes could be induced in cells penetrated by the probe, thereby allowing evaluation of the elastic properties, the half-time for water exchange, and the hydraulic conductivity of individual cells, in addition to measurements of turgor pressure (Steudle, 1989a). Additionally, cell contents can be removed and their osmotic pressure measured (Tomos et al., 1994). The pressure probe technique has now become standard for the measurement of cell water relations and has been used on more than 20 different species and in cells ranging from 10 to 100 pm in diameter (Steudle, 1993). It has been used in glasshouses and in intact plants (Palta et al., 1987) and complements the tissue water relations of the pressure chamber technique. The pressure probe technique has been used to measure the pressure in the xylem vessels of a number of crop and tree species (Balling and Zimmermann, 1990; Benkert et al., 1991; Zimmermann et al., 1993). Although negative pressures as low as -0.3 MPa have been measured, in all cases the average values were +0.02 to +0.06 MPa, even when the leaf water potentials measured by the pressure chamber technique were -2 to -3 MPa (Zimmermann et al., 1993). Furthermore, the pressure in the xylem vessels did not change with transpiration rate and did not always change by an equivalent amount when the tissue was pressurized in a pressure chamber. These observations have raised questions about the reliability of the pressure chamber technique (Balling and Zimmermann, 1990) and the current framework of our understanding of plant water relations and cohesion theory (Passioura, 1991; Zimmermann et al., 1993). Alternatively, the observations raise questions on the reliability of the pressure probe to measure the pressure in the xylem vessels, particularly in detached plants (Steudle, 1993). In the face of overwhelming evidence (e.g., Baughn and Tanner, 1976), the reliability of the pressure chamber technique for the measurement of leaf water potential is robust, provided suitable precautions as detailed by Turner (1988) are heeded. The pressure probe technique has also been modified to study not just individ-

NEIL C. TURNER

ual plant cells but whole roots and root systems (Steudle, 1989b, 1993). The root pressure probe, as it is called, depends on the fact that excised roots exude xylem sap due to the active accumulation of solutes in the xylem. If the pressure probe is sealed to the root xylem, root pressure can be measured directly. By changing either the external osmoticum or the pressure in the xylem with the pressure probe, the elastic, osmotic, and hydraulic properties of the tissues as a whole can be determined (Steudle, 1993). The root pressure probe has been extensively used to study the water transport and osmotic responses of roots of crop plants (Steudle and Jeschke, 1983; Peterson and Steudle, 1993; Steudle et al., 1987).

111. “SENSING” WATER DEFICITS The conceptual understanding that water moves in response to a difference in water potential between the wet soil and dry atmosphere has led to improved understanding of the factors influencing transpiration. However, the physiological responses of crop plants are not determined by the difference in water potential between the soil and atmosphere or by the total water potential of the leaf. This can be seen in Fig. 1 in which adjacent plots of lupin and wheat were monitored as the soil dried. The photosynthetic rate of lupin was higher than that in wheat when water in the soil was nonlimiting, but as the soil dried the rate of photosynthesis in lupin decreased markedly and at higher water potentials (less leaf stress) than that in the wheat (Turner and Henson, 1989). This could be because the lupin had shallower roots than wheat, but observations in this field study and in others (Hamblin and Hamblin, 1985) showed that this is not the case, and studies in controlled environments with restricted rooting volumes showed that lupin photosynthetic rates decreased at higher water potentials and higher turgor pressures than those of wheat (Turner and Henson, 1989; Henson et al., 1989a). This occurred despite a higher hydraulic conductance in lupin than wheat (Gallardo er a/., 1996). Indeed, stomata of lupin closed and assimilation was reduced before there was any measurable decrease in leaf water potential (Turner and Henson, 1989; Henson et al., 1989a). This strongly suggests that leaf hydraulic relations do not always control leaf photosynthesis or leaf conductance. A definitive study was conducted by Gollan et a/. ( I 986) in which wheat and sunflower plants were grown in containers that could be placed inside a pressure chamber so that the roots were able to be pressurized to maintain leaf turgor as the soil dried. As Fig. 2 illustrates, the stomata1 conductance decreased as the soil water content declined whether or not the leaf water content and leaf turgor were maintained by applying hydrostatic pressure to the roots. Similarly, leaf elongation has been shown to decrease as the soil dries, even when the leaf water

FURTHER PROGRESS IN CROP WATER RELATIONS

297

40,

n

P-n

Rainfed lupin

L

g!

Irrigated lupin Rainfed wheat Irrigated wheat

3 -2.4 -3.21

I 220

I

I

I

I

240

260

280

L

l

I 300

Day of year Figure 1 The change with time in (a) the rate of net photosynthesis and (b) the leaf water potential of irrigated (solid symbols) and rain-fed (open symbols) narrow-leafed lupin (A,A) and crops in the field. Adapted from Turner and Henson (1989). wheat (0,0)

relations were maintained in a fully hydrated state by increasing the pressure applied on the roots (Passioura, 1988; Kuang et al., 1990). These studies suggest that leaf turgor does not always control the stomatal conductance or leaf expansion in crop species, but that soil water conditions can override leaf water relations in determining the physiological activity of the shoot. Clearly, leaf water relations can influence stomatal conductance and leaf growth directly as the midday closure of stomata and diurnal changes in leaf elongation in fully watered plants demonstrates (Kramer, 1988). However, soil water relations can also provide a feed-forward signal to the shoot that conditions are deteriorating and reductions in growth and transpiration are required to maintain hydration. Whether this signal is hydraulic or chemical has been the subject of conjecture, but the current evidence strongly supports the presence of a chemical signal, abscisic acid.

A. ABSCISIC ACIDAS A STRESSSIGNAL For several years abscisic acid (ABA) has been known to induce stomatal closure (Mittelheuser and Van Steveninck, 1969; Raschke, 1975; Mansfield,

298

NEIL C. TURNER

. .

500 -

- 2.0 -

*.

p

2ool

s 0 100

0 0.04

- 1.0

0

m

'0 c

0

0.05

0.07 0.08 Water content (g/g)

0.06

Ea

- 0.5

0.09

3

I

0.10

Figure 2 Relationship between leaf conductance and soil water content of wheat in which the maintained by pressurizing the soil/root system. The hydrostaleaf hydration was (0)or was not (0) tic pressure required to maintain leaf hydration is given (0). Lines fitted by eye. Adapted from Gollan er al. (1986).

1976; Zeevaart and Creelman, 1988). Redistribution of ABA within the leaf can lead to stomatal closure (Dorffling and Tietz, 1985; Hartung and Slovik, 1991), so that total leaf content of ABA does not need to change. However, stomatal closure is frequently associated with an increase in leaf ABA content. In the study illustrated in Fig. 1, for example, the stomata closed in both wheat and lupin as the endogenous ABA concentration in the leaf increased (Turner and Henson, 1989). Studies in controlled environments clearly showed that the stomatal conductance of wheat and lupin leaves was halved as the ABA concentration in the leaf doubled (Henson et af., 1989b). ABA is also produced in roots (Lachno and Baker, 1986; Zeevaart and Boyer, 1984; Cornish and Zeevaart, 1985; Zhang and Davies, 1987; Zhang et al., 1987; Zhang and Davies, 1989; Ban0 et al., 1993; Hartung and Davies, 1993) and has been shown to move in the xylem sap to the leaves in which it induced stomatal closure (Davies et af., 1994). However, not all the ABA in the xylem necessarily is produced in the root. Studies by Wolf et af. (1990) have shown that in wellwatered plants of white lupin, 28% of the ABA in the xylem sap originated in the leaves and was transferred to the roots in the phloem. Decreasing the solute potential around the root induced a 10-fold increase of free ABA in the xylem sap and, of this, 50% of the ABA originated in the leaves and was transferred to the roots in the phloem before being transferred back to the shoot in the xylem sap. An analysis by Slovik et al. (1995) has concluded that ABA synthesis in leaves

FURTHER PROGRESS IN CROP WATER RELATIONS

2 99

and transport to the root, in addition to synthesis and redistribution of ABA in the root, is necessary for ABA to act as the signal for deteriorating soil water conditions. Although ABA is produced in both the root and the shoot, the production in the shoot does not preclude the roots of crop plants in drying soil acting as the primrvy sensors of soil water content and ABA acting as the primary root-toshoot message. These suggestions are supported by the relationship between the concentration of ABA in the xylem sap and predawn leaf water potential (Tardieu et al., 1992b), which is often taken as an indication of the integrated soil water potential (Begg and Turner, 1970). However, some of the evidence is equivocal (Munns and Sharp, 1993). First, it has often been observed that stomatal conductance and photosynthesis are frequently maintained high until two-thirds of the available water in the root zone is depleted (Ritchie, 1974, 1981; Turner et al., 1985; Turner, 1990a). In a study in which the root zone of lupin was divided into three compartments that could be dried and watered independently, the stomatal conductance and rate of photosynthesis did not decrease significantly until more than 60% of the roots were in dry soil (Gallardo et al., 1994). Leaf ABA concentrations were not affected by soil drying unless the entire soil profile was allowed to dry. Thus, ABA produced by roots in drying soil may not reach the xylem of the leaf in sufficient amounts to close stomata because of reduced water uptake from dry soil and dilution by water supplied from roots in soil with adequate water. This is consistent with the observations by Tardieu et al. (1992a) who observed that the concentration of ABA in the xylem sap of field-grown maize was higher in compacted soil than in noncompacted soil and that the concentration of ABA in the xylem sap only increased when the water in the entire soil profile was depleted and not when the surface soil layers dried. A second line of evidence against ABA being the root-to-shoot message is that the ABA concentration in the xylem sap does not always correlate well with leaf conductance. There may be several reasons for this. ABA may not be the only inhibitor of stomatal conductance in the xylem sap. Munns and King (1988) showed that when ABA was removed from xylem sap and the sap fed back into the xylem stream, stomatal closure still occurred. Subsequent work has shown that the xylem sap of wheat and barley contains a large molecular weight inhibitor that may be closely related to ABA, although its exact structure has eluded identification (Munns, 1992; Netting et af., 1992; Munns and Sharp, 1993; Munns et af.,1993). Soil drying also markedly affects the ionic balance and pH of the xylem sap and, although none of the ions analyzed in the xylem sap increased or decreased with soil water content in a way that would indicate a direct role in root-to-shoot communication (Gollan et af., 1992), the sensitivity of guard cells to ABA was affected by the concentration of calcium and nitrate in the xylem sap and by its pH (Schurr er al., 1992). Thus, the sensitivity of stomata to ABA is modulated by the nutritional status of the plants. The sensitivity of the

3 00

NEIL C. TURNER

stomata to ABA is also affected by the water status of the leaf itself (Tardieu and Davies, 1992). At low leaf water potentials, the stomata of Cornmelina cornrnunis in the glasshouse and maize in the field were more sensitive to the ABA concentration in the bathing medium or xylem sap than they were at high water potentials (Tardieu and Davies, 1992). Whether this response is mediated by the concentration of other ions in the xylem sap is not clear, but it is likely. Finally, the sensitivity of stomata to ABA is also genetically determined. Differences among species of lupin (Henson and Turner, 1991) and maize lines (Tuberosa et al., 1994) in stomatal sensitivity to ABA have been recorded and adaxial and abaxial stomata have been shown to differ in their sensitivity to ABA (Henson and Turner, 1991). Thus, although not all the evidence conclusively confirms that ABA is the major hormone sensing soil water depletion and providing the feed-forward rootto-shoot signal of soil conditions, the general consensus is that it is the strongest candidate as the operative phytohormone. Further research is required on its site of action in the guard cell and the variation in sensitivity of cells to its presence.

B. ABSCISICACIDAND GROWTH Leaf growth is more sensitive to water deficits than stomatal conductance in many species (Turner and Begg, 1978). Does ABA play a role in leaf expansion? Application of exogenous ABA certainly inhibits leaf growth (Quarrie and Jones, 1977; Van Volkenburgh and Davies, 1983). However, the relevance of the exogenous application studies has been questioned because the concentration of ABA in the xylem sap is much lower than that needed to induce a slowing of growth (Munns and Sharp, 1993). Furthermore, it has also been shown that expanding leaves contain higher concentrationsof ABA than mature leaves (Zeevaart, 1977; Zeevaart and Boyer, 1984). Nevertheless, leaf growth slows in drying soil even when leaf turgor is maintained by use of split-root experiments or pressurization of the root to maintain leaf water potential (Passioura, 1988; Saab and Sharp, 1989; Kuang er a [ . , 1990). This strongly suggests that root-to-shoot signals are involved in leaf expansion. Indeed, Saab et al. (1990) showed that use of an inhibitor of carotenoid biosynthesis, fluridone, to limit endogenous ABA content in the growing zone of the shoot maintained at a low water potential approximately doubled the rate of shoot growth in maize and this was associated with a halving of the ABA content of the uppermost 15 mm of the mesocotyl. Although ABA inhibited leaf growth, it aided in the maintenance of root growth in maize seedlings in dry soil (Saab et al., 1990). When the ABA content of the root tip maintained at a water potential of - 1.6 MPa was reduced by twothirds by the inhibition of ABA synthesis, root growth was also reduced by nearly 67%. When ABA synthesis was not inhibited, the content of the ABA in

FURTHER PROGRESS IN CROP WATER RELATIONS

301

the root tips was about 10-fold higher than that in unstressed roots at a water potential of -0.03 MPa and the rate of root growth was only reduced by about 33%. Analysis of the growth patterns of the root and mesocotyl suggested that the ABA protected cell expansion in the root and inhibited cell expansion in the leaf (Saab et a / . , 1992). The data also suggest that ABA may play a role in maintaining cell production of the root at low water potentials. The way in which ABA affects cell expansion is not clear from these studies and an explanation requires an understanding of recent advances in cell growth.

C . TURGOR AND GROWTH For cells and, hence, tissues to grow, turgor pressure is required in order to stretch the walls at a rate determined by their yielding properties. Lockhart (1965) and Green et a / . (1971) described this requirement of expansion growth ( E ) in terms of the following parameters: E

=

m(P - Y)

(1)

where m is the cell extensibility or “yielding tendency,” P is the turgor pressure of the cell, and Y is a threshold turgor pressure or yield threshold required for cell expansion. Equation ( 1 ) suggests that water deficits that reduce turgor pressure also reduce growth. Although there is considerable evidence that leaf expansion and root extension are often linearly related to turgor pressure, there are cases in which leaf growth is not correlated with leaf turgor (Meyer and Boyer, 1981; Michelena and Boyer, 1982; Barlow, 1986; Nonami and Boyer, 1989; Pritchard and Tomos, 1993; Spollen et af., 1993; Neumann, 1995). Nonami and Boyer (1990a,b) showed that water deficits reduced the yielding properties [m in Eq. (I)] of the cell wall. By contrast, water deficits were shown to increase the yielding properties of roots (Hsiao and Jing, 1987; Itoh et al., 1987; Kuzrnanoff and Evans, 1981; Neumann, 1995), thereby allowing growth at low water potentials as observed by Saab er al. (1990). Careful spatial analysis of growth and measurement of turgor in individual cells along a root using the pressure probe technique (see Section 11) has shown that the extensibility of the cell wall varied with the position of the cell within the growing zone, i.e., with distance from the root tip (Spollen and Sharp, 1991). This dynamic behavior of cell wall extensibility has been explained in terms of the changes in the binding of hemicelluloses and cellulose microfibrils in the cell wall. Green et a / . (1971) and Fry (1986, 1989) suggested that the cell wall simultaneously tightened and loosened during cell expansion. Hemicelluloses are considered to tether and bind adjacent cellulose microfibrils and bear much of the stress in the cell wall due to turgor pressure. Enzymes cleave the binding hemi-

3 02

NEIL C . TURNER

celluloses, thereby allowing expansion and wall loosening, whereas expansion causes slack tethers to become taut. Passioura and Fry (1992) have modeled this process and suggest that the number of taut tethers increases in shoots, but decreases in roots, in response to low water potentials. The hemicelluloses in the cell wall are predominantly xyloglucan and p-glucan, which have the necessary properties to act as tethers (Fry, 1989). The enzyme xyloglucan endotransglyosylase(XET) can both cut and rejoin xyloglucan polymers (Fry et al., 1992). The spatial distribution of XET activity closely mirrored the growth rate in roots of maize at low water potential in the region that the wallyielding properties increased (Spollen et al., 1993) when the roots were exposed to low water potentials of - 1.6 MPa but not when exposed to mannitol at - 1.O MPa (Pritchard and Tomos, 1993). Both Spollen et al. (1993) and Pritchard and Tomos (1993) provide evidence that the slowing and cessation of growth in roots coincides with an increase in peroxidase activity, an enzyme that is believed to make the wall less yielding (Fry, 1986). It has been known for some years that the phytohormone, indole acetic acid, increases wall extensibility (Cleland, 1986). Other hormones have been shown to have varying effects. The role of ABA is currently unclear (Spollen et al., 1993). There are reports that ABA decreases wall yielding (Van Volkenburgh and Davies, 1983) and inhibits XET activity (Spollen et al., 1993). Whether ABA affects XET activity directly is not known. Because soil drying increases the pH of the xylem sap (Gollan et al., 1992) and acidification is known to increase cell expansion (Taiz, 1984), the influence of soil drying on leaf cell expansion may not be mediated by ABA, but rather by other constituents of the xylem sap that also change with soil drying. Considerable advances have been made in the past decade in the role of photohormones, particularly ABA, in the growth and stomata1behavior of crops. The importance of this to understanding of adaptation to water deficits is now becoming clear. Leaf and root growth are fundamental to crop performance when water supply is limited and ultimately affect yield, the subject of the following section.

