Advances in Agronomy, Volume 75

Agronomy DVANCES I N VOLUME 75 Advisory Board Martin Alexander Ronald Phillips Cornell University University of...

Author: Donald L. Sparks

64 downloads 1298 Views 3MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!

Report copyright / DMCA form

DOWNLOAD PDF

Agronomy

DVANCES I N

VOLUME

75

Advisory Board Martin Alexander

Ronald Phillips

Cornell University

University of Minnesota

Kenneth J. Frey

Kate M. Scow

Iowa State University

University of California, Davis

Larry P. Wilding Texas A&M University

Prepared in cooperation with the American Society of Agronomy Monographs Committee Jerry M. Bigham Jerry L. Hatfield David M. Kral Linda S. Lee

Diane E. Stott, Chairman David Miller Matthew J. Morra John E. Rechcigl Donald C. Reicosky

Wayne F. Robarge Dennis E. Rolston Richard Shibles Jeffrey Volenec

Agronomy

DVANCES IN

VOLUME

75

Edited by

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

San Diego San Francisco New York Boston

London

Sydney

Tokyo

This book is printed on acid-free paper.

∞

C 2002 by ACADEMIC PRESS Copyright

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. The appearance of the code at the bottom of the first page of a chapter in this book indicates the Publisher’s consent that copies of the chapter may be made for personal or internal use of specific clients. This consent is given on the condition, however, that the copier pay the stated per copy fee through the Copyright Clearance Center, Inc. (222 Rosewood Drive, Danvers, Massachusetts 01923), for copying beyond that permitted by Sections 107 or 108 of the U.S. Copyright Law. This consent does not extend to other kinds of copying, such as copying for general distribution, for advertising or promotional purposes, for creating new collective works, or for resale. Copy fees for pre-2002 chapters are as shown on the title pages. If no fee code appears on the title page, the copy fee is the same as for current chapters. 0065-2113/2002 $35.00 Explicit permission from Academic Press is not required to reproduce a maximum of two figures or tables from an Academic Press chapter in another scientific or research publication provided that the material has not been credited to another source and that full credit to the Academic Press chapter is given.

Academic Press An Elsevier Science Imprint 525 B Street, Suite 1900, San Diego, California 92101-4495, USA http://www.academicpress.com

Academic Press Harcourt Place, 32 Jamestown Road, London NW1 7BY, UK http://www.academicpress.com International Standard Book Number: 0-12-000793-2 PRINTED IN THE UNITED STATES OF AMERICA 02 03 04 05 06 07 SB 9 8 7 6

5

4

3

2

1

Contents CONTRIBUTORS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . PREFACE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vii ix

PHYTOREMEDIATION OF METALS, METALLOIDS, AND RADIONUCLIDES S. P. McGrath, F. J. Zhao, and E. Lombi I. II. III. IV. V.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Phytoextraction Using Hyperaccumulator Plants . . . . . . . . . . . . . . . . . . . . . . Chemically Enhanced Phytoextraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Phytovolatilization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary and Future Directions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

2 6 26 39 44 46

THE SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES: A NOVEL UNDERSTANDING OF HUMUS CHEMISTRY AND IMPLICATIONS IN SOIL SCIENCE Alessandro Piccolo I. II. III. IV. V. VI.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Paradigmatic View of Humus Chemistry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Dilemma of the Conformational Structure of Humic Substances Size-Exclusion Chromatography of Humic Substances . . . . . . . . . . . . . . . . Supramolecular Associations of Self-Assembling Humic Molecules . . Chemical and Spectroscopic Evidence of Supramolecular Associations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . VII. Turning Loose Humic Superstructures into Stable Polymers . . . . . . . . . VIII. Role of Hydrophobic Humic Superstructures in Soil . . . . . . . . . . . . . . . . . . IX. Future Perspectives in Research and Technology . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

58 59 63 66 75 105 109 115 125 126

WATER-SAVING AGRICULTURE IN CHINA: AN OVERVIEW Huixiao Wang, Changming Liu, and Lu Zhang I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . II. Water-Saving Agriculture as a System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v

136 139

vi III. IV. V. VI. VII. VIII.

CONTENTS Water-Use Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A Rationale for the Use of Water Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . Water-Saving Engineering Measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Water-Saving Agronomic Practices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Water-Saving Management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

140 143 150 154 163 167 168

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES E. Ben-Dor I. II. III. IV. V. VI. VII. VIII.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Principles of Quantitative Remote Sensing of Soils . . . . . . . . . . . . . . . . . . . . Mechanisms of the Soil–Radiation Interactions. . . . . . . . . . . . . . . . . . . . . . . . . Problems in Quantitative Remote Sensing of Soil . . . . . . . . . . . . . . . . . . . . . . Parameters Affecting the Remote Sensing of Soil . . . . . . . . . . . . . . . . . . . . . . High-Spectral-Resolution Sensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . General Analytical Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Closing Remarks and Recent Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