IV. WATER DEFICITS AND YIELD Improvement in the performance of a crop under water shortage can arise both from genetic changes to the crop that enable it to perform better in drought-prone environments and from agronomic changes that enable the crop to perform better under these conditions. Performance will usually be measured as yield, either biomass or seed yield, but may also be measured as yield stability in droughtprone environments in which maintenance of food supply or return on borrowed

FURTHER PROGRESS IN CROP WATER RELATIONS

303

money are paramount. This is best illustrated by reference to the work of French and Schultz (1984a,b). These authors showed that wheat grown in water-limited environments in southern Australia had a potential seed yield set by water use of 20kglhalmm for values of water use above 110 mm, the amount of water assumed lost by soil evaporation during the growing season. French and Schultz (1984b) cited experimental data showing that earlier sowing, use of nitrogen fertilizer, control of weeds, and diseases could all improve crop yields to close to the potential values set by water availability. The question then arises as to how much of the yield improvement in the past has arisen from improved agronomy and how much from breeding. Yields of wheat in Australia have risen on average by 13 kgihalyear between 1940 and 1990. This contrasts with the United Kingdom where yields have risen over the same period by an average of 93 kg/ha/year. One of the reasons for the poorer yield increases in Australia has been the opening up of new and more marginal land for agriculture (Turner and Begg, 1981), but limitations imposed by water availability play a much more significant role in rain-fed Australian agriculture than in the United Kingdom where water shortage rarely limits production. An analysis of trends in Australian wheat production has shown that yields have increased from - I to 39 kg/ha/year between 1950 and 1991 and that yield increases have been lower and harder to maintain in the drier zones of the wheatbelt than in the wetter ones (Hamblin and Kyneur, 1993). The impact of breeding on crop performance can be evaluated by growing genotypes released over a particular period under similar conditions and evaluating their performance. This has been done for wheat genotypes released between the 1830s and 1985 in the United Kingdom (Austin et al., 1980, 1989) and for wheat genotypes released between the 1860s and 1982 in Western Australia (Perry and D’Antuono, 1989). Since the turn of the century, the rate of genetic improvement in the United Kingdom with adequate rainfall has been between 30 and 38 kg/ha/year compared to 6 kg/ha/year in the water-limited Mediterranean climate of Western Australia (Fig. 3a). These results are consistent with those of other studies with wheat and barley conducted in Australia, the United States, and the United Kingdom (Table I). For the period from 1935 to 1978, the rate of increase in the United Kingdom was 48 kg/ha/year (Perry and D’Antuono, 1989) suggesting that, despite the lower absolute rates of gain in water-limited environments, in both water-limited and adequately watered environments about half of the yield improvement over the past 50 years can be attributed to genetic improvement and the remaining half to improvement in management. However, it must be recognized that there is some inherent risk in interpreting these data simply in terms of the genetic and agronomic improvements because time of sowing will influence the outcome and because use of current management practices will favor the modem cultivars that have been bred for such conditions. Nevertheless, Austin et a / . (1980) showed that the yield rankings obtained at a

3 04

NEIL C. TURNER

,,

Adequate rainfall (UK)

1 D m

F

/

4OOo-

O

.

2ooo-

-: 0

I

1

I

I

I

.

;/

"30[

.,

I

1860

,

1

1900

1

1

1940

1

1980

Year of introduction

Figure 3 Grain yield (a) and harvest index (b) of wheat genotypes in relation to their year of release in adequately watered (0, 0) and water-limited (0)environments. Data are from Austin et al. (1980) (O), Austin et al. ( 1989) (U),Perry and D'Antuono, (1989),and Siddique et al. (l989a) (0).

Table I

Rates of Yield Increase Due to Genotypic Selection" Crop

Period

Locality

Wheat Wheat Wheat Wheat Wheat Wheat Wheat Wheat Wheat Wheat Wheat Barley Barley

1884-1982 1898- I977 1917- I967 1900- 1950 1911- 1979 1919-1987 1934- 1978 1950-1982 1947- 1977 1908- 1978 1908-1985 1880- 1953 1953-1 980

Australia Australia Australia United States United States United States United States Mexico United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom

Rate (kg/halyear) 6 4 2 5-11

10 16 8-22 59 40 30 38 14

45

Reference Perry and D'Antuono (1989) O'Brien (1982) Russell (1973) Salmon et al. (1953) Deckerd ef al. (1985) Cox et al. (1988) Feyerherm et al. (1984) Waddington et al. (1986) Silvey (1981) Austin et al. (1980) Austin et al. (1989) Riggs et al. (1981) Riggs et al. (1981) ~

a

Adapted from Perry and D'Antuono (1989) and Austin et al., (1989).

FURTHER PROGRESS IN CROP WATER RELATIONS

305

high fertility site were similar to those at a low fertility site when lodging was prevented and diseases were controlled, but the rate of genetic improvement was less at the low fertility site than at the high fertility site because the modem cultivars were able to respond to the increased soil fertility better than the older cultivars. In both Australia, with its water-limited environment, and the United Kingdom, with its adequate water availability, the increase in grain yield with cultivar development over the past century has not been accomplished with an increase in above-ground biomass (Austin et al., 1980; Perry and D’Antuono, 1989; Siddique et al., 1989a). Rather, the increase in yield has been due to an increase in the harvest index, i.e., the ratio of grain to total above-ground biomass (Fig. 3b). The improvement in harvest index of modem wheat cultivars grown in the waterlimited environment appears to be due to a number of factors. These include a shortening of the period from sowing to anthesis and maturity (Perry and D’Antuono, 1989; Siddique et al., 1989a; Loss et al., 19891, an increase in ear development (Siddique et al., 1989a,b), and a pattern of dry matter production and water use that is closer to the pattern of rainfall and water availability (Turner 1993; Loss and Siddique, 1994). This ensures that ear emergence and grain filling in modem cultivars does not occur so late that soil water is depleted (Siddique er al . , 1990) and high temperatures induce premature senescence and pinched grain (Kuroyanagi and Paulsen, 1988). In addition to avoiding severe water deficits, crops growing in terminal stress situations, such as Mediterranean-type climates, need to have the availability to maintain grain filling by utilizing stored preanthesis assimilates (Turner and Nicolas, 1987). This is the subject of Section V.

V. USE OF RESERVES TO MAINTAIN YIELDS UNDER WATER DEFICITS The role of reserves of preanthesis assimilates in maintaining grain filling when water deficits develop during grain filling has long been recognized. Turner and Begg (1981) concluded from a survey of the literature that preanthesis assimilates contributed little to grain yield in cereals when the water supply was abundant, but could contribute up to 30% of the final grain yield when water deficits develop during grain filling. Clearly, the contribution from preanthesis assimilates will vary with the timing and intensity of the deficit. Use of radioactive carbon, ‘4C, suggested that although the proportion of preanthesis assimilates transferred to the grain was higher under water deficits, this was due to a reduction in the proportion of postanthesis assimilates and not an absolute increase in the amount of preanthesis assimilates transferred to the grain (Turner and Begg, 1981).

3 06

NEIL C. TURNER

Subsequent work by Blum et af. (1983a) showed that application of the dessicating agent, magnesium chlorate, which reduced current assimilation in imgated plants, induced a reduction in kernel size that was similar to that induced by drought. The reduction in kernel size was least in those genotypes in which loss of stem dry weight was greatest. This suggests that genotypes that translocate more reserves to the grains are able to maintain stable grain size and that genetic variation for this characteristic exists, at least in wheat (see Section VII). Nicolas and Turner (1993), using potassium iodide as a rapid senescing agent, showed that wheat genotypes varied in their ability to maintain grain size, that the ranking of genotypes was stable from year to year, and that those genotypes that were better able to maintain kernel size were those that had the highest content of sugars in their stems at the time that the rapid senescing agent was applied. There is evidence that stem sugars continue to accumulate in the first 2 or 3 weeks after anthesis provided soil water is adequate and photosynthesis maintained (Nicolas et af., 1985; Pheloung and Siddique, 1991). The data strongly suggests that absolute increases in the transfer of preanthesis and early postanthesis carbohydrates to the grain are possible. Kobata et al. (1992) showed that the final grain size of wheat subjected to water shortage varied with the rate of development of water deficits. When the rate of development of the water deficit was rapid, the grain yield was reduced and grain size was smaller than when the water deficit was imposed more slowly by allowing the soil to dry in a humid atmosphere. With a rapid development of water deficit, grain growth was slowed earlier and photosynthesis decreased earlier so that the proportion, but not the absolute amount, of preanthesis dry matter transferred to the grain was higher compared to the slow development of water deficits. Furthermore, the analysis suggested that late-formed tillers, even when they were fertile and had grain, transferred dry matter to the main stem and first-formed tillers (Kobata et af., 1992). Labeling with the stable isotopes of carbon and nitrogen confirmed this and showed that rapid stress development reduced postanthesis carbon assimilation by 57% but increased the remobilization of preanthesis carbon by 36% compared to slow stress development (Palta et al., 1994). There was an absolute increase in the contribution of preanthesis carbon to grain yield. Total grain nitrogen was not affected by the rate of stress development because at rapid stress development there was a reduction in loss of nitrogen (presumably from ammonia volatilization) from 25 to 6% and an absolute increase in the preanthesis nitrogen remobilized to the grain. The preanthesis contributions of carbon and nitrogen to the grain were 64 and 81%, respectively (Palta et al., 1994), considerably higher than those observed previously. Studies by Palta and Fillery (1995a,b) with nitrogen fertilizer treatments suggest that remobilization of stored preanthesis assimilates also aids in maintenance of grain yield in field situations in which terminal water shortage occurs. Although cereals are determinate, water deficits also influenced carbon alloca-

FURTHER PROGRESS IN CROP WATER RELATIONS

307

tion patterns in the indeterminate legume, narrow-leafed lupin. French and Turner (1991) showed that when lupins in the field were given a mild water deficit, the allocation pattern was altered from further vegetative development of axillary branches in favor of the developing seeds. This resulted in slightly more pods and a higher yield in those lupins from which irrigation was withheld for 23 days and in which the midday leaf water potential fell from -0.7 to - 1. I MPa during early flowering on the main stem than in fully irrigated lupins. The water deficit also almost doubled the rate of dry matter accumulation of the seeds on the main stem. Subsequent work by J. A. Palta (personal communication) has shown that a mild water deficit of - 1.4 MPa imposed for 5 days and an increase in temperature from 20 to 25°C for 5 days induced a change in carbon allocation from branches to the main stem in lupin. The overall effect was a significant increase in yield of lupin. Similar changes of assimilate partitioning in favor of reproductive growth induced by water deficits have been observed in peanut (Ong, 1984), cotton (Turner ct af., 1986b), sunflower (Sobrado and Turner, 1986), and faba bean (Grashoff, 1990). This has led to the conclusion that water deficits can provide benefits to yield in a range of species (Turner, 1990b). Indeed, restricted irrigation is often utilized to improve early boll development in cotton (Hearn, 1975, 1979) and to improve yields of sunflower (Turner, 1990b) and faba bean (Grashoff, 1990). One reason why lupin crops are adapted to deep sandy soils of the warm northern agricultural areas of southwestern Australia (Hamblin, 1987) may be that the coarse textured soils induce mild water deficits during flowering that do not occur on fine textured soils in the cool wet areas of the southern and eastern regions of Australia (Turner, 1990b).

VI. WATER USE EFFICIENCY As pointed out by Turner ( 1 986a), water use efficiency can be observed at several levels and care is needed in defining its use. Here, water use efficiency will refer to the dry matter or grain yield produced per unit evapotranspiration, i.e., water loss by soil evaporation plus crop transpiration. Transpiration efficiency (Fischer, 1979) will be used for the crop assimilation or dry matter production per unit of water lost by transpiration. For a leaf, the instantaneous transpiration efficiency, i.e., the ratio of net photosynthetic rate ( A ) to rate of transpiration ( E ) is given by

308

NEIL C. TURNER

where pa and pi are the partial pressures of carbon dioxide in the atmosphere surrounding the leaf and in the substomatal cavity, respectively, and e, and ei are the vapor pressures of water surrounding and inside the leaf, respectively. Integrated over the life of the plant, the transpiration efficiency (TE) is

where & represents the losses of carbon dioxide by leaves, stems, and roots by respiration and 9, represents water lost through partially closed stomata at night (Turner et al., 1978) and cuticular losses by day (Hubick and Farquhar, 1989). Transpiration efficiency is thus affected by both plant factors and environmental factors. The key environmental factor influencing transpiration efficiency is the vapor pressure deficit or relative humidity of the air (Fischer and Turner, 1978; Richards, 1991; Condon er al., 1992). The more humid the atmosphere, the greater the transpiration efficiency, all other things being equal. The key plant factor influencing transpiration efficiency is the ratio of the partial pressure of CO, in the leaf and that outside the leaf. Because the external partial pressure is stable in the short term, decreasing the internal partial pressure of carbon dioxide will increase water use efficiency. Thus, increasing the photosynthetic efficiency of the plant or closing the stomata will decrease the partial pressure of CO, in the substomatal cavity. C, plants typically have higher rates of assimilation and lower stomatal conductances than C, plants and have a ratio of internal to external partial pressures of carbon dioxide of about 0.3 compared to values in C, plants of about 0.7 (Wong et al., 1979). For C, species, variation in the ratio of internal to external partial pressure of carbon dioxide is reflected in variation in the carbon isotope composition of dry matter. The carbon dioxide in the atmosphere typically comprises 98.9% W O , and 1.1% 13C02.Plants take up the lighter l2CO, at a faster rate than the heavier 13C0, (O'Leary, 1981), in part due to the slower diffusion of ' T O , through the stomatal pore and also because of the discrimination against the 'TO, by the primary carboxylation enzyme, ribulose biphosphate carboxylase/oxygenase (Rubisco), which utilizes l2C0, 1.030 times faster than it utilizes WO,. Thus, the discrimination would be 30 X or 30°0 (thirty thousandths) if the discrimination by the enzyme were fully expressed. Because carbon isotope discrimination in C, plants is determined by the internal to ambient carbon dioxide partial pressures, as is transpiration efficiency [Eq.(3)], transpiration efficiency is correlated with carbon isotope discrimination (Farquhar er al., 1982). Farquhar and Richards (1984) predicted that the discrimination (A) of WO, relative to WO, should be

FURTHER PROGRESS IN CROP WATER RELATIONS

A

=

(4.4

+ 22.6 e) X Pa

3 09 (4)

and by substitution into Eq. (3), TE is

i.e., transpiration efficiency is negatively related to A. Studies with a wide range of species have confirmed that transpiration efficiency is negatively correlated with carbon isotope discrimination, as shown in Table 11. Furthermore, the carbon isotope discrimination technique has been used to clearly demonstrate that transpiration efficiency varies with genotype (see Section VII). The variation in transpiration efficiency arises from either differences in photosynthetic efficiency or stomatal conductance, or both. Studies in common bean (Phuseolus vulgaris L.) showed that the variation in transpiration efficiency was largely due to differences in stomatal conductance (Ehleringer, 1990; Ehleringer et al., 1990), whereas in peanut the genotypic variation in transpiration efficiency was largely due to differences in photosynthetic efficiency at the level of the enzyme (Hubick er al., 1986). Variation in the transpiration efficiency of wheat appears to arise from both variation in stomatal conductance and photosynthetic efficiency (Condon et al., 1990). Improvements in transpiration efficiency do not always translate into improvements in water use efficiency (Turner, 1993). Under water-limited conditions, an improvement in transpiration efficiency would be expected to give an increased water use efficiency and grain yield. However, studies by Condon and Richards ( 1993) showed that dry matter production was negatively correlated with carbon isotope discrimination in dry sites and seasons, positively correlated with carbon isotope discrimination at wet sites and seasons, and not correlated at all in intermediate sites and seasons. In part this can be explained on the basis that in wheat the carbon isotope discrimination is influenced by both stomatal conductance and photosynthetic efficiency so that in wet years the lower stomatal conductance limits leaf growth and biomass production, whereas it is only in dry seasons that the improvements in transpiration efficiency are exhibited. If differences in carbon isotope discrimination largely arise from differences in stomatal conductance rather than photosynthetic capacity, differences in transpiration efficiency may not be maintained in the field (Condon and Richards, 1993). A smaller stomatal conductance will lead to a wanner canopy and greater vapor pressure deficits, thereby increasing transpiration and reducing transpiration efficiency. Comparisons of wheat lines with high and low carbon isotope discrimination values in field canopies have shown that indeed a wheat genotype with the high transpiration efficiency had higher canopy temperatures and a higher vapor pressure deficit compared to a genotype with a low transpiration

Table I1 Relationship between Carbon Isotope Discrimination (A) and Transpiration Eficiency in a Range of Speciesa Crop

Location

Relationship

Range in A

( O h )

Reference

Wheat

Glasshouse

Negative

19.1-23.2 (4)

Barley

Glasshouse Glasshouse Glasshouse

Negative Negative Negative

19.1-20.9 (14) 19.6-22.8 (16) 19.8-22.4 (12)

Rice

Glasshouse Field

Negative Negative

16.6-19.0 (10) 19.7-21.5 (9)

Sunflower Peanut

Glasshouse Glasshouse

Negative Negative

17.8-20.8 (6) 16.6- 19.4 (9)

Field Pots-outdoors or in glasshouse Pots-under rainout shelter Pots in field

Negative Negative

18.1-20.9 (8) 20.0-23.2 (16)

Negative

18.4-21.6 (6)

Vos and Groenwold (1989)

Negative

24.5-21.9 (3)

Altai wild rye Orchard grass

Glasshouse Glasshouse Glasshouse Glasshouse

Negative Negative Negative Negative

19.5-24.9 21.0-24.4 20.0-24.5 14.6-15.6

Tall fescue

Glasshouse

Negative

14.1-15.7 (11)

Perennial ryegrass

Glasshouse

Negative

14.2-14.6 (3)

Cotton

Glasshouse

Negative

16.1-19.4 (6)

Coffee Cowpea

Field Pots in field

Negative Negative

16.3-20.3 (5) 19.6-21.5 (5)

Pots in field

Negative

16.6-21.6 (3)

Pots in field

Negative

18.4-20.8 (8)

Martin and Thorstenson ( I 988) Johnson et al. (1990) Read ef al. (1991) Johnson et al. (1990) Johnson and Bassett (1991) Johnson and Bassett (1991) Johnson and Bassett (1991) Hubick and Farquhar ( 1987) Meinzer et al. (1990) Ismail and Hall ( 1992) Ismail and Hall ( 1992) Ismail and Hall (1993)

Common bean

Potato

Tomato Crested wheatgrass

(28) (16) (22) (7)

Farquhar and Richards (1984) Condon et al. (1990) Ehdaie et af. (1991) Hubick and Farquhar (1989) Acevedo (1993) Dingkuhn er al. (1991) Virgona et al. (1990) Huhick er el. (1986, 1988) Wright et al. (1988) Ehleringer er al. (1991)

a Values are given to indicate maximum range of A observed in the study. The number of genotypes in the study is indicated in parentheses. Adapted from Turner (1993).

FURTHER PROGRESS IN CROP WATER RELATIONS

311

efficiency (F. X. Dunin, personal communication). Nevertheless, the genotype with high transpiration efficiency withdrew water from the profile more slowly than the line with low transpiration efficiency (Condon and Richards, 1993; Passioura et al., 1993). The greater warming of the canopy, due to the lower stomatal conductances, reduced the transpiration efficiency in the high transpiration efficiency line by about one-third but did not eliminate differences in transpiration efficiency (Condon and Richards, 1993). Although improvements in transpiration efficiency arising from an increase in photosynthetic capacity are not subject to the same feedback as reduced stornatal conductance, there is some evidence that genotypes with high photosynthetic efficiency have smaller leaves. Bhagsari and Brown (1986) showed that high photosynthetic efficiency is linked to small leaf size and high specific leaf weight as a means of concentrating the available nitrogen. Thus, the association between high photosynthetic efficiency and small leaves may be inherent, leading to slow early growth, low radiation interception, and greater water loss by soil evaporation relative to plant transpiration. Whether high or low transpiration efficiency is required will depend on circumstances. It is clear from the previous discussion and from studies of natural ecosystems (Ehleringer, 1993) that increased transpiration efficiency is useful as a survival mechanism in water-limited environments. However, due to the inherent trade-off between transpiration efficiency and potential productivity, high transpiration efficiency will not be beneficial in all environments (Jones, 1993). Although it is useful in conserving water in the preanthesis period for use in the postanthesis period (Condon et al., 1992), this will benefit crops grown on stored soil moisture but may reduce biomass at anthesis to such an extent that it reduces yields in Mediterranean environments (Turner et al., 1989). Indeed, in Mediterranean-type climates in which temperate crops are grown during the cool wet winters and fill grain in spring, low transpiration efficiency and rapid growth during the period with low vapor pressure deficits in winter may be preferable. As pointed out by Fischer and Turner (1978) and Turner (1986a), there are several management options for improved crop water use efficiency. The studies by French and Schultz (1984a,b) referred to under Section IV showed that over a 10-year period in water-limited field environments wheat crops rarely reached the potential set by water use or rainfall (Fig. 4). They showed that the potential biomass and grain yield should be 55 and 20 kg/ha for every millimeter of water use above 110 mm lost by soil evaporation. Figure 4 also summarizes the relationship between grain yield of wheat and water use for crops grown in Western Australia (Perry, 1987; Siddique et al., 1990; Gregory er al., 1992) as well as South Australia. On the sandy-surfaced soils that prevail in Western Australia (Turner, 1992; Tennant et al., 1992), the water lost by soil evaporation is likely to be lower than that on the heavier textured soils of South Australia (French and Schultz, 1984a; Gregory et al., 1992). Thus, in Fig. 4 the dashed

3 12

NEIL C. TURNER

. .. .. . ... * .

~5.2000 C

*

1000 n U

100

200

300

400

500

Water use (mm)

Agure 4 Relationships between grain yield and water use for wheat grown in South Australia

(0)and Western Australia (0). Data from the study of historical (0) and modern (m) wheats by

Perry and D’Antuono (1989) are also shown. The dashed and solid lines give the transpiration efficiency of 20 kg/ha/mm with a soil evaporation of 80 mm (dashed line) and 110 mm (solid line). Data are from French and Schultz (1984a), Perry (1987), Siddique et al. (1990), and Gregory et af. (1992).

line gives the transpiration efficiency when soil evaporation during the season is 80 mm. The figure also shows the gains made in water use efficiency by breeding for improved yields over the past 120 years (see Section IV). Although breeding over the past 120 years has doubled the transpiration efficiency of wheat under optimum management from about 10 to 20 kg/ha/mm by improvements in harvest index (Fig. 3) and better matching of water use to water supply (Siddique et al., 1990), in many cases the yields per unit of water use are still very low and significant advances in efficiency can be achieved by improved management. Much of the improvement in crop yield achieved in southern Australia in the past decade has been as a result of farmers managing their crops to achieve their potential water use efficiency (Edwards, 1992). Practices that aid in the improvement of water use efficiency that have been adopted by farmers in the Mediterranean-climatic zone of Australia include earlier planting, use of increased levels of fertilizer (especially nitrogen and phosphorus), stubble retention, minimum tillage, and use of rotations to improve the nutrition and root penetration of cereal crops, Extensive studies in Syria have shown that nitrogen and phosphorus increased the water use efficiency of barley by at least 50% over several sites and seasons (Cooper et al., 1987) and studies in Australia have shown that application of 30 kg N/ha increased the water use efficiency from 8 to 11 kg of grain/ha/mm in a year with a growing season rainfall of 165 mm (Turner et al., 1987). One of the major reasons for the increased water use efficiency was the greater proportion of water used by the crop in transpiration rather than being lost in soil evaporation.

FURTHER PROGRESS IN CROP WATER RELATIONS

313

French and Schultz (1984a,b) also showed that earlier sowing improved yield per millimeter of growing-season rainfall. Anderson and Smith (1990a) showed that earlier planting did not affect water use, but did increase yield of wheat to that near the optimum set by water availability. Indeed, the combination of selecting a cultivar suited to the length of growing season, earlier sowing with minimum tillage, use of higher levels of nitrogen fertilization, and better weed control all aid in the utilization of growing-season rainfall and improved water use efficiency (Anderson and Smith, 1990b). However, although it is possible to optimize water use efficiency when the growing season rainfall is less than 400 mm per year, it is much more difficult to do this when runoff, throughflow, deep drainage, and/or waterlogging occur as a result of higher rainfall and uneven rainfall distribution (Gregory et al., 1992). Passioura (1977) and Fischer (1979, 1981) have argued that too great a water use prior to anthesis in cereals can reduce yields due to insufficient water being available for postanthesis water use. Although data collected by Shepherd er al. (1987) over several sites and seasons in west Asia showed a general increase in barley yield with increase in water use after anthesis (Fig. 3,the increased yields and water use always occurred at wetter sites and in wetter seasons. As Fig. 5 shows, at any one site or in any one season grain yield could be approximately doubled by fertilization of the crop without the use of significant additional amounts of water in the preanthesis and postanthesis periods. Water use efficiency was significantly improved on both a dry matter basis and a grain yield basis whether taken on a total seasonal water use or a postanthesis water use basis (Shepherd et al., 1987). The data in Fig. 5 suggest that considerable scope exists

4ooo

U

.

t

I

I

20

40

0

60

I

I

I

80

100

120

140

Water use after anthesis (mm) Figure 5 Relationship between grain yield and water use between anthesis and maturity for and fertilized ( 0 )barley. Data are from Shepherd e? a / . (1987). unfertilized (0)

3 14

NEIL C . TURNER

in Mediterranean climatic regions to increase yields by increasing growth prior to anthesis by management or breeding without any detrimental effects on yield.

VII. DROUGHT RESISTANCE The ability of a crop to produce satisfactorily in areas subjected to water deficits has been termed its drought resistance. One measure of drought resistance of a crop is the comparison of yield with and without water shortage (Fischer, 1981). Although this is a relatively robust measure for determinate plants that complete their growth cycle and mature even when adequately watered, it is less meaningful for indeterminate plants in which irrigation can result in continued production of new flowers and seeds. The mechanisms of drought resistance have been categorized in various ways. 'Ibrner (1979, 1982, 1986a,b) showed that they could be classified into three major categories: Drought escape Dehydration postponement Dehydration tolerance The characteristics adopted by plants in nature and sought by the plant breeder depend on the drought environment. Turner (1982) identified the characteristics or traits required for different water-limited environments. This was further developed by Ludlow and Muchow (1990) for both sorghum and cowpea grown in both modem (Table 111) and subsistence agriculture. Of concern to Ludlow and Muchow (1990) was the observation that although many characteristics or traits had been identified as putatively useful for yield improvement in water-limited environments, very few traits had actually been tested for their influence on yield or survival. Because a characteristic of drought resistance may be useful in one environment but not in another environment, Passioura et al. (1993) have suggested abandoning the concept of drought resistance. However, breeders still consider the framework useful (Arraudeau, 1989; McWilliam, 1989). An alternative framework for considering the characteristics that are important in yield improvement under water-limited conditions was proposed by Passioura (1977). He suggested that it was appropriate to consider the grain yield (GY) under water-limited conditions as a function of three variables: GY = W

X

WUE

X

HI

(6)

where W is water used, WUE is water use efficiency, and HI is the harvest index. Such a framework is a truism because WUE by definition is the biomass production per unit of water use and HI is the ratio of grain yield to biomass production. For the framework to be most useful to the breeder, each component in the

FURTHER PROGRESS IN CROP WATER RELATIONS

315

Table 111 Traits for Improved Drought Resistance in lntermittent and Terminal Stress Environments“ Intermittent stress

Terminal stress

Matching phenology to water supply High degree of osmotic adjustment of shoots and roots Increased rooting depth and density Developmental plasticity Early vigor Leaf area maintenance lncreased leaf reflectance Low lethal water status ~~

a

Matching phenology to water supply High degree of osmotic adjustment of shoots and roots Increased rooting depth and density Increased leaf reflectance Early vigor Good mobilization of preanthesis dry matter

~

Adapted from Turner (1982) and Muchow and Ludlow (1990).

equation must be independent of the others and improvement in any one component should result in an improvement in grain yield without any feedback. Clearly, this is unlikely for an identity such as grain yield, which is subject to a series of controls, buffers, and feedbacks (Richards, 1989). Nevertheless, it has provided a useful framework for considering agronomic and genetic improvements to yield under water-limited conditions (Fischer and Turner, 1978; Turner and Nicolas, 1987; Richards 1989; Passioura et al., 1993).

A. BREEDINGFOR IMPROVED DROUGHT RESISTANCE Turner ( 1986a) predicted that “the next decade will determine whether these [morphophysiological] characters successfully find their way into released varieties and will clarify whether incorporation of individual traits or multiple, but less proven, characters into the gene pool will more rapidly improve yields under water-limited conditions.” Certainly, there has been progress in identifying characters useful in breeding for improved yield or yield maintenance under drought and some of these are now part of mainstream breeding programs. Some of the key characteristics that have been studied and progressed in the past decade follow. 1. Phenology

Matching the phenology to the water supply is the primary way in which yields have been improved under water-limited environments. Genetic variation in growth duration is usually large in crop plants and can be readily selected by

3 16

NEIL C. TURNER

observing the days to flowering or maturity. In environments in which terminal stress is likely, selection for a shorter time to flowering has been highly successful in improving the drought resistance of a crop. The improvement in yields with more modem wheat genotypes observed by Perry and D’Antuono (1989) was shown to be largely, but not solely, due to earlier flowering and matching the water use pattern to the seasonal rainfall (Siddique et al., 1990; Turner, 1993). However, selection for an earlier flowering date may not always increase yields if this makes the genotypes flower during a period of high frost ‘incidence or if the biomass at anthesis is very low (Fischer, 1979, 1981). Additionqly, if the crop is determinate, selection for earliness may result in the genotype being unable to respond to longer and more favorable seasons. Furthermore, in‘ tnvironments with a bimodal rainfall distribution or more erratic rainfall pattern, sbort-duration genotypes may not yield as well as longer season genotypes that can take advantage of the late rains. A specific example in which modification of the phenology has resulted in improved yield under drought is that provided by tropical maize. One of the major advances in improving yields of tropical maize grown in water-limited environments has been the shortening of the interval between anthesis and silking (Edmeades et al., 1989). Selection through eight breeding cycles for improved yield, greater stem and leaf elongation, reduced anthesis-to-silking interval, and canopy temperature was applied under drought conditions to a tropical maize population (Bolaiios and Edmeades, 1993a). Selection resulted in yields being improved by 0.85 t/ha at the selection site and 0.65 t/ha across all testing locations in the tropics with mean yield levels ranging from 1 to 8 t/ha. The major reason for the improved yield under drought was a reduction in the anthesis-to-silking interval from 35 to 10 days, increased dry matter partitioning to the ear, and reduced tassel weight (Bolaiios and Edmeades, 1993b). The improved yield was also associated with reduced rooting in the upper 50 cm of the profile (Bolafios ef al., 1993). The shorter anthesis-to-silking interval improved yields under drought through an improved harvest index.

2. Early Vigor Studies in a range of localities have shown that soil evaporation can be a major source of water loss in water-limited environments. Studies of water use in a Mediterranean environment suggest that soil evaporation can account for between one-fifth and two-thirds of growing season rainfall or water use (French and Schultz, 1984a,b; Perry, 1987; Shepherd et al., 1987; Cooper et al., 1987; Siddique et al., 1990; Greenwood et al., 1992; Gregory and Eastham, 1996). When the soil surface is wet, evapotranspiration is the same whether a crop is present or not and lupin crops prior to flowering were shown to have transpiration rates only one-third of the total evapotranspiration (Greenwood et al., 1992). Turner and Nicolas (1987) showed that grain yields of wheat on a deep coarse-

FURTHER PROGRESS IN CROP WATER RELATIONS

317

textured soil were linearly related to early vigor, measured as dry matter production at the five- or six-leaf stage, in a Mediterranean climatic region of Australia. Subsequently, Whan et al. (1991a) confirmed in a wide range of genotypes at several locations that early growth had the potential to increase grain yields, and Siddique et al. (1989a) showed that, compared to wheat, barley had a faster emergence rate and a more rapid development of leaf area that was related to its higher grain yield. Likewise, Van Oosterom and Acevedo (1992) found a strong correlation between early vigor and grain yield in barley. In both barley and wheat, the early vigor was associated with greater leaf area and ground cover (Regan et al., 1992; Van Oosterom and Acevedo, 1992) and probably a greater proportion of total evapotranspiration used in crop transpiration rather than soil evaporation (Shepherd er al., 1987; Cooper et al., 1987). However, not all the gain is likely to derive from improved water use by the crop because early vigor also increased biomass at anthesis (Regan et al., 1992; Whan et al., 1991a,b, 1993; Palta and Fillery, 1995a) and studies in a Mediterranean environment showed that grain yields were correlated with dry matter at anthesis up to at least a biomass yield at anthesis of 12 t/ha (Turner and Nicolas, 1987). The correlation between ground cover and dry matter production has been used to rank lines of cereals differing in early vigor using field spectroscopy (Elliot and Regan, 1993; Smith et al., 1993). Increased biomass or ground cover resulted in increased reflectance in the visible range and decreased reflectance in the near infrared and mid-infrared parts of the spectrum. The methodology used was able to distinguish biomass differences in the order of 4 g/m2, but may be confounded by soil type, crop type, and growth stage if comparisons are made across soil types and at different stages of development (Bellairs et al., 1996). Lines of wheat with more rapid rates of early growth have been identified (Whan et al., 1991a,b, 1993; Richards, 1991), and although these genotypes did not always give higher grain yields (Regan et al., 1992), they were not adapted to local conditions. Whan ef al. (1993) measured the biomass production and grain yield of the progeny of these genotypes when crossed with standard locally adapted genotypes and showed that high biomass was associated with high grain yields at dry short-season sites, but no relationship was evident at cool wet sites. Broad sense heritability for biomass production varied from 60 to 80% depending on the cross (Loss and Siddique, 1994). Early vigor may result in greater water use prior to anthesis and this may be detrimental to yield if it depletes the water in the profile by anthesis. Although this might be the case in which crops are grown on stored soil moisture, data presented in Fig. 5 (see Section V1) suggests that in Mediterranean-type climates considerable scope exists for increasing early vigor without decreasing yield. Moreover, the observation that increased biomass at anthesis up to 12 t/ha was positively correlated with yield on deep sandy soils (Turner and Nicolas, 1987) is inconsistent with yields being limited by vigorous early growth and suggests that vigorous shoot growth may be correlated with vigorous root growth and greater

318

NEIL C. TURNER

water use. Thus, early vigor appears to increase water use at least on deep sandy soils by deeper rooting, increases water use efficiency by increasing the proportion of water lost by transpiration compared to that lost by soil evaporation, and indirectly appears to increase the harvest index by increasing ear weight at anthesis (M. E. Nicolas and N. C. Turner, unpublished results).

3. Osmotic Adjustment Osmotic adjustment is the accumulation of solutes that occurs as a result of exposure to water shortage (Turner and Jones, 1980; Morgan, 1984). The dzgree of osmotic adjustment varies with species (Turner and Jones, 1980) and genotypic variation within a species has been demonstrated in wheat (Morgan, 1983), sorghum (Ludlow et al., 1990; Santamaria et al., 1990), millet (Henson, 1982), rice (Turner et al., 1986a), chick-pea (Morgan et al., 1991), and pigeon pea (Flower and Ludlow, 1987). Data for wheat suggest that only one or a few genes are involved and the characteristic is simply inherited (Morgan, 1983; Morgan et al., 1986). Osmotic adjustment has been shown to enable photosynthesis to be maintained at low leaf water potentials (Jones and Rawson, 1979; Turner et al., 1996), to delay leaf senescence (Hsiao et al., 1984), improve floret survival, and to improve yield under water-limited conditions. Sorghum genotypes with high osmotic adjustment had 34 and 24% higher yields when water stress occurred before and after anthesis, respectively, compared to genotypes with low osmotic adjustment (Ludlow and Muchow, 1990). Likewise, wheat genotypes with high osmotic adjustment had 1-60% higher yields than those with low osmotic adjustment when water supply was limited (Morgan, 1983; Morgan et al., 1986). Osmotic adjustment does not prevent photosynthesis decreasing (Jones and Rawson 1979; Turner et al., 1996) but helps to maintain low rates of photosynthesis at low water potentials. Although this may be perceived as a survival mechanism, it does enable physiological activity to be maintained, albeit at a low level, throughout a period of water deficit and may enable carbon and nitrogen reallocation to the grain during a terminal stress (Kobata et al., 1992; Palta et al., 1994). The major way in which osmotic adjustment affects yield is through increased root growth and water extraction from the soil (Morgan and Condon, 1986). Thus, osmotic adjustment improves water use by the crop and, because it also delays senescence and maintains assimilate transfer to the grain, it increases harvest index.

4. Carbon Partitioning The work by Blum et al. (1983a,b) showing that a rapid desiccating agent applied to irrigated wheat 10-14 days after anthesis (see Section V) provided a

FURTHER PROGRESS IN CROP WATER RELATIONS

3 19

means of simulating postanthesis water deficits and of distinguishing between lines with differing abilities to transfer assimilates to the grain. Using potassium iodide, a rapid senescing agent that better simulates the postanthesis water deficits common in water-limited environments, studies over 2 years confirmed that there was a good correlation between the reduction in grain size due to potassium iodide treatment and that due to water deficits (Nicolas and Turner, 1993). The lines with high assimilate transfer to the grain were ones with high soluble sugar levels in the stem at the time of treatment (Nicolas and Turner, 1993). The technique is now routinely used in the Israeli wheat breeding program to select lines under irrigation that have high assimilate transfer to the grain under dryland conditions (Blum et af., 1991; Blum, 1993) and is used under glasshouse conditions in the United States to select parents and advanced lines to maintain high assimilate transfer when high-temperature stresses increase leaf senescence (Hossain el af., 1990). High assimilate transfer is a trait that affects only the harvest index. Studies have been conducted to evaluate the use of the technique in a conventional dryland breeding program in the Mediterranean-type environment of southwestern Australia. Ninety-six lines of wheat in two separate groups and at two locations were treated with the chemical senescing agent at 7- 14 days after anthesis, and the influence on grain size and on grain yield was evaluated. These studies differed from previous studies in that no irrigation was applied because the use of irrigation is not feasible for routine screening in a rainfed breeding program. The reduction due to treatment vaned from year to year and from location to location (Regan et al., 1993). However, differences between lines in response to the chemical desiccation were only evident in those environments that gave large reductions in grain yield or grain weight. Differences could not be detected when the overall reductions were small. Regan et al. (1993) concluded that the use of chemical senescence to identify breeding lines with a high transfer of assimilates to the grain will be limited in the absence of irrigation because the expression of differences will depend on the environment, but that it continues to be a useful technique where selection under irrigated conditions is practical.

5. Grain Growth Kuroyanagi and Paulsen ( 1988) showed that, even with adequate water supply, wheat subjected to air temperatures above 35°C rapidly senesces, leading to pinched grain. Genotypes with an ability to fill their grain quickly should have an advantage in water-limited environments with short hot and dry grain-filling periods (Bruckner and Frohberg, 1987). Studies by Loss et af. (1989), Whan et a/. (1991b), and Duguid and BriilC-Babel (1994) have shown that'wheat genotypes differ in their rate of grain growth and that selection for faster rates of grain growth than those in the current genotypes is feasible. Genotypes with high

320

NEIL C . TURNER

rates of grain growth achieved this by having a short duration of growth and either a high grain weight or a high number of grains per unit area. Rapid grain growth is associated with a reduced number of grains so that the two characteristics need to be considered together (Whan et al., 1993). Selections from crosses between introduced genotypes with rapid grain growth and locally adapted genotypes have been identified that maintain the high grain number of the locally adapted genotypes but also have faster grain growth derived from parents with high grain growth (Whan et al., 1993). The impact of these selections on yield has not yet been determined, but it is clear that any improvement in yield will derive from an improved harvest index.

6. Hydraulic Conductance Where crops have to produce their yield primarily on water stored in the soil prior to seeding, eking out the water supply throughout the life of the crop is of paramount importance (Passioura, 1972). Based on the observation that wheat has only one major xylem vessel in each seminal root, theoretical considerations suggested that reducing the diameter of this vessel would reduce water use by the wheat plant during vegetative growth, thereby providing more available water during grain filling (Richards and Passioura, 1981a). Richards and Passioura (1981b) searched for genotypes of wheat with small vessel diameters in their seminal roots and identified a landrace from Turkey with a vessel diameter of about 50 pm compared with vessel diameters of 60-65 pm in genotypes in use in Australia at the time. The landrace with the narrow vessels was backcrossed to two locally adapted genotypes to give lines of wheat similar to the parent, but with a reduced diameter of the main vessel in the xylem of the seminal roots. These lines yielded 3- 11% better than the unselected genotypes at dry sites and in dry seasons, and showed no yield reduction at wet sites and in wet seasons because in wet seasons the nodal roots developed and water uptake was not impaired (Richards and Passioura, 1989). The genotypes with restricted hydraulic conductance had greater biomass and a higher harvest index than the standard cultivars, suggesting that the water conserved in the vegetative phase enabled a better seed set and a longer period of grain filling in the reproductive phase.

7. Transpiration Efficiency The discovery by Farquhar and Richards ( 1984) that discrimination against the stable isotope of carbon, 13C, by plants with the C3 pathway of photosynthesis was correlated with their efficiency of transpiration has provided a ready means of evaluating the transpiration efficiency of species and genotypes. As pointed out under Section VI, the carbon isotope discrimination technique has been widely used over the past decade to determine the genotypic variation in transpi-

FURTHER PROGRESS IN CROP WATER RELATIONS

32 1

ration efficiency of a wide range of species (Table 11). Considerable progress has been made over the past decade in breeding for improved transpiration efficiency (Ehleringer et al., 1993). Studies on the heritability of transpiration efficiency indicate that broadsense heritability was generally high in crested wheatgrass (Johnson et al., 1990), peanut (Hubick et al., 1988), and wheat (Ehdaie et a/., 1991, 1993; Condon and Richards, 1992), but low to moderate in cowpea (Hall et af., 1993). Martin and Thorstenson (1988) suggested that transpiration efficiency may be under the control of relatively few genes in tomato. However, high transpiration efficiency appears to be associated with late flowering in cowpea (Hall et af., 1990), peanut (Hubick et al., 1986), barley (Craufurd et al., 1991), and bean (White, 1993) and low harvest index in cowpea (Hall et af., 1993) and peanut (Wright et af., 1993). Both these associations suggest that selection for high transpiration efficiency may result in lower grain yield under water-limited conditions. Furthermore, although early studies suggested that genotype by environment interactions were small for transpiration efficiency (Hall el al., 1990, 1992; Richards and Condon, 1993), other studies have suggested that rankings of genotypes vary with location (Condon ef al., 1992; Hall et al., 1993) and from experiment to experiment even at the same location (Masle et al., 1993). Breeding for improved transpiration efficiency using carbon isotope discrimination has been initiated (Hall et af., 1993; Richards and Condon, 1994) and it remains to be determined whether the associations between high transpiration efficiency and low harvest index and lateness can be broken and improved transpiration efficiency be used to improve drought resistance. It is clear from the previous discussion that considerable progress has been achieved in the past decade in using morphophysiological characters in breeding for improved drought resistance (Whan et af.,1991b). However, as Whan et al. (1993) have pointed out, the success of such breeding programs is dependent on the development of suitable screening techniques and the development of small teams that include agronomists, physiologists, biochemists, and breeders. Deveiopment of selection techniques or their surrogates is an important component of any “directed” breeding program in which physiological traits are utilized. The lack of suitable techniques has hampered the use of physiological attributes in plant breeding programs to date (Whan et af., 1993; Turner and Takeda, I993), but the use of molecular markers to track morphophysiological traits may provide potentially simpler methodologies in the future. The 1980s saw the development of teams, particularly at the International Agricultural Research Institutes, devoted to developing new genotypes based on physiological attributes. With the reduced funding for agricultural research in the more developed nations and in the International Agricultural Research Institutes, and the increased focus on molecular biology, many of these teams have been disbanded in the 1990s, just as progress was being observed (Turner and Takeda, 1993). However, they are still required to ensure future progress (Hay, 1993).

322

NEIL C. TURNER

B. USEOF MOLECULAR TECHNIQUES IN BREEDING FOR IMPROVED DROUGHT RESISTANCE Progress has been made during the past decade in the use of molecular techniques to improve the disease, insect, and herbicide resistance and quality aspects of several crop species. Many of the agronomically and economically important crops have been transformed so that genes for pest and disease resistance and, for example, improved amino acid composition, from a variety of sources can be and are being inserted. However, drought resistance is not a single gene (see Section VILA) and progress in improving drought resistance by the use of molecular techniques has not been as rapid. Water deficits have been shown to induce a wide range of molecular, cellular, and biochemical responses in plants (Fig. 6 ) . Although protein synthesis is generally suppressed by water deficits, some proteins increase with dehydration and disappear when the deficit is relieved. One group of proteins that are consistently induced by dehydration are the dehydrins (Close et al., 1993), which are part of the late embryogenesis abundant (LEA) class of proteins that have been shown to be stimulated by dehydration (Dure, 1993). Synthesis of LEA proteins has been shown to occur in the aboveground parts of a wide range of monocotyledenous and dicotyledenous plants, mosses, liverworts, and resurrection plants in response to desiccation (Bartels et al., 1993). Similar proteins are

lhnscrlptlon Control transcription factors; highmobility group proteins; chromatin structure; DNA methylation; cell cycle control

Osmolyte Biosynthesis proline; polyols; quaternary amines; sugars; ions; polyamines

Metabollte Balance ATP/ADP ratio; photosystem reduction; cytosol-organelletransport; one-carbon transfer

FURTHER PROGRESS IN CROP WATER RELATIONS

323

synthesized in the developing seeds as water content declines and the embryo matures (Dure et al., 1981; Bartels er al., 1993). LEA proteins and dehydrin production can also be induced by ABA, even in the absence of water stress, in some but not all plants (Bray et ul., 1993; Chandler et al., 1993). However, the function of the LEA and dehydrin genes is not clear, as overexpression or downregulation of the genes had no influence on the plants’ ability to withstand water deficits, and the introduction of three genes encoding for specific LEA proteins from resurrection plants into tobacco did not induce any change in response to stress in the transgenic tobacco (Ituriaga er al., 1992). Water deficits also induce accumulation of the compatible solutes, glycine betaine and proline betaine, and other quaternary ammonium compounds that are considered to act as osmoprotectants (Hanson, 1993). Genes for their overproduction have been identified and genetic modification of species to increase the content of compatible solutes is now feasible (Bartels and Nelson, 1994). Because so many biochemical pathways have been shown to respond to water deficits (Fig. 6), it is the identification of relevant mechanisms that is the challenge for the molecular biologist. Currently, modification of gene expression by genetic engineering is being used to test which traits are important in conferring drought resistance and results indicate that higher drought tolerance is only marginally increased by a single trait and that multiple mechanisms will need to be engineered to improve performance under moisture stress (Bartels and Nelson 1994; Van den Bulcke, 1994; Bohnert et al., 1995). Although Bohnert et al. (1995) are optimistic that progress is being made, even though the number of important genes involved exceeds the current resolving power of genetic screens, Van den Bulcke (1994) is far less optimistic and concludes that “successful engineering of drought- or salt-stress resistant crops is not on the horizon.” Because several genes are likely to be involved in each trait for drought resistance, molecular biology can aid in identifying and selecting these genes and in determining their influence on yield. O’Toole (1989) pointed out the value of using restriction fragment-length polymorphism (RFLP) technology as markers and selection tools for physiological, morphological, or developmental traits for drought resistance. This field has moved rapidly forward with a range of marker technologies now available (Quarrie, 1996). This has enabled high-density genetic maps to be made and the location of genes regulating the expression of physiological and agronomic characteristics that are inherited in a quantitative manner to be determined using a combination of statistical analysis, multiple regression, and maximum likelihood techniques (Jansen and Stam, 1994; Zeng, 1994). By analyzing these quantitative trait loci (QTL) for coincidence among traits, it is now possible to test whether the characteristics are causally related (Quarrie et a / . , 1995). QTL analysis has been used to identify a range of drought-resistance characteristics in wheat (Quarrie et af., 1994, 1995) and maize (Lebreton et al., 1995;

324

NEIL C . TURNER

Quarrie et al., 1995). In maize, 84 RFLP markers were mapped in an F, population from a cross between a drought-resistant and a drought-sensitive parent that was also measured for a range of putative drought-resistance characteristics after 21 days without water. Significant differences among the F2 population were found in stomatal conductance, leaf ABA, xylem ABA, osmotic pressure, turgor pressure, seminal and nodal root number, root pulling force, chlorophyll fluorescence, and flowering time, but not in leaf water potential or specific leaf area (Lebreton et al., 1995). By comparing the coincidence of QTL and a particular characteristic, it was suggested that xylem ABA rather than leaf ABA had a greater regulating effect on stomatal conductance, and that xylem ABA content was significantly associated with the root characteristics, suggesting that xylem ABA was likely regulated by the number of nodal roots that also determined the root pulling force. Subsequent QTL analysis, and measurement of ABA contents and yields in the F, generation, indicated that the yield under drought and ABA genes were located on different chromosomes in maize and hence are unlikely to be causally related (Quarrie et al., 1995). However, a strong QTL for flowering date and anthesis to silking interval was identified on the same chromosome as the genes for yield under drought. This strongly suggests that the anthesis-tosilking interval is linked to yield in maize and confirms the observations of Bolaiios and Edmeades (1993b) using eight cycles of recurrent selection (see Section VI1,A). Although molecular genetics has not yet had a major impact on improving the drought resistance of crops, the potential for significant gains is present. However, the warning of Blum (1994) that molecular biologists and agronomists are currently operating at different scales must be heeded, and for significant progress to be made, molecular biologists and agronomists or plant breeders will need to find common ground to test genes for drought resistance and to identify genes that will be useful for yield improvement under water-limited conditions.

VIII; CONCLUDING REMARKS The population of the world continues to increase at an alarming rate. Although approximately 100 million people are added to the global population every year and the population is expected to reach 8.5 billion by the year 2025, the area of cropable land per person will decrease as the area of arable land is stable or declining. Furthermore, the potential for increased irrigation is limited, so that future population increases will need to be fed from higher food production per unit land area and without the aid of increased irrigation resources. Thus, for improved food production, rainfall and irrigation water will need to be used more efficiently and the importance of understanding and managing crop water

FURTHER PROGRESS IN CROP WATER RELATIONS

325

deficits is a necessity. Clearly, improving the drought resistance of our food and fiber crops is of strategic importance and progress will need to be sustained. Although the new technologies of genetic engineering hold considerable promise, they will need to be coupled to traditional breeding and agronomy if these new technologies are to be fully utilized. Additionally, increased food production will need to be achieved without damage to the environment and without encroaching on the worlds’ forests, native vegetation, and other centers of biodiversity. Thus, cropping practices that lead to land degradation will need to be avoided. Because one of the factors that leads to land degradation is irrigation and poor water use by crops (Sadler and Turner, 1994), improving our understanding of crop water relations and its impact on irrigation management needs to be highlighted. In a world in which food limitations and sustainability issues are important, our understanding and management of crop water relations needs to continue to increase.

ACKNOWLEDGMENTS I thank Drs. Stephen Loss, Jairo Palta, and Tony Condon for comments and Vicki McSharer for typing the manuscript.

REFERENCES Acevedo, E. (1993). Potential of carbon isotope discrimination as a selection criteria in barley breeding. In “Stable Isotopes and Plant Carbon-Water Relations” (J. R. Ehleringer, A. E. Hall, and G. D. Farquhar, eds.), pp, 399-417. Academic Press, San Diego. Anderson, W: K . , and Smith, W. R. (1990a). Yield advantage of two semi-dwarf compared to two tall wheats depends on sowing time. Ausf. J. Agric. Res. 41, 811-826. Anderson, W. K., and Smith, W. R. (1990b). Incre’asing wheat yields in a high rainfall area of Western Australia. Ausf. J. Exp. Agric. 30, 607-614. Arraudeau, M. A. (1989). Breeding strategies for drought resistance. I n “Drought Resistance in Cereals” (F. W. G. Baker, ed.), pp. 107-116. ICSU Press, Paris. Austin, R. B., Bingham, J., Blackwell, R. D., Evans, L. T., Ford, M. A., Morgan, C. L., and Taylor, M. (1980). Genetic improvements in winter wheat yields since 1900 and associated physiological changes. J. Agric. Sci. 94, 675-689. Austin, R. B., Ford, M. A,, and Morgan, C. L. (1989). Genetic improvement in the yield of winter wheat: A further evaluation. J . Agric. Sci. 112, 295-301. Balling, A., and Zirnmermann, U. (1990). Comparative measurements of the xylem pressure of Nicofiana plants by means of the pressure bomb and pressure probe. Planfa 182, 325-328. Bano, A,, Dofling, K . , Bettin, D., and Hahn, H. (1993). Abscisic acid and cytokinins as possible root-to-shoot signals in xylem sap of rice plants in drying soil. A m . J. PIanf Physiol. 20, 10911s.

Barlow, E. W. R. (1986). Water relations of expanding leaves. Ausf. J . Plant Physiol. 13, 45-58. Bartels, D., and Nelson, D. (1994). Approaches to improve stress tolerance using molecular genetics. Plant Cell Environ. 17. 659-667.

326

NEIL C. TURNER

Bartels, D., Alexander, R., Schneider, K., Elster, R., Velasco, R., Alamillo, J., Bianchi, G., Nelson, D., and Salamini, F. (1993). Desiccation-related gene products analyzed in a resurrection plant and in barley embryos. In “Plant Responses to Cellular Dehydration during Environmental Stress” (T. J. Close and E. A. Bray, eds.), pp. 119-127. American Society of Plant Physiologists, Rockville, MD. Baughn, J. W., and Tanner, C. B. (1976). Leaf water potential: Comparison of pressure chamber and in situ hygrometer on five herbaceons species. Crop Sci. 16, 181-184. Begg, 1. E., and “ h e r , N. C. (1970). Water potential gradients in field tobacco. Plant Physiol. 46, 343-346. Begg, J. E., and ’hrner, N. C. (1976). Crop water deficits. Adv. Agron. 28, 161-217. Bellairs, S. M., Hick, P. T., Turner, N. C., and Smith, R. C. G. (1996). Plant and soil influences on estimating biomass of wheat in plant breeding plots using field spectral radiometers. Ausi. J. Agric. Res. 47, in press. Benkert, R., Balling, A., and Zimmermann, U. (1991). Direct measurements of the pressure flow in the xylem vessels of Nicotiana tabacum and their dependence on flow resistance and transpiration rate. Bat. Acta 104, 423-432. Bhagsari, A. S., and Brown, R. H. (1986). Leaf photosynthesis and its correlation with leaf area. Crop Sci. 26, 127-132. Blum, A. (1993). Selection for sustained production in water-deficit environments. In “International Crop Science I” (D. R. Buxton, R. Shibles, R. A. Forsberg, B. L. Blad, K. H. Asay, G. M. Paulsen, and R. F. Wilson, eds.), pp. 343-547. Crop Science Society of America, Madison, WI. Blum, A. (1994). Stress tolerance in plants: What are we looking for? In “Biochemical and Cellular Mechanisms of Stress Tolerance in Plants” (1. H. Cherry, ed.), pp. 315-324. Springer-Verlag. Berlin. Blum, A., Poiarkova, H., Gozlan, G., and klayer, J. (1983a). Chemical desiccation of wheat plants as a simulator of post-anthesis stress. I . Effects on translocation and kernel growth. Field Craps Res. 6, 51-58. Blum, A., Mayer, J., and Gozlan, G. (1983b). Chemical desiccation of wheat plants as a simulator of post-anthesis stress. 11. Relations to drought stress. Field Crops Res. 6, 149- 155. Blum, A., Shipler, L., Golan, J., Mayer, J., and Sinmena, B. (1991). Mass selection of wheat for grain filling without transient photosynthesis. Euphyticu 54, 11 1-1 16. Bohnert, H. I., Nelson, D. E., and Jensen, R. G. (1995). Adaptations to environmental stresses. Plant Cell 7, 1099- I 1 I 1, BolaAos, J., and Edmeades, G. 0. (1993a). Eight cycles of selection for drought tolerance in lowland tropical maize. I. Responses in grain yield, biomass, and radiation interception. Field Crops Res. 31, 244-252. BolaAos, J., and Edmeades, G. 0. (1993b). Eight cycles of selection for drought tolerance in lowland tropical maize. 11. Responses in reproductive behaviour. Field Crops Res. 31, 253-268. BolaAos, J., Edmeades, G. 0..and Martinez, L. (1993). Eight cycles of selection for drought tolerance in lowland tropical maize. 111. Responses in drought adaptive physiological and morphological traits. Field Crops Res. 31, 269-286. Bray, E. A. (1993). Molecular responses to water deficit. Plant Physiol. 103, 1035-1040. Bray, E. A., Moses, M. S., Imai, R., Cohen, A., and Plant, A. L. (1993). Regulation of gene expression by endogenous abscisic acid during drought stress. In “Plant Responses to Cellular Dehydration during Environmental Stress” (T. J. Close and E. A. Bray, eds.), pp. 167-176. American Society of Plant Physiologists, Rockville, MD. Bruckner, P. 0.. and Frohberg, R. C. (1987). Rate and duration of grain fill in spring wheat. Crop Sci. 27, 451-455. Chandler, P. M., Munns, R., and Robertson, M. (1993). Regulation of dehydnn expression. In “Plant Responses to Cellular Dehydration during Environmental Stress” (T. J. Close and E. A. Bray, eds.), pp. 159- 166. American Society of Plant Physiologists, Rockville, MD.

FURTHER PROGRESS IN CROP WATER RELATIONS

327

Cleland, R. E. (1986). The role of hormones in wall loosening and plant growth. Aust. J. Plant. Physiol. 13, 93-103. Close, T. J., Fenton, R. D., Yang, A., Asghar, R., DeMason, D. A,, Crone, D. E., Meyer, N. C., and Moonan, F. (1993). Dehydrin: The protein. In “Plant Responses to Cellular Dehydration during Environmental Stress” (T. J. Close and E. A. Bray, eds.), pp. 104-118. American Society of Plant Physiologists, Rockville, MD. Condon, A. G., and Richards, R. A. (1993). Exploiting genetic variation in transpiration efficiency in wheat: An agronomic view. In “Stable Isotopes and Plant Carbon-Water Relations” (J. R. Ehleringer, A. E. Hall, and G. D. Farquhar, eds.), pp. 435-450. Academic Press, San Diego. Condon, A. G . , Farquhar, G. D., and Richards, R. A. (1990).Genotypic variation in carbon isotope discrimination and transpiration efficiency in wheat. Leaf gas exchange and whole plant studies. Aust. J. Plant Physiol. 17, 9-22. Condon, A. G.. Richards, R. A., and Farquhar, G. D. (1992). The effect of variation in soil water availability, vapour pressure deficit and nitrogen nutrition on carbon isotope discrimination in wheat. Ausr. 1. Agric. Res. 43, 935-947. Cooper, P. J. M.. Gregory, P. J., Tully, D., and Harris, H. C. (1987).Improving water use efficiency of annual crops in the rainfed farming systems of West Asia and North Africa. Exp. Agric. 23, 113-158. Cornish, K., and Zeevaart, J. A. D. (1985). Abscisic acid accumulation by roots of Xanthium strumarium L. and Lycopersicon esculentum Mill. in relation to water stress. Plant Physiol. 79, 653-658. Cox, T. S., Shroyer, J. P., Lui, B.-H., Sears, R. G., and Martin, T. I.(1988). Genetic improvement in agronomic traits of hard red winter wheat cultivars from 1919 to 1987. Crop Sci. 28, 756-760. Craufurd, P. Q., Austin, R. B., Acevedo, E., and Hall, M. A. (1991).Carbon isotope discrimination and grain yield in barley. Field Crops Res. 27, 30 1-3 13. Davies, W. J . , Tardieu, F., and Trejo, C. L. (1994). How do chemical signals work in plants that grow in drying soil? Plant Physiol. 104, 309-314. Deckerd, E. L., Busch, R. H., and Koford, K. D. (1985). Physiological aspects of spring wheat improvement. In “Exploitation of Physiological and Genetic Variability to Enhance Crop Productivity” (J. E. Harper, L. E. Schrader, and R. w. Howell, eds.), pp. 46-54. American Society of Plant Physiology. Rockville, MD. Dingkuhn, M., Farquhar, G . D., De Datta, S. K., and O’Toole, J. C. (1991). Discrimination of I3C among upland rices having different water use efficiencies. Aust. J. Agric. Res. 42, 1123-1 131. Dofling, K., and Tietz, D. (1985). Abscisic acid in leaf epidermis of Commelina communis: Distribution and correlation with stomata1 closure. J. Plant Physiol. 117, 297-305. Duguid. S. D., and BriilC-Babel, A. L. (1994). Rate and duration of grain filling in five spring wheat (Triticum aestivum L.) genotypes. Can. J. Plant Sci. 74, 681-686. Dure, L. ( 1993). Structural motifs in Lea proteins. In “Plant Responses to Cellular Dehydration during Environmental Stress” (T. J. Close and E. A. Bray, eds.), pp. 91-103. American Society of Plant Physiologists, Rockville, MD. Dure, L., Greenway, S. C., and Galau, G. A. (1981). Developmental biochemistry of cottonseed embryogenesis and germination. XIV. Changing mRNA populations as shown by in virro and in vivo protein synthesis. Biochemistry 20, 4162-4168. Edmeades, G . O., Bolafios, J., Lafitte, H. R., Rajaram, S., Pfeiffer, W., and Fischer, R. A. (1989). Traditional approaches to breeding for drought resistance in cereals. In “Drought Resistance in Cereals” (F. W. G. Baker, ed.), pp. 27-52. ICSU Press, Paris. Edwards, I. (1992).Farming duplex soils: A farmer’s perspective. Aust. J. Exp. Agric. 32,811-814. Ehdaie, B.. Hall, A. E., Farquhar, G. D., Nguyen, H. T., and Waines, J. G. (1991). Water-use efficiency and carbon isotope discrimination in wheat. Crop Sci. 31, 1282-1288. Ehdaie, B . , Barnhart, D., and Waines, I. G. (1993). Genetic analyses of transpiration efficiency, carbon isotope discrimination, and growth characteristics in bread wheat. In “Stable Isotopes

328

NEIL C. TURNER

and Plant Carbon-Water Relations” (1. R. Ehleringer, A. E. Hall, and G. D. Farquhar, eds.), pp. 419-434. Academic Press, San Diego. Ehleringer, J. R. (1990). Correlations between carbon isotope discrimination and leaf conductance to water vapor in common beans. Plant Physiol. 93, 1422-1425. Ehleringer, J. R. (1993). Gas-exchange implications of isotopic variation in arid-land plants. In “Water Deficits: Plant Responses from Cell to Community” (J. A. C. Smith and H. Griffiths, eds.), pp, 265-284. Bios Scientific, Oxford, UK. Ehleringer, J. R., White, J. W., Johnson, D. A., and Brick, M. (1990). Carbon isotope discrimination, photosynthetic gas exchange and transpiration efficiency in beans and range grasses. Acta Oecol. 11, 1611-1615. Ehleringer, J. R., Klassen, S . , Clayton, C., Shemll, D., Fuller-Holbrook. M., Fu, Q., and Cooper, T. A. (1991). Carbon isotope discrimination and transpiration efficiency in common bean. Crop Sci. 31, 1611-1615. Ehlinger, J. R., Hall, A. E., and Farquhar, G. D. (1993). “Stable Isotopes and Plant Carbon-Water Relations.” Academic Press, San Diego. Elliott, G. A,, and Regan, K. L. (1993). Use of reflectance measurements to estimate early cereal biomass production on sandplain soils. Aust. J. Exp. Agric. 33, 179-183. Farquhar, G. D., and Richards, R. A. (1984). Isotopic composition of plant carbon correlates with water-use efficiency of wheat genotypes. Ausr. J. Planr Physiol. 11, 539-552. Farquhar, G. D., O’Leary,M. H., and Berry, 1. A. (1982). On the relationship between carbon isotope discrimination and the intercellular carbon dioxide concentration in leaves. Aust. J. Plant Physiol. 9, 121-137. Feyerhenn, A. M., Paulsen, G. M., and Sebaugh, J. L. (1984). Contribution of genetic improvement to recent wheat yields increases in the USA. Agron. J. 76, 985-990. Fischer, R. A. (1979). Growth and water limitations to dryland wheat yield in Australia in a physiological framework. J. Ausr. fnst. Agric. Sci. 45, 83-89. Fischer, R. A. (1981). Optimising the use of water and nitrogen through breeding of crops. Plant Soil 58, 249-278. Fischer, R. A., and Turner, N. C. (1978). Plant productivity in the arid and semi-arid zones. Annu. Rev. Plant Physiol. 29, 277-317. Flower, D. J., and Ludlow, M. M. (1987). Variation among accessions of pigeonpea (Cajanus cajan) in osmotic adjustment and dehydration tolerance of leaves. Field Crops Res. 17, 229-243. French, R. J., and Schultz, J. E. (1984a). Water use efficiency of wheat in a mediterranean-type environment. I. The relation between yield, water use and climate. Ausr. J. Agric. Res. 35,743764. French, R. J., and Schultz, 1. E. (1984b). Water use efficiency of wheat in a mediterranean-type environment. 11. Some limitations to efficiency. Aust. J . Agric. Res. 35, 765-775. French, R. J., and Turner, N. C. (1991). Water deficits change dry matter partitioning and seed yield in narrow-leafed lupins (Lupinusangustifolius L.). Aust. J . Agric. Res. 42, 471-484. Fry, S. C. (1986). Cross-linking of matrix polymers in the growing cell walls of angiosperms. Annu. Rev. Plant Physiol. 37, 165-186. Fry, S. C. (1989). Cellulases, hemicelluloses and auxin-stimulated growth: A possible relationship. Physiol. Plant. 15, 532-536. Fry, S . C., Smith, R. C., Renwick, K. F., Martin, D. J., Hodge, S. K., and Matthews, K. I. (1992). Xyloglucan endotransglycosylase, a new wall-loosening enzyme activity from plants. Biochem. J . 282, 821-828. Gallardo, M., Turner, N. C., and Ludwig, C. (1994). Water relations, gas exchange and abscisic acid content of Lupinus cosentinii leaves in response to drying different proportions of the root system. J. Exp. Bot. 45, 909-918. Gallardo, M., Eastham, J., Gregory, P. J., and Turner, N. C. (1996). A comparison of plant hydraulic conductances in wheat and lupins. J. Exp. Bot. 47, 233-239.

FURTHER PROGRESS IN CROP WATER RELATIONS

329

Gollan, T., Passioura, J. B., and Munns, R. (1986). Soil water status affects the stomata1 conductance of fully turgid wheat and sunflower leaves. Aust. J . Plant Physiol. 13, 459-464. Gollan, T.,Schurr, U., and Schulze, E.-D. (1992). Stomatal response to drying soil in relation to changes in the xylem sap composition of Helianthus annuus. I. The concentration of cations, anions, amino acids in, and pH of, the xylem sap. Planr Cell Environ. 15, 551-559. Grashoff, C. (1990). Effect of pattern of water supply on Vicia faba L. 11. Pod retention and filling, and dry matter partitioning, production and water-use. Netherlands J. Agric. Sci. 38, 131-134. Green, P. B., Erickson, R. 0..and Buggy, J. (1971). Metabolic and physical control of cell elongation rate. In vivo studies of Nirella. Plant Physiol. 47, 423-430. Greenwood, E. A. N., Turner, N. C., Schulze, E.-D., Watson, G. D., and Venn, N. R. (1992). Groundwater management through increased water use by lupin crops. J . Hydrol. 134, 1-11. Gregory, P. J., and Eastham, J. (1996). Growth of shoots and roots, and interception of radiation by wheat and lupin crops on a shallow duplex soil in response to time of sowing. Ausr. J. Agric. Res. 47, 427-447. Gregory, P. J., Tennant, D., Hamblin, A. P., and Eastharn, J. (1992). Components of the water balance on duplex soils of Western Australia. Ausr. J. Exp. Agric. 32, 845-855. Hall, A. E., Mutters, R. G., Hubick, K. T., and Farquhar, G. D. (1990). Genotypic differences in carbon isotope discrimination in cowpea under wet and dry field conditions. Crop Sci. 30,300-305. Hall, A. E., Mutters, R. G . , and Farquhar, G. D. (1992). Genotypic differences and drought-induced differences in carbon isotope discrimination and gas exchange of cowpea. Crop Sci. 32, 1-6. Hall, A. E., Ismail, A. M., and Menendez, C. M. (1993). Implications for plant breeding of genotypic and drought-induced differences in water-use efficiency, carbon isotope discrimination, and gas exchange. In “Stable Isotopes and Plant Carbon-Water Relations” (J. R. Ehleringer. A. E. Hall, and G. D. Farquhar, eds.), pp. 349-369. Academic Press, San Diego. Hamblin, A . P., and Hamblin, I. (1985). Root characteristics of some temperate legume species and varieties on deep, free-draining entisols. Aust. J. Agric. Res. 36, 63-72. Hamblin, A. P., and Kyneur, G. (1993). “Trends in Wheat Yields and Soil Fertility in Australia.” Australian Government Publishing Service, Canberra. Hamblin, J. (1987). Grain legumes in Australia. In “Agronomy 1987-Responding to Change. Proceedings of the 4th Australian Agronomy Conference” (T. G . Reeves, ed.), pp. 65-82. Australian Society of Agronomy, Parkville. Hanks, R. 1.. and Brown, R. W. (1987a). “Proceedings of International Conference on Measurement of Soil and Plant Water Status,” Vol. 1. Utah State Univ., Logan. Hanks, R. J., and Brown, R. W. (1987b). “Proceedings of International Conference on Measurement of Soil and Plant Water Status,” Vol. 2. Utah State Univ., Logan. Hanks, R. I., and Brown, R. W. (1987~).“Proceedings of International Conference on Measurement of Soil and Plant Water Status,” Vol. 3. Utah State Univ., Logan. Hanson, A. D. (1993). Accumulation of quaternary ammonium and tertiary sulfoniurn compounds. In “Plant Responses to Cellular Dehydration during Environmental Stress” (T. I. Close and E. A. Bray, eds.), pp. 30-36. American Society of Plant Physiologists, Rockville, MD. Hartung, W., and Davies, W. J. (1993). Abscisic acid under drought and salt stress. In “Handbook of Plant and Crop Stress” (M. Pessarakli, ed.). pp. 401-41 1. Dekker, New York. Hartung, W., and Slovik. S. (1991). Physiochemical properties of plant growth regulators and plant tissues determine their distribution and redistribution: Stomatal regulation by abscisic acid in leaves. New Phytol. 119, 361-382. Hay, R. K. M. (1993). The contribution of physiology to breeding progress. In “Physiology of Varieties” (E. White, P. S. Kettlewell, M. A . Parry, and R. P. Ellis, eds.) (Aspecrs Appl. B i d . 34, pp. 17-24). Association of Applied Biologists, Wellesbourne, Australia. Heam, A. B. (1975). Response of cotton to water and nitrogen in a tropical environment. 1. Frequency of water and method of application of nitrogen. J. Agri. Sci. 84, 407-417. H e m , A. B. (1979). Water relationships in cotton. Outlook Agric. 10, 159-166.

NEIL C. TURNER Henson, I. E. (1982). Osmotic adjustment to water stress in pearl millet (Pennnisetum americanum (L.) Leeke) in a controlled environment. J. Exp. Bot. 33, 78-87. Henson, I. E., and ’hrner, N. C. (1991). Stomata1 responses to abscisic acid in three lupin species. New Phytol. 117, 529-534. Henson, I. E., Jensen, C. R., and ’hrner, N. C. (1989a). Leaf gas exchange and water relations of lupins and wheat. I. Shoot responses to soil water deficits. Aust. J. Plant Physiol. 16,401-413. Henson, 1. E., Jensen, C. R., and Turner, N. C. (1989b). Leaf gas exchange and water relations of lupins and wheat. 111. Abscisic acid and drought-induced stomata1 closure. Aust. J. Plant Physiol. 16, 429-442. Hossain, A. B. S., Sears, R. G., Cox, T. S., and Paulsen, G. M. (1990). Desiccation tolerance and its relationship to assimilate partitioning in winter wheat. Crop Sci. 30,622-627. Hsiao, T. C., and Jing. J. (1987). Leaf and root expansive growth in response to water deficits. In “Physiology of Cell Expansion During Plant Growth” (D. J. Cosgrove and D. P. Knievel, eds.), pp. 180- 192. American Society of Plant Physiologists, Rockville, MD. Hsaio, T. C., O’Toole, J. C., Yambao, E. B., and Turner, N. C. (1984). Influence of osmotic adjustment on leaf rolling and tissue death in rice (Oryza sativa L.). Planr Physiol. 75, 338341. Hubick, K. T., and Farquhar, G. D. (1987). Carbon isotope discrimination-Selecting for water use efficiency. Aust. Cotton Grower 8, 66-68. Hubick, K. T., and Farquhar, G. D. (1989). Genetic variation in carbon isotope discrimination and the ratio of carbon gained to water lost in barley. Plant Cell Environ. 12, 795-804. Hubick, K. T., Farquhar, G. D., and Shorter, R. (1986). Correlation between water-use efficiency and carbon isotope discrimination in diverse peanut (Arachis) germplasm. Aust. J . Plant Physi01. 13, 803-816. Hubick, K. T., Shorter, R., and Farquhar, G. D. (1988). Heritability and genotype X environment interactions of carbon isotope discrimination and transpiration efficiency in peanut (Arachis hypogeae L.). Aust. J. Plant. Physiol. 15, 799-813. Hiisken, D., Steudle, E., and Zimmermann, U. (1978). Pressure probe technique for measuring water relations of cells in higher plants. Plant Physiol. 61, 158-163. Ismail, A. M., and Hall, A. E. (1992). Correlation between water-use efficiency and carbon isotope discrimination in diverse cowpea genotypes and isogenic lines. Crop Sci. 32, 7-12. Ismail, A. M., and Hall, A. E. (1993). Inheritance of carbon isotope discrimination and water-use efficiency in cowpea. Crop Sci. 33, 498-503. Itoh, K., Nakamura, Y., Kawata, H., Yamada, T., Ohta, E., and Sakata, M. (1987). Effect of osmotic stress on turgor pressure in mung bean root cells. Plant Cell Physiol. 28, 987-994. Itumaga, G., Schneider, K., Salamini, F., and Bartels, D. (1992). Expression of desiccation-related proteins from the resurrection plant Craterostigma plantagineum in trangenic tobacco. Plant Mol. Biol. 20, 555-558. Jackson, R. D. (1982). Canopy temperature and crop water stress. Adv. Irrigation 1, 43-85. Jackson, R. D. (1987). The crop water stress index: A second look. In “Proceedings of International Conference on Measurement of Soil and Plant Water Status” (R. J. Hanks and R. W. Brown, eds.), Vol. 2, pp. 87-91. Utah State Univ., Logan. Jansen, R. C., and Stam, P. (1994). High resolution of quantitative traits into multiple loci via interval mapping. Genetics 136, 1447- 1455. Johnson, D. A., Asay, K. H., Tieszen, L. L., Ehleringer, J. R., and Jefferson, P. G. (1990). Carbon isotope discrimination: Potential in screening cool-season grasses for water-limited environments. Crop Sci. 30, 338-343. Johnson, R. C., and Bassett, L. M. (1991). Carbon isotope discrimination and water use efficiency in four cool-season grasses. Crop Sci. 31, 157-162. Jones, H. G. (1993). Drought tolerance and water use efficiency. In “Water Deficits: Plant Responses

FURTHER PROGRESS IN CROP WATER RELATIONS

331

from Cell to Community” (J. A. C. Smith and H. Griffiths, eds.), pp. 193-203. Bios Scientific, Oxford, UK. Jones, M. M., and Rawson, H. M. (1979). Influence of the rate of development of leaf water deficits upon photosynthesis, leaf conductance, water use efficiency, and osmotic potential in sorghum. Phvsiol. Planr 45, 103- I 11. Kobata, T.. Palta, J. A., and Turner, N. C. (1992). Rate of development of postanthesis water deficits and grain filling in spring wheat. Crop Sci. 32, 1238-1242. Kramer, P. J. (1988). Changing concepts regarding plant water relations. Plant Cell Environ. 11, 565-567. Kuang, J.-B., Turner, N. C., and Henson, I. E. (1990). Influence of xylem water potential on leaf elongation and osmotic adjustment of wheat and lupin. J . Exp. Bot. 41, 217-221. Kuroyanagi, T., and Paulsen, G. M. (1988). Mediation of high-temperature injury by roots and shoots during reproductive growth of wheat. Plant Cell Environ. 11, 517-523. Kuzmanoff, K. M.. and Evans, M. L. (1981). Kinetics of adaptation of osmotic stress in lentil (Lens culinaris Med.) roots. Planr Phvsiol. 68, 244-247. Lachno, D. R.. and Baker, D. A. (1986). Stress induction of abscisic acid in maize roots. Physiol. Plant. 68, 215-221. Lebreton, C., LaziC-JantiC, V., Steed. A,, PekiC. S . , and Quarrie, S. A. (1995). Identification of QTL for drought responses in maize and their use in testing causal relationships between traits. J . EX^. Bat. 46, 853-865. Lockhart, J. A. (1965). An analysis of irreversible plant cell elongation. J. Theor. Biol. 8, 264275. Loss, S. P., and Siddique, K. H.M. (1994). Morphological and physiological traits associated with wheat yield increases in mediterranean environments. Adv. Agron. 52, 229-276. Loss, S . P.. Siddique, K. H. M., and Perry, M. W. (1989). Grain growth and development of old and modern Australian wheats. Field Crops Res. 21, 131-146. Ludlow, M. M.. and Muchow, R. C. (1990). A critical evaluation of traits for improving crop yields in water-limited environments. Adv. Agron. 43, 107-153. Ludlow, M. M., Santamaria, J. M., and Fukai, S. (1990). Contribution of osmotic adjustment to grain yield in Sorghum bicolor (L.) Moench under water-limited conditions. 11. Water stress after anthesis. Ausr. J. Agric. Res. 41, 67-78. Mansfield. T. A. ( 1976). Stomatal behaviour: Chemical control of stomata1 movements. Philos. Trans. R . Soc. London Ser. B 273, 541-550. Martin, B., and Thorstenson, Y. R. (1988). Stable carbon isotope composition (8 I T ) , water use efficiency and biomass productivity of Lvcopersicon esculentum. Lycopersicon pennelli. and the F, hybrid. Planr Physiol. 88, 213-217. Masle, J., Shin, J. S . , and Farquhar, G. D. (1993). Analysis of restriction fragment length polymorphisms associated with variation of carbon isotope discrimination among ecotypes of Arabidopsis thaliana. In “Stable Isotopes and Plant Carbon-Water Relations” (J. R. Ehleringer, A.E. Hall, and G. D. Farquhar, eds.). pp. 371-386. Academic Press, San Diego. McWilliam, J. R. (1989). The dimensions of drought. In “Drought Resistance in Cereals” (F. W. G. Baker, ed.), pp. 1- I 1. ICSU Press, Pans. Meinzer, F. C., Goldstein, G., and Grantz, D. A. (1990). Carbon isotope discrimination in coffee genotypes grown under limited water supply. Planr Physiol. 92, 130-135. Meyer, R. F., and Boyer, J. S. (1981). Osmoregulation, solute distribution and growth in soybean seedlings having low water potentials.Planta 151, 482-489. Michelena, V. A,, and Boyer, J. S. (1982). Complete turgor maintenance at low water potentials in the elongating region of maize leaves. Plant Phvsiol. 69, 1145- 1149. Mittelheuser, C. I., and Van Steveninck, R. F. M. (1969). Stomatal closure and inhibition of transpiration induced by (RS)-abscisic acid. Nature (London) 221, 281-282.

332

NEIL C. TURNER

Morgan, J. M. (1983). Osmoregulation as a selection criterion for drought tolerance in wheat. Aust. J. Agric. Res. 34, 607-614. Morgan, J. M. (1984). Osmoregulation and water stress in higher plants. Annu. Rev. Plant Physiol. 35, 299-319. Morgan, J. M., and Condon, A. G. (1986). Water use, grain yield and osmoregulation in wheat. Aust. J . Plant Physiol. 13, 523-532. Morgan, J. M., Hare, R. A,, and Fletcher, R. J. (1986). Genetic variation in osmoregulation in bread and durum wheats and its relationship to grain yield of field environments. Aust. J . Agric. Res. 37,449-457. Morgan, J. M., Rodriguez-Maribona, B., and Knights, E. J. (1991). Adaptation to water-deficit in chickpea breeding lines by osmoregulation: Relationship to grain-yields in the field. Field Crops Res. 27, 61-70. Munns, R. (1992). A leaf elongation bioassay detects an unknown growth inhibitor in xylem sap from wheat and barley. Aust. J. Plant Physiol. 19, 127-135. Munns, R., and King, R. W. (1988). Abscisic acid is not the only stomata1 inhibitor in the transpiration stream of wheat plants. Plant Physiol. 88, 703-708. Munns, R., and Sharp, R. E. (1993). Involvement of abscisic acid in controlling plant growth in soils of low water potential. Aust. J . Plant. Physiol. 20, 425-437. Munns, R., Passioura, J. B., Milborrow, B. V., James, R.,and Close, T. J. (1993). Stored xylem sap from wheat and barley contains a transpiration inhibitor that is retained by high molecular weight exclusion filters. Plant Cell Environ. 16, 867-872. Netting, A. G., Williams, R. D., and Milborrow, B. V. (1992). The isolation and characterization of the prosthetic group released from a bound form of abscisic acid. Plant Growth Regul. 11,327334. Neumann, P. M. (1995). The role of cell wall adjustment in plant resistance to water deficits. Crop Sci. 35, 1258-1266. Nicolas, M. E., and Tbmer, N. C. (1993). Use of chemical desiccants and senescing agents to select wheat lines maintaining stable grain size during post-anthesis drought. Field Crops Res. 31, 155-171. Nicolas, M. E., Gleadow, R. M., and Dalling, M. J. (1985). Effects of post-anthesis drought on cell division and starch accumulation in developing wheat grains. Ann. Bot. 55, 433-444. Nonami, H., and Boyer, J. S. (1989). ’hrgor and growth at low water potentials. Plant Physiol. 89, 798-804. Nonami, H., and Boyer, J. S. (1990a). Primary events regulating stem growth at low water potentials. Plant Physiol. 93, 1601-1609. Nonami, H.,and Boyer, J. S. (1990b). Wall extensibility and cell wall hydraulic conductivity decrease in enlarging sIem tissues at low water potentials. Plant Physiol. 93, 1610-1619. O’Brien, L. (1982). Victorian wheat yield trends, 1898-1977. J. Aust. Inst. Agric. Sci. 48, 163168. O’Leary, M. H. (1981). Carbon isotope fractionation in plants. Phytochemisrry 20, 553-557. Ong, C. K. (1984). The influence of temperature and water deficits on the partitioning of dry matter in groundnut (Arachis hypogaea L.) J . Exp. Bot. 35, 746-755. O’Toole, J. C. (1989). Breeding for drought resistance in cereals: Emerging new technologies. In “Drought Resistance in Cereals” (F. W. G. Baker, ed.), pp. 81-106. ICSU Press, Paris. Palta, J. A., and Fillery, I. R. P. (1995a). N application increases pre-anthesis contribution of dry matter to grain yield in wheat grown on a duplex soil. Aust. J. Agric. Res. 46, 507-518. Palta, J. A., and Fillery, I. R. P. (1995b). N application enhances remobilization and reduces losses of pre-anthesis N in wheat grown on a duplex soil. Aust. J. Agric. Res. 46, 519-531. Palta, J. A , , Wyn-Jones, R. G., and Tomos, A. D. (1987). Leaf diffusive conductance and tap root cell turgor pressure of sugar beet. Plant Cell Environ. 10, 735-740.

FURTHER PROGRESS IN CROP WATER RELATIONS

333

Palta, J. A., Kobata, T., Turner, N. C., and Fillery, 1. R. (1994). Remobilization of carbon and nitrogen in wheat as influenced by postanthesis water deficits. Crop Sci. 34, 118-124. Passioura, J. B. (1972). The effect of root geometry on the yield of wheat growing on stored water. Aust. J. Agric. Res. 23, 745-752. Passioura, J. B. (1977). Grain yield, harvest index and water-use of wheat. J. Aust. Insr. Agric. Sci. 43, 117-120. Passioura, J. B. (1988). Root signals control leaf expansion in wheat seedlings growing in drying soil. Aust. J. Plant Physiol. 15, 687-693. Passioura, J. B. (1991). An impasse in plant water relations? Bot. Acta 104, 405-411. Passioura, J. B., and Fry, S. C. (1992). Turgor and cell expansion: Beyond the Lockhart equation. Aust. J . Plant Physiol. 19, 565-576. Passioura, J. B., Condon, A. G., and Richards, R. A. (1993). Water deficits, the development of leaf area and crop productivity. In “Water Deficits: Plant Responses from Cell to Community” (J. A. C. Smith and H. Griffiths, eds.), pp. 253-264. Bios Scientific, Oxford, UK. Perry, M. W. (1987). Water use efficiency of non-irrigated field crops. In “Agronomy 1987Responding to Change. Proceedings of the 4th Australian Agronomy Conference” (T. G. Reeves, ed. 1, pp. 83-99. Australian Society of Agronomy, Parkville. Perry. M. W., and D’Antuono, M. F. (1989). Yield improvement and associated characteristics of some Australian spring wheat cultivars introduced between 1860 and 1982. Aust. J . Agric. Res. 40, 457-472. Peterson, C. A., and Steudle, E. (1993). Lateral hydraulic conductivity of early metaxylem vessels in Zea mays L. roots. Planta 189, 288-297. Pheloung, P. C., and Siddique, K. H. M. (1991). Contribution of stem dry matter to grain yield in wheat cultivars. Aust. J. Plant Physiol. 18, 53-64. Pritchard, J., and Tomos. A. D. (1993). Correlating biophysical and biochemical control of root cell expansion. In “Water Deficits: Plant Responses from Cell to Community” (J. A. C. Smith and H. Criffiths, eds.), pp. 53-72. Bios Scientific, Oxford, UK. Quame, S. A. (1996). New molecular tools to improve the efficiency of breeding for increased drought resistance. In “Interdrought 95--Proceedings of the First International Meeting on Integrated Studies on Drought Tolerance of Higher Plants, Montpellier, France, August/September 1995.” In press. Quarrie, S. A,, and Jones, H. G. (1977). Effects of abscisic acid and water stress on development and morphology in wheat. J. Exp. Bot. 28, 182-203. Quarrie, S. A,, Gulli, M.,Calestani, C., Steed, A,, and Marmiroli, N. (1994). Location of a gene regulating drought-induced abscisic acid production on the long arm of chromosome 5A of wheat. Theor. Appl. Genet. 89, 794-800. Quarrie, S . A , , PekiC, S., Lazic-JanfiC. V., IvanoviC, M., and VasiC, G. (1995). ABA and drought tolerance in maize: Future prospects. In “Maize Breeding, Production and Utilization: 50 Years of Maize Research Institute, Zemun Polje” (Z. Videnovic, ed.), pp. 121-137. Maize Research Institute, Belgrade. Raschke, K. (1975). Stomata1 action. Annu. Rev. Plant Physiol. 26, 309-340. Read, J. J., Johnson, D. A , , Asay, K. H . , and Tieszen. L. L. (1991). Carbon isotope discrimination, gas exchange, and water-use efficiency in crested wheatgrass clones. Crop Sci. 31, 12031208. Regan, K . L., Siddique, K. H. M., a r n e r . N . C., and Whan, B. R. (1992). Potential for increasing early vigour and total biomass in spring wheat. 11. Characteristics associated with early vigour. Aust. J. Agric. Res. 43, 541-553. Regan, K . L., Whan, B. R . , and Turner, N. C. (1993). Evaluation of chemical desiccation as a selection technique for drought resistance in a dryland wheat breeding program. Aust. J. Agric. Res. 27, 1683-1691.

334

NEIL C. TURNER

Richards, R. A. (1989). Breeding for drought resistance-Physiological approaches. In “Drought Resistance in Cereals” (F. W. G. Baker, ed.), pp. 65-79. ICSU Press, Paris. Richards, R. A. (1991). Crop improvement for temperate Australia: Future opportunities. Field Crops Res. 26, 141-169. Richards, R. A,, and Condon, A. G. (1993). Challenges ahead in using carbon isotope discrimination in plant-breeding programs. In “Stable Isotopes and Plant Carbon-Water Relations” ( J . R. Ehleringer, A. E. Hall, and G. D. Farquhar, eds.), pp. 451-462. Academic Press, San Diego. Richards, R. A., and Condon, A. G. (1994). New ‘yield genes’ for wheat. In “Proceedings of the 7th Assembly, Wheat Breeding Society of Australia, September 1994” (J. Paull, I. S. Dundas, K. 1. Shepherd, and G. J. Hollamby, eds.), pp. 159-163. Univ. of Adelaide, Adelaide, Australia. Richards, R. A., and Passioura, J. B. (1981a). Seminal root morphology and water-use in wheat. I. Environmental effects. Crop Sci. 21, 249-252. Richards, R. A,, and Passioura, J. B. (1981b). Seminal root morphology and water-use in wheat. 11. Genetic variation. Crop Sci. 21, 253-255. Richards, R. A,, and Passioura, J. B. (1989). A breeding program to reduce the diameter of the major xylem vessel in the seminal roots of wheat and its effect on grain yield in rain-fed environments. Ausr. J. Agric. Res. 40,943-950. Riggs, T. J., Hanson, P. R., Start, N. D., Miles, D. N., Morgan, C. L., and Ford, M. A. (1981). Comparison of spring barley cultivars grown in England and Wales between 1880 and 1980. J . Agric. Sci. 95, 599-610. Ritchie, J. T. (1974). Atmospheric and soil water influences on the plant water balance. Agric. Meteorol. 14, 183- 198. Ritchie, J. T. (1981). Water dynamics in the soil-plant-atmosphere system. Plant Soil 58, 81-96. Russell, I. S. (1973). Yield trends of different crops in different areas and reflections on the sources of yield improvement in the Australian environment. J. Aust. lnst. Agric. Sci. 39, 156-166. Russell, M. B. (1959). Water and its relation to soils and crops. Adv. Agron. 11, 1-131. Saab, I. N., and Sharp, R. E. (1989). Non-hydraulic signals from maize roots in drying soil: Inhibition of leaf elongation but not stomata1 conductance. Pluntu 179, 466-474. Saab, I. N., Sharp, R. E., Pritchard, J., and Voetberg, G. S. (1990). Increased endogenous abscisic acid maintains primary root growth and inhibits shoot growth of maize seedlings at low water potentials. Plant Physiol. 93, 1329- 1336. Saab, I. N., Sharp, R. E., and Pritchard, J. (1992). Effect of inhibition of abscisic acid accumulation on the spatial distribution of elongation in the primary root and mesocotyl of maize at low water potentials. Plant Physiol. 99, 26-33. Sadler, E. I., and lhrner, N. C. (1994). Water relationships in a sustabilable agriculture system. In “Sustainable Agriculture Systems” (J. L. Hatfield and D. L. Karlen, eds.). pp. 21-46. Lewis, Baton Rouge. LA. Salmon, S. C., Mathews, 0. R., and Leukel, R. W. (1953). A half century of wheat improvement in the United States. Adv. Agron. 5 , 1-151. Santamaria, J. M., Ludlow, M. M., and Fukai, S. (1990). Contributions of osmotic adjustment to grain yield in Sorghum bieoior (L.) Moench under water-limited conditions. 1. Water stress before anthesis. Aust. J . Agric. Res. 41, 51-65. Schurr, U . , Gollan, T., and Schulze, E.-D. (1992). Stomatal response to drying soil in relation to changes in the xylem sap composition of Heliunthus unnuus. 11. Stomatal sensitivity to abscisic acid imported from the xylem sap. Plant Cell Environ. 15, 561-567. Shepherd, K. D., Cooper, P. J. M., Allan, A. Y., Drennan, D. S. H., and Keatinge, J. D. H. (1987). Growth, water use and yield of barley in Mediterranean-type environments. J . Agric. Sci. 108, 365-378. Siddique, K. H. M., Belford, R. K., Perry, M. W., and Tennant, D. (1989a). Growth, development

FURTHER PROGRESS IN CROP WATER RELATIONS

335

and light interception of old and modem wheat cultivars in a mediterranean-type environment. Aust. J. Agric. Res. 40, 473-487. Siddique, K. H. M., Kirby, E. J. M., and Perry, M. W. (1989b). Earstem ratio in old and modem wheat varieties: Relationship with improvement in number of grains per ear and yield. Field Crops Res. 21, 59-78. Siddique, K. H. M., Tennant, D., Perry, M. W., and Belford, R. K. (1990). Water use and water use efficiency of old and modem wheat cultivars in a mediterranean-type environment. Ausr. J . Agric. Res. 41, 431-447. Silvey, V. (1981). The contribution of new wheat, barley and oat cultivars to increasing yield in England and Wales, 1947-78. J . Natl. Inst. Agric. Bot. 15, 399-412. Slovik, S., Daeter, W., and Hartung, W. ( I 995). Compartmental redistribution and long-distance transport of abscisic acid (ABA) in plants as influenced by environmental changes in the rhizosphere-A biomathematical model. J. Exp. Bor. 46, 88 1-894. Smith, R. C. G . , Wallace, J. F., Hick, P. T., Gilmour, R. F., Belford, R. K., Portmann, P. A,, Regan, K. L., and lhmer, N. C. (1993). Potential for using field spectroscopy during early growth for ranking biomass in cereal breeding trials. Aust. J . Agric. Res. 44, 1713-1730. Sobrado, M. A., and ’hmer, N. C. (1986). Photosynthesis, dry matter accumulation and distribution in the wild sunflower Helianthus periolaris and the cultivated sunflower Helianthus annuus as influenced by water deficits. Oecologin 69, 181-187. Spollen, W. G., and Sharp, R. W. (1991). Spatial distribution of turgor and root growth at low water potentials. Plant Physiol. 96,438-443. Spollen, W. G., Sharp, R. W., Saab, I. N., and Wu, Y. (1993). Regulation of cell expansion in roots and shoots at low water potentials. I n “Water Deficits: Plant Responses from Cell to Community” (J. A. C. Smith and H. Grifliths, eds.), pp. 37-52. Bios Scientific, Oxford, UK. Steudle, E. ( 1989a). Water flow in plants and its coupling to other processes: An overview. Methods Enzymol. 174, 183-225. Steudle, E. (1989b). Water transport in roots. I n “Structural and Functional Aspects of Transport in Roots” (B. C. Loughman, 0. Gasparikova, and J. Kolek, eds.), pp. 139-145. Kluwer Academic, Amsterdam. Steudle, E. (1993). Pressure probe technique: Basic principles and application to studies of water and solute relations at the cell, tissue and organ level. I n “Water Deficits: Plant Responses from Cell to Community” (J. A. C. Smith and H. Griffiths, eds.), pp. 5-36. Bios Scientific, Oxford, UK. Steudle, E., and Jeschke, W. D. (1983). Water transport in barley roots. Planta 158, 237-248. Steudle, E., and Zimmermann, U. (1974). Determination of hydraulic conductivity and of reflection coefficients in Nitella JIexilis by means of direct cell-turgor pressure measurements. Biochem. Bivphys. Acra 322, 399-412. Steudle, E., Oren, R., and Schulze, E.-D. (1987). Water transport in maize roots. Plantfhysiol. 84, 1220- 1232. Taiz, L. (1984). Plant cell expansion: Regulation of cell wall mechanical properties. Annu. Rev. Plant Physiol. 35, 585-657. Tardieu, F., and Davies, W. J. (1992). Stomata1response to abscisic acid is a function of current plant water status. Plant Phvsiol. 98, 540-545. Tardieu, F.. Zhang, I., and Davies, W. J. (1992a). What information is conveyed by an ABA signal from maize roots in drying field soil? Plan1 Cell Environ. 15, 185-191. Tardieu, F., Zhang, J., Katerji, N . , Bethenod, 0.. Palmer, S., and Davies, W. J. (1992b). Xylem ABA controls the stomata1 conductance of field-grown maize subjected to soil compaction or soil drying. Plant Cell Environ. 15, 193-197. Tennant, D., Scholz. G.. Dixon, J., and Purdie, B. (1992). Physical and chemical characteristics of duplex soils and their distribution in the south-west of Westem Australia. Aust. J . Exp. Agric. 32, 827-843.

336

NEIL C . TURNER

Tomos, A. D., Hinde, P., Richardson, P., Pritchard, J., and Fricke, W. (1994). Micro-sampling and measurements of solutes in single cells. In “Plant Cell Biology-A Practical Approach (N. Harris and K. J. Oparka, eds.), pp. 149-160. IRL Press, Oxford. Thberosa, R., Sanguineti, M. C., and Landi, P. (1994). Abscisic acid concentration in leaf and xylem sap, leaf water potential and stomata1 conductance in maize. Crop Sci. 34, 1557-1563. Turner, N. C. (1979). Drought resistance and adaptation to water deficits in crop plants. In “Stress Physiology in Crop Plants” (H. Mussell and R. C. Staples, eds.), pp. 343-372. Wiley, New York. Turner, N. C. (1982). The role of shoot characteristics in drought resistance of crop plants. In “Drought Resistance in Crops with Emphasis on Rice,” pp. 115-134. International Rice Research Institute, Los Bafios, The Philippines. Turner, N. C. (1986a). Crop water deficits: A decade of progress. Adv. Agron. 39, 1-51. h r n e r , N. C. (1986b). Adaptation to water deficits: A changing perspective. Aust. J . Plan? Physiol. 13, 175-190. Turner, N. C. (1988). Measurement of plant water status by the pressure chamber technique. Irrigarion Sci. 9, 289-308. Turner, N.C. (1990a). Plant water relations and irrigation management. Agric. Water Management 17, 59-93. Turner, N. C. (1990b). The benefits of water deficits. In “Proceedings of the International Congress of Plant Physiology, New Delhi, India, February 1988” (S. K. Sinha, P. V. Sane, S. C. Bargava, and P. K. Agrawal, eds.), Vol. 2, pp. 806-815. Indian Society of Plant Physiology and Biochemistry, New Delhi. Turner, N. C. (1992). Crop production on duplex soils: An introduction. Ausr. J . Exp. Agric. 32, 797-800. Turner, N. C. (1993). Water use efficiency of crop plants: Potential for improvement. In “International Crop Science I” (D. R. Buxton, R. Shibles, R. A. Forsberg, B. L. Blad, K. H. Asay, G. M. Paulsen, and R. F. Wilson, eds.), pp. 75-82. Crop Science Society of America, Madison, WI. nrner, N. C., and Begg, J. E. (1978). Responses of pasture plants to water deficits. In “Plant Relations in Pastures” (J. R. Wilson, ed.), pp. 50-66. CSIRO, Melbourne, Australia. Turner, N. C., and Begg, J. E. (1981). Plant-water relations and adaptation to stress. Plan? Soil 58, 97- 131. Turner, N. C., and Henson, I. E. (1989). Comparative water relations and gas exchange of wheat and lupins in the field. In “Structural and Functional Responses to Environmental Stresses: Water Shortage” (K. H. Kreeb, H. Richter, and T. M. Hinckley, eds.), pp. 293-304. SPB Academic, The Hague, The Netherlands. nrner, N. C., and Jones, M.M. (1980). Turgor maintenance by osmotic adjustment: A review and evaluation. I n “Adaptation of Plants to Water and High Temperature Stress” (N. C. Turner and P. J. Kramer, eds.), pp. 87-103. Wiley, New York. ’hrner, N. C., and Nicolas, M. E. (1987). Drought resistance of wheat for light-textured soils in a mediterranean climate. In “Drought Tolerance in Winter Cereals” (J. P. Srivastava, E. Porceddu, E. Acevedo, and S. Varma, eds.), pp. 203-216. Wiley, Chichester, UK. Thrner, N. C., and Takeda, G . (1993). Use of physiological attributes in plant breeding-Discussion. In “International Crop Science I” (D. R. Buxton, R. Shibles, R. A. Forsberg, B. L. Blad, K. H. Asay, G. M. Paulsen, and R. F. Wilson, eds.), pp. 619-620. Crop Science Society of America, Madison, WI. Turner, N. C., Begg, J. E., Rawson, H. M., English, S. D., and Hearn, A. B. (1978). Agronomic and physiological responses of soybean and sorghum to water deficits. 111. Components of leaf water potential, leaf conductance, I4CO2 photosynthesis and water use efficiency of leaves. Aust. J. Plant Physiol. 5 , 195-209. Turner, N. C . , Schulze, E.-D., and Gollan, T. (1985). The responses of stomata and leaf gas

FURTHER PROGRESS IN CROP WATER RELATIONS

337

exchange to vapour pressure deficits and soil water content. 11. In the mesophytic herbaceous species Helianthus annuus. Oecologia 65, 348-355. Turner, N. C., O’Toole, J. C., Cruz, R. T., Yambao, E. B., Ahmad, S., Namuco, 0. S., and Dingkuhn, M. (1986a). Responses of seven diverse rice cultivars to water deficits. 11. Osmotic adjustment, leaf elasticity, leaf extension, leaf death, stornatal conductance and photosynthesis. Field Crops Res. 13, 273-256. Turner, N . C., Hearn, A. B., Begg, J. E., and Constable, G. A. (1986b). Cotton (Gossypium hirsutum L.): Physiological and morphological responses to water deficits and their relationship to yield. Field Crops Res. 14, 153-170. Turner, N. C., Tennant, D., Hamblin, A . P.. Henson, 1. E., Jensen, C. R., and Perry, M. W. (1987). Influence of nitrogen on the yield, water use and water deficits of wheat grown in a mediterranean climate. In “Agronomy 1987-Responding to Change. Proceedings of the 4th Australian Agronomy Conference” (T. G. Reeves, ed.), p. 254. Australian Society of Agronomy, Parkville. Turner, N. C., Nicolas, M. E., Hubick, K. T., and Farquhar, G. D. (1989). Evaluation of traits for the improvement of water use efficiency and harvest index. In “Drought Resistance in Cereals” (F. W. G. Baker, ed.), pp. 177-189. ICSU Press, Pans. Turner, N. C., Leport, L., French, R. J., Thomson, B. D., and Siddique, K. H. M. (1996). Adaptation of cool-season pulses to water-limited environments. In “Proceedings of the 7th Australian Agronomy Conference, Toowoomba, Queensland, January 1996,” pp. 554-557. Australian Society of Agronomy, Carlton. Van den Bulcke, M. ( 1994). Engineering stress-resistant plants through biotechnological approaches. In “Mechanisms of Plant Growth and Improved Productivity” (A. S. Basra, ed.), pp. 435-462. Dekker, New York. Van Oosterom, E. J., and Acevedo, E. (1992). Adaptation of barley (Hordeum vulgare L.) to harsh mediterranean environments. 1. Morphological traits. Euphyrica 62, I- 14. Van Volkenburgh, E., and Davies, W. J. (1983). Inhibition of light stimulated leaf expansion by abscisic acid. J . Exp. Bof. 34, 835-845. Virgona, 3. M., Hubick, K. T., Rawson, H. M., Farquhar, G. D., and Downes, R. W. (1990). Genetic variation in transpiration efficiency, carbon isotope discrimination and carbon allocation during early growth in sunflower. Aust. J. Planr Physiol. 17, 207-214. Vos, J., and Groenwold, J. (1989). Genetic differences in water-use efficiency, stornatal conductance and carbon isotope fractionation in potato. Potam Res. 32, 113-121. Waddington, S. R., Ransom, J. K.,Osmanzai, M., and Saunders, D. A. (1986). Improvement in the yield potential of bread wheat adapted to north-west Mexico. Crop Sci. 26, 699-703. Whan, B. R., Carlton, G. P., and Anderson, W. K. (1991a). Potential for increasing early vigour and total biomass in spring wheat. I. Identification of genetic improvements. Aust. J . Agric. Res. 42, 347-361. Whan, B. R., Anderson, W. K., Gilmour, R. F., Regan, K. L., and Turner, N. C. (I991b). A role for physiology in breeding for improved wheat yield under drought stress. In “Physiology-Breeding of Winter Cereals for Stressed Mediterranean Environments” (E. Acevedo, A. P. Conesa, P. Monneveux, and J. P. Srivastava, eds.), Les Colloques No. 55, pp. 179-194. INRA, Paris. Whan, B. R., Carlton, G. P., Siddique, K. H. M., Regan. K. L., Turner, N. C.. and Anderson, W. K. (1993). Integration of breeding and physiology: Lessons from a water-limited environment. In “International Crop Science I” (D. R. Buxton, R. Shibles, R. A. Forsberg, B. L. Blad, K. H. Asay, G. M. Paulsen, and R. F. Wilson, eds.), pp. 607-614. Crop Science Society of America, Madison, WI. White, J. W. (1993). Implications of carbon isotope discrimination studies for breeding common bean under water deficits. In “Stable Isotopes and Plant Carbon-Water Relations” (I. R. Ehleringer, A. E. Hall, and G. D. Farquhar, eds.), pp. 387-398. Academic Press, San Diego.

338

NEIL C. TURNER

Wolf, O., Jeschke, W. D., and Hartung, W. (1990). Long distant transport of abscisic acid in NaCItreated intact plants of Lupinus albus. J. Exp. Bot. 41, 593-600. Wong, S. C., Cowan, 1. R., and Farquhar, G. D. (1979). Stomata1 conductance correlates with photosynthetic capacity. Nature (London) 282, 424-426. Wright, G. C., Hubick, K. T.,and Farquhar, G. D. (1988). Discrimination in carbon isotopes of leaves correlates with water-use efficiency of field-grown peanut cultivars. Aust. J. Plant Physiol. 15, 815-825. Wright, G. C., Hubick, K. T., Farquhar, G. D., and Rao, R. C. N. (1993). Genetic and environmental variation in transpiration efficiency and its correlation with carbon isotope discrimination and specific leaf area in peanut. In “Stable Isotopes and Plant Carbon-Water Relations” (J. R. Ehleringer, A. E. Hall, and G. D.Farquhar, eds.), pp. 247-267. Academic Press, San Diego. Zeevaart, I. A. D. (1977). Sites of ABA synthesis and metabolism in Ricinus communis L. Plant Physiol. 59, 788-791. Zeevaart, J. A. D., and Boyer, G. L. (1984). Accumulation and transport of abscisic acid and its metabolites in Ricinus and Xanthium. Plant Physiol. 74, 934-939. Zeevaart, J. A. D., and Creelman, R. A. (1988). Metabolism and physiology of abscisic acid. Annu. Rev. Plant Physiol. 39, 439-473. Zeng, 2.-B. (1994). Precision mapping of quantitative trait loci. Generics 136, 1457-1468. Zhang, J., and Davies, W. J. (1987). Increased synthesis of ABA in partially dehydrated root tips and ABA transport from roots to leaves. J. Exp. Bot. 38, 2015-2023. Zhang, J., and Davies, W. J. (1989). Abscisic acid produced in dehydrating roots may enable the plant to measure the water status of the soil. Plant Cell Environ. 12, 73-81. Zhang, J., Schurr, U., and Davies, W. J. (1987). Control of stomata1 behaviour by abscisic acid which apparently originates in roots. J . Exp. Bot. 38, 1174-1181. Zimmemann, U., Rade, H., and Steudle, E. (1969). Kontinuierliche Druckmessung in Pflanzenzellen. Naturwissenschafren 56, 634. Zimmerman, U., Benkert, R., Schneider, H., Rygol, J., Zhu, J. I.,and Zimmermann, G. (1993). Xylem pressure and transport in higher plants and tall trees. I n “Water Deficits: Plant Responses from Cell to Community” (J. A. C. Smith and H. Griffiths, eds.), pp. 87-108. Bios Scientific, Oxford, UK.

Index A

Applications geostatistics to soil survey problems, 241Aberrations. karyological, effect on variation, 29 1 211-212 silicate slag, 183 Abscisic acid silicon effect on growth, 300-301 foliar, 187 as stress signal, 297-300 rate and time, 160, 166-168 and yield under drought, 323-324 Approximation, first-order, to sorption kinetics, Acetate buffer, Si extracted by, 180-181 15-21 Acetolactate synthase Ash, black-gray, rice hulls, 164-166, 173-174 herbicides, 68-69 Assimilation, preanthesis, and yields under wainhibitor resistance genes, 78-79 ter deficit, 305-307 Acetyl-coenzyme A carboxylase, herbicides, Attachment, role in biodegradation of sorbed 67-68 chemicals, 37-38 S-Adenosylhomocysteine, ratio with S-adenosyl- Australia methionine, 213 adoption of IWM strategies, 82-83 S-Adenosylmethionine, in DNA methylation, herbicide resistance, 73-74 213 soil water loss, 311-312 Advection, coupled with sorption and biodewheat yields, 303-305 gradation, 28-29 Auxin, herbicides disrupting, 69-70 Age, tissue culture, effect on variability, 207208 B Aggregates geometry, I 10- 1 I3 Bacteria, biodegradation kinetics, 18-20 size, decreasing, 43-44 Barley, early vigor and grain yield, 317 spatial distribution at aggregate scale, 39-40 Batch conditions, coupled process models unAging der, 14-28 and bound residue formation, 11-12 Batch systems, extrapolation of microbial pacontinuum, 47 rameters from, 29-30 relation to bioavailability, 41-42 Bedding, rice straw and hull, 163-161 Agrochemical research, new herbicides, 75-76 Benefits Agronomic benefits, silicon management, 165silicon management, 165-177, 188 171, 177-179 transgenic herbicide-resistant crops, 79-8 I Agronomic performance, regenerants, 222-223 Bioavailability Alculigenes, strain NP-ALK, 19-20, 36-37 reduced, regulatory implications, 46 Algorithm silicon in silicate slag, 160-161 kriging, 269-270 temporal changes in, 41-42 simulation, 281-286 Biodegradation Aluminum toxicity, in acid soil, silicon amendcoupled with sorption, controlling factors, ment, 176-177 32-42 Aminomethylphosphonic acid, glyphosate desorbed chemicals, 12-31 graded to, 76 Bioremediation, sorbed chemicals, 42-46

339

3 40

INDEX

Boundary conditions, at mobile-immobile interface, 108-109 Bound residue formation and aging, 11-12 effect on biodegradation rates, 41 Breakthrough curves Ca and Mg, 130 from individual cells, 96- 107 large- and small-pulse, 117-121 Breeding for improved drought resistance, 3 15-324 role in yield improvement, 303-305 Bromide, breakthrough, 102-103, 107 Bromoxynil, resistance genes, 78

Cross-pollination, transgenic crops with weeds,

81 Cross-validation p-field, 266 simple cokriging, 262-264 Crushing, in decrease of aggregate size, 43-

44 Cultivars canola OAC Triton, 75 from tissue culture-derived somaclonal variation, 218-220 Cuticle, rice, silicon double layer, 153 Cycling, rice plant silicon, 163-165

D C Capillary bundle models, 122- 124 Carbofuran, soluble pools, 41 Carbon flow in soil, 30-31 isotope discrimination, 308-31 1, 320-321 organic, sequestration, 2 partitioning, 318-3 19 preanthesis, 306 Cell lines, mutation-specific, selection, 216-

217 Cereals, global herbicide use, 61 Chemicals, sorbed biodegradation, 12-3 I bioremediation, 42-48 Chimeras, regenerants in plant species, 204 Cokriging maps, EC,, 270-274 simple, 262 Colorimetry, silicon in rice plant tissue, 183 Column studies, coupling advection, sorption, and biodegradation, 28-29 Composting, rice straw, 163-164 Convective-dispersive equations, 101- 103,

117-119 Convective flow, in diffusion model, 108-

109 Cotton, global herbicide use, 63 Crops improvement, and tissue culture-induced variation, 201-223 major, and herbicide markets, 57-63 transgenic herbicide-resistant, 74-8 1 Crop water relations, 293-325

Data set, salinity, Broadview, 243-244 Deficiency, silicon, in rice, 154 Degradation abilities of organisms, 36-38 kinetics, naphthalene, for bacterial strains,

19-20 sorbed chemical, analogy with carbon flow in soil, 30-31 Desorption, kinetics, in stirred reactor, 26 Deterministic models, accounting for physical nonequilibrium, 138-139 3,4-Dichloroaniline, biodegradation rate, 35-36 Dichlorophenol, mineralization, 31 Differentiation, soil type, 248-250 Diffusion effect on coupled biodegradation and sorption, 33-34 equations, 8- I 1 model, dispersion coefficient in, 110-1 12 radial, combined with biodegradation kinetics, 21-27 Dimensionless groups, 23-24 Dinitroaniline, biodegradation kinetics, 17- 18 Diseases, rice, suppression, 17I- 174 Dispersion coefficient, in diffusion model, 110-

112 Dissolution silicon in soil, kinetics, 156-158 from solids and nonaqueous-phase liquids,

27-28 Distribution EC,, mapping, 254-268 microbes, 39-40 probability, in simulation algorithm, 28 1-284

341 rate constant, shape, 8 spatial filtering structures from, 269-276 at pore and aggregate scale, 39-40 xyloglucan endotransglyosylase, 302 DNA methylation, effect on variation, 212-214 Drought, resistance, 314-324

E Electrical conductivity data, exploratory data analysis, 245-254 distribution mapping, 254-268 Electromagnetic data EM,, effect on EC, maps, 267-269 exploratory data analysis, 248-254 Embryogenesis, somatic, 204-206 Emulsifiers, produced by microorganisms, 38 Environments terminal stress, 315 water-limited, 31 1-312 Epigenetic variation, transient, 203 EPSP synthase, glyphosate-resistant, 76-77 Equations convective-dispersive, 101-103, 117-1 19 diffusion, 8- I 1 first-order mass-transfer, 7-8, I 11-1 12 Michaelis-Menten, 12-15, 22, 26 Monod, 12-14. 22, 27 Equilibrium ion-exchange model, 128-131 Equilibrium partition coefficient, chemical, 4-6 Equilibrium sorption, coupled process models,

Field applications, physical nonequilibrium models, 138-140 Field screening, somaclonal variation. 221-222 Field selection, variants, 217, 223 Filtering, structures, from spatial distribution, 269-276 First-order mass-transfer equations, 7-8, 1 1 1I12 Fitness penalty, dominant allele with, 66 Fractionation, chemical, caveat, 21 Freundlich equilibrium model, 126- 127 Fruits, global herbicide use, 63

G

Gamma function model, 20 Gaussian random function model, 256-257 Genes acetolactate synthase inhibitor-resistance, 7879 bar, 77-78 drought-resistance, 322-324 glyphosate-resistant, 76-77 PsbA, triazine-resistant, 75 resistance, degree of dominance, 65-66 Genotypes culture-regenerated, 222 earlier flowering, 316-317 rice, selection for resistance, 174 and somaclonal variation. 209-210 transpiration efficiency, 320-32 1 wheat 14-15 grain growth, 319-320 Evaporation, soil, in Australia, 31 1-312 and grain yield, 303-304 Evolution, herbicide resistance, and selection for transpiration efficiency, 309-31 1 pressure, 64-67 Geostatistical analysis, soil salinity data set, Explant source, generation of somaclonal varia24 I -29 1 tion, 207 Global sales, herbicides, and crop production, Exploratory data analysis, E C , and EM data, 58-63 245-254 Glufosinate, resistance genes, 77-78 Extraction, silicon Glyphosate, resistance genes, 76-77, 80-81 acetate buffer method, 180-181 Grain, growth, and genotype selection, 319-320 HCI method, 185 Grayscale plots EC, and EM data, 245-254 F prior probability values, 282-284 Growth Fertilization. silicon, 159-161, 166-168, 179abscisic acid effects, 300-301 I85 grain, and genotype selection, 319-320 Fertilizers, interactions with rice silicon, 168rice plant, and silicon management, 165- I66 171 and turgor, 301-302

3 42

INDEX H

Hamaker's model, 16-19 Hard data, core EC, values as, 243-244 Herbicide Resistance Action Committee, 74 Herbicides acetolactate synthase, 68-69 acetyl-coenzyme A carboxylase, 67-68 auxin- and photosystem I-disrupting, 69-70 markets, and major crops, 57-63 photosystem 11, 69 resistance, management, 7 1-83 Heritability, transpiration efficiency, 321 Heritable variation, stability, 203-204 Heterogeneity physical, during solute transport, 96 spatial, soil, 48 Histograms EC, and EM data, 249-251 target, 267 Histosols, amendment with silicate, 176 Hormones, effect on tissue culture-induced variability, 208-209 Hull, rice, silicon source, 161-165 Hydraulic conductance, effect on yield, 320 Hydraulic property, application of three poreregion model, 124-125 Hydrophobic compounds, sorption thermodynamics, 4-5 Hydrophobicity, microorganisms, effect on attachment, 38

I Imaging, stochastic, 280-286, 290 Improvement crops, and tissue culture-induced variation, 201-223 transpiration efficiency, 309-3 12 yield, 302-305, 315-316 Incubation method, flooded soil, 181-182 Infrared thermometry, 294-295 Inoculation, with microorganisms, 43 Integrated pest management, 179 Integrated weed management, 71-84 Iron toxicity, in acid soil, silicon amendment, 176-177 Irrigation water, source of silicon, 158, 165 Isotherms, linear and nonlinear, 5-6 IWM, see Integrated weed management

J Juvenility vigor trait, 203

K Karickhoffs model, 7-8, 16 Karyological aberrations, effect on variation, 21 1-212 Kernel size, reduction, 306 Kinetics biodegradation combined with radial diffusion, 21-27 dinitroaniline, 17- 18 naphthalene degradation, for bacterial strains, 19-20, 36-37 silicon dissolution in soil, 156-158 sorption first-order approximation to, 15-21 processes and models, 6- I 1 Kinetic two-site model, 127-128 Kriging algorithm, 269-270 simple, 259-262

Late embryogenesis abundant proteins, 322-323 Leaf transpiration efficiency, 307-3 12 turgor, control of leaf expansion, 296-297 Lodging, rice plants, silicon effect, 175-176 Lolium rigidurn, herbicide resistance, 82-83

M Mahalanobis distance, 277-279 Maize global herbicide use, 60-61 imidazolinone-resistant,78-79 RFLP marker mapping, 324 Management herbicide resistance, 57-84 for improved crop water use efficiency, 3 11314 integrated pest, for rice, 179 integrated weed, 71-84 silicon, and sustainable rice production, 15 II89

INDEX Manganese toxicity, in acid soil, silicon amendment, 176- I77 Mapping EC, distribution, 254-268 RFLP marker, for maize, 324 Mass transfer coefficient, 23-24 as rate-limiting step, 2-3 in sorbing porous media, 106-107 Measurement, water deficits, 294-296 Metabolism rate, effect on coupled biodegradation and sorption, 34-38 Methane emission, from lowland rice fields, 186 Methylation, DNA, effect on variation, 212214 Michaelis-Menten equations describing biodegradation, 22, 26 for non-growth-linked metabolism, 12- 15 Microbes distribution, 39-40 population density, 34-36 processes, and soil structure, 26-27 rate parameters, extrapolation, 29-30 Mineralization carbofuran, 4 1 dichlorophenol, 3 I dimensionless, 25 Mixing, in decrease of aggregate size, 43-44 Mixtures, herbicides, to minimize resistance, 72-74 Mobile-immobile models modified, 135-138 second-order, I3 1 - 134 two-region, 101-1 16 Mode of action, herbicides classification, 59-60 new, 8 1-84 significance, 71-74 Models capillary bundle, 122- 124 coupled physical and chemical nonequilibrium. 126-138 distributed reactivity, 5-6 Gaussian random function, 256-257 mathematical, sorptioddesorption kinetics, 7-11 mobile-immobile modified, 135-138 second-order, 131-134 two-region, 101-1 16

343

multiple-Row domain, 124-126 quantitative conceptual, 47 radial diffusion-biodegradation, 2 1-27 two-compartment, first-order approximation, 15-21 two-Row domain, 116-121 variogram, 269-270 Moisture, soil, effect on biodegradation, 40-41 Molecular techniques, in plant breeding, 322324 Monod equations describing biodegradation, 22, 27 for growth-linked metabolism, 12- 14 Monosilicic acid, polymerization, 153 Multiple-flow domain models, 124-126 Multiple peaks, breakthrough curves, 99 Multiple resistance, herbicide, 70-71, 83

N Naphthalene, degradation kinetics, for bacterial strains, 19-20, 36-37 Nitrogen effect on water use efficiency, 3 12 fertilizers, effect on rice silicon, 168-170 preanthesis, 306 Nonaqueous-phase liquids, dissolution from, 27-28 Nonequilibrium, physical, in modeling of solute transport, 95-144 Nugget effect, in variogram model, 270 Nutrition, silicon, in rice, 152-155, 172-173 Nuts. global herbicide use, 63

0 Oilseed rape, global herbicide use, 62 Oily waste, contaminated soil, 44 Organisms, degradation abilities, 36-38, 47-48 Organogenesis, root, 204-206 Osmotic adjustment, effect on yield, 318

P Paraquat, resistance, 70 Partition coefficient, sorption, 23-24, 32-33 Partitioning, carbon, 318-319 Pesticides biodegradation rates, 17-18, 20 loss of chemical identity, 31 reduction with silicon management, 187

344

INDEX

Pests, rice integrated management, 179 suppression, 173-175 p-field approach estimates, 262-266 in mapping EC, distribution, 257 pH, silicon adsorption dependent on, 156 Phenanthrene, biodegradation rate, 35 Phenology, and yield improvement, 315-316 Phenotype, variation, in somaclones, 217-221 Phosphate fertilizers, effect on rice silicon, 170 silicon-amended sources, I87 Photosynthetic efficiency, and transpiration efficiency, 309-3 1 I Photosystem I1 herbicides, 69 Physical nonequilibrium, in modeling of solute transport, 95-144 Physical trapping aging resulting from, 11-12 pollutants, 31 Plants regeneration competence, 208-210 from somatic cells, 204-206 rice lodging, 175- 176 silicon recycling, 161-165 testing for silicon fertilization, 182-184 Ploidy, effect on tissue culture-induced variability, 210-21 1 Pollutants different pools in soil, 21 reduced bioavailability, 2 Pools carbofuran, soluble, 41 extractable fractions, 21 labile and resistant, 16-17 Population density, microbial, 34-36 Pores intraaggregate, and diffusion equations, LO macro and micro, definitions, 113-1 15 size, and water flow geometry, 141 spatial distribution at pore scale, 39-40 Potassium, fertilizers, effect on rice silicon, 171 Preanthesis carbon, 306 water use, 313-314, 317-318 Preanthesis assimilates, reserves, 305-307 Pressure head, soil-water, 98-99

Pressure probe, for measurement of turgor, 29S296 Probability distribution. in simulation algorithm, 281-284 Probability maps, in assessment of spatial uncertainty, 287-289 Productivity, problem soils, silicon effect, 176177 Proteins, late embryogenesis abundant, 322-323 Pseudomonas purida, strain ATCC 17484, 1920, 36-37

Q Quantile maps, in assessment of spatial uncertainty, 287-289 Quantitative trait loci, analysis, 323-324

R Radial diffusion, coupling with sorption and biodegradation, 2 1-27 Ranking, stochastic images, 290 Rate constant distribution, shape, 8 Rate-limiting step in bioremediation, 45-46 mass transfer as, 2-3 Rate parameters, microbial, extrapolation, 2930 Recycling, plant silicon, 161- I65 Regeneration and phenotypic variation in somaclones, 221 plant competence, 208-210 from somatic cells, 204-206 Regulatory implications, reduced bioavailability, 46 Research agrochemical. new herbicides, 75-76 on silicon nutrition, 185-187 Reserves, preanthesis assimilates, 305-307 Resistance drought, 314-324 herbicide evolution, 64-67 management, 71-83 specific groups, 67-70 to stresses in rice, induction, 171-176 Restriction fragment-length polymorphism, 323324

INDEX Retardation factor, in diffusion model, 1 I 1 Rice global herbicide use, 61 hull and straw, silicon recycling, 161-165 plant testing for silicon fertilization, 182-184 silicon-accumulating cultivars, 187 silicon management agronomic essentiality, 177- I79 benefits, 165-171 silicon nutrition, 152- 155 stresses, induced resistance, 171-176 Richards-type model, and non-Richards-type model, 101 Risks, transgenic herbicide-resistant crops, 7981 Roots, organogenesis, 204-206 Root-to-shoot message, abscisic acid as, 298300

Rotations, herbicides, 72-74 RZWQM model. 139

S Salinity, Broadview data set, 243-244 Sampling cases, in mapping EC, distribution, 258-259 Scattergram, regional components of EM, and EC,, 271-277 Second-order two-site model, modified, 134I35 Seedbank, soil, 66 Selection genotype grain growth, 319-320 resistance, 174 in vitro. somaclonal variation, 215-222 pressure. and evolution of herbicide resistance, 64-67 Sensing, water deficits, 296-302 Sequestration, organic carbon, 2 Short-scale structure, on cokriging maps, 2702 74 Silicon fertilization, 159- 161, 179- 185 management agronomic essentiality, 177- 179 benefits, 165- 177 nutrition. in rice. 152-155 in soil and water, 155-158 Simulation, algorithm. 281-286

345

Slags chemical composition. 159 silicate applications, 183 silicon bioavailability, 160- 161 as silicon source, 186 Smoothing effect correction, 280-281 kriging, 259-262 Sodium chloride, as selective agent, 216 Soils carbon flow, analogy with sorbed chemical degradation, 30-3 I drying, effect on abscisic acid levels, 299 flooded, 162-163, 181-182 moisture, effect on biodegradation, 40-4 I problem, increased productivity, 176- 177 salinity data set, geostatistical analysis, 24129 1 silicon in, 155-158 solute transport, nonequilibrium modeling, 95- I44 spatial heterogeneity, 48 testing for silicon fertilization, 179- 182 type, differentiation, 248-250 Solids, dissolution from, 27-28 Solute transport, in soils, nonequilibrium modeling, 95-144 Somaclonal variation causes and range, 202-206 methodological basis, 207-2 14 rate and in v i m selection, 214-222 Sorption coupling with advection and biodegradation, 28-29 biodegradation, 32-42 radial diffusion, 21-27 equilibrium, coupled process models, 14- 15 kinetics, 6-12, 15-21 thermodynamics, 3-6 Sowing, earlier, effect on yield. 313 Soybeans, global herbicide use, 62 Spatial analysis, and variogram analysis, 250254 Spatial clustering, analysis, 276-280 Spatial distribution filtering structures from, 269-276 at pore and aggregate scale, 39-40 xyloglucan endotransglyosylase, 302 Spatial uncertainty, assessment, 287-289

3 46

INDEX

Stochastic imaging, 280-286, 290 Stochastic models, accounting for physical nonequilibrium, 138-139 Stomata1 conductance induced by abscisic acid, 297-300 and transpiration efficiency, 309-31 I Stoves, rice hull-fired, 165 Straw, rice, silicon source, 161-165 Stresses biotic and abiotic, in rice, 171-176 environmental, induced variation, 2 14 signaled by abscisic acid, 297-300 Substrate, availiability, and microbial physiology, 37-38 Sugar beet, global herbicide use, 62 Surfactants, addition, in bioremediation, 44

T Tension, water, and estimation of macroporosity, 114-116 TETrans model, 139-140 Thermodynamics, sorption, 3-6 Thermometry, infrared, 294-295 Three-parameter mixing model, 103- 106 Time, dependence of first-order transfer coefficient, 112-113 Tissue culture, somaclonal variation induced by, 20 1-223 Toxicities, acid soils with, silicon amendment, 176-177 Traditional similarity measure, 277 Transgenic crops, herbicide-resistant, 74-8 I Transpiration efficiency genotypic variation, 320-321 leaf, 307-3 12 rate, rice plants, 175 Transport processes, sorption kinetics, 6-7 Transposable elements, effect on variation, 21 2 Trapping, physical aging resulting from, 1 I- 12 pollutants, 31 Triazine, resistant canola cultivar, 75 Tbrbulent flow, in macropores, 120-121 Turgor control of leaf expansion, 296-297 and growth, 301-302 Two-flow domain models, 116-121

U Uscore transform, simple cokriging, 272, 274, 276

V Vapor pressure deficit, affecting transpiration efficiency, 308-312 Variation genotypic, in transpiration efficiency, 320321 somaclonal causes and range, 202-206 methodological basis, 207-214 rate and in v i m selection, 214-222 Variogram analysis, and spatial analysis, 250254 Variogram model, 267-270 Vegetables, global herbicide use, 63 Velocity, dependence of first-order transfer coefficient, 113-116 Vigor early, yield related to, 316-318 juvenility vigor trait, 203

W Waste oily, contaminated soil, 44 slag, as silicon fertilizers, 184-185 Water silicon in, 155-158 tension, and estimation of macroporosity, 114-116 Water deficits associated proteins, 322-324 measurement and sensing, 294-302 and yield, 302-307 Water phase, mobile and immobile, 106-1 16 Water use efficiency, and crop water relations, 307-314 Weeds herbicide-resistant, threat to crop production, 64-7 1 integrated management, 7 1-84 Wheat early vigor and grain yield, 317

INDEX genotypes grain growth, 319-320 grain yield, 303-304 transpiration efficiency, 309-3 1 I Wounding, primary embryos, 206

X Xylem vessels, pressure probe technique, 295296 Xyloglucan endotransglyosylase, spatial distribution, 302

347 Y

Yield improvement, and phenology, 315-316 osmotic adjustment effect, 318 rice declining trends, I78 silicon management, 166-168, 187 and water deficits, 302-307