174 183 187 206 218 222 225 228 231

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

245

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

E. BEN-DOR (173), The Remote Sensing and GIS Laboratory, Department of Geography and the Human Environment, Tel-Aviv University, Ramat Aviv, Tel-Aviv 69978, Israel E. LOMBI (1), Agriculture and Environment Division, IACR-Rothamsted, Harpenden, Herts AL5 2JQ, United Kingdom C. LIU (135), Institute of Geographic Sciences and Natural Resources Research, The Chinese Academy of Sciences, Beijing 100101, China S. P. McGRATH (1), Agriculture and Environment Division, IACR-Rothamsted, Harpenden, Herts AL5 2JQ, United Kingdom A. PICCOLO (57), Dipartimento di Scienze Chimico-Agrarie, Universit`a Degli Studi Di Napoli “Federico II,” 80055 Portici, Italy Via Universita 100, Naples, Italy H. WANG (135), State Key Laboratory of Water Environment Simulation, Key Laboratory for Water and Sediment Sciences, Ministry of Education, Beijing Normal University, Beijing 100875, China L. ZHANG (135), CSIRO Land and Water, Canberra Laboratory, P.O. Box 1666, Canberra, ACJ 2601, Australia F. J. ZHAO (1), Agriculture and Environment Divison, IACR-Rothamsted, Harpenden, Herts AL5 2JQ, United Kingdom

vii

This Page Intentionally Left Blank

Preface Volume 75 contains four outstanding reviews dealing with phytoremediation, issues related to water use in China, humic substances, and remote sensing. Chapter 1 is an extensive review on phytoremediation of metals, metalloids, and radionuclides including discussion on phytoextraction technologies, hyperaccumulator plants, and chemically induced phytoextraction and phytovolatilization. Chapter 2 covers the conservation and use of water in Chinese agriculture including engineering, economic, and agronomic aspects and considerations. Chapter 3 presents advances in understanding the structure of humic substances, particularly the concept of a supramolecular structure. Analytical and molecular scale evidence for this latter structure are presented as well as discussions on the role of humic superstructures in soils. Chapter 4 presented frontiers in quantitative remote sensing of soil properties including principles, methods, mechanisms, and limitations. I thank the authors for their ﬁrst-rate reviews. DONALD L. SPARKS

ix

This Page Intentionally Left Blank

PHYTOREMEDIATION OF METALS, METALLOIDS, AND RADIONUCLIDES S. P. McGrath, F. J. Zhao, and E. Lombi Agriculture and the Environment Division IACR-Rothamsted, Harpenden, Herts AL5 2JQ, United Kingdom

I. Introduction A. Risks of Metals and Metalloids in Soils B. The Need for Cleanup of Contaminated Soils C. Phytoextraction, Phytomining, and Removal Technologies II. Phytoextraction Using Hyperaccumulator Plants A. Metal Hyperaccumulators B. Phytoextraction Using Hyperaccumulator Plants C. Mechanisms of Metal Hyperaccumulation III. Chemically Enhanced Phytoextraction A. Potential Applications B. Chemically Enhanced Phytoextraction of Lead C. Chemically Enhanced Phytoextraction of Other Heavy Metals D. Chemically Enhanced Phytoextraction of Radionuclides E. Chemically Enhanced Phytomining F. Chemically Enhanced Phytoextraction versus Natural Hyperaccumulation G. Possible Concern Relating to the Use of Chelating Agents IV. Phytovolatilization A. Selenium B. Mercury V. Summary and Future Directions References

Phytoremediation is a developing technology that can potentially address the problems of contaminated agricultural land or more intensely polluted areas affected by urban or industrial activities. Three main strategies currently exist to phytoextract inorganic substances from soils using plants: (1) use of natural hyperaccumulators; (2) enhancement of element uptake of high biomass species by chemical additions to soil and plants; and (3) phytovolatilization of elements, which often involves alteration of their chemical form within the plant prior to volatilization to the atmosphere. Concentrating on the techniques that potentially remove inorganic pollutants such as Ni, Zn, Cd, Cu, Co, Pb, Hg, As, Se, and radionuclides, we review the progress in the understanding of the processes involved and 1 Advances in Agronomy, Volume 75 C 2002 by Academic Press. All rights of reproduction in any form reserved. Copyright 0065-2113/02 $35.00

2

S. P. McGRATH ET AL. the development of the technology. This includes the advances made in the study of the physiology and biochemistry of metal uptake, transport and sequestration by hyperaccumulator plants, as well as the investigation of the processes occurring in soil and plant systems subject to the chemical enhancement approach. Enough work has been carried out in the last decade to allow some assessment of the situations and elements in which phytoremediation is likely to be most successful. However, we also identify where there is lack of knowledge. Finally, the likely future directions C 2002 Academic Press. for research and application are discussed.

I. INTRODUCTION Phytoremediation can be loosely deﬁned as the use of plants to improve the environment. Obviously this is an enormous subject and here we will concentrate on the phytoremediation of metals, metalloids, and radionuclides. Phytoremediation of organic compounds in soil and water is a related and rapidly expanding area, which is covered elsewhere (Kruger et al., 1997; Salt et al., 1998; Wenzel et al., 1999). It is very appropriate to review this subject at this time because it was around 1990 that the ﬁrst ﬁeld experiments began examining phtyoremediation of metals and Se (reported in Ba˜nuelos et al., 1993; McGrath et al., 1993); and now a decade has passed. We will examine the different strategies that have evolved for phytoremediation and the progress that has been made on the physiology of metal accumulation. On a more practical level, the attempts at ﬁeld application will be evaluated, and the likely future directions of the science and technology will be discussed.

A. RISKS OF METALS AND METALLOIDS IN SOILS Metals and metalloids such as As and Se can pose risks when they build up in soils due to many forms of anthropogenic inﬂuences. Some such as Zn, Cu, Mn, Ni, Se, Co, Cr, and Mo are essential for living organisms, and therefore deﬁciency situations exist either because of very low total amounts of these metals in soil or because of low bioavailability caused by soil chemical conditions. In these cases, when metals are added, there may be positive biological responses in terms of growth and health of organisms. However, these metals and those that are thought to be nonessential such as Pb, As, Hg, and Cd tend to build up in soils; and when their bioavailability becomes high, toxicity can result. These negative effects can occur in soil microbes, soil fauna, higher animals, plants, and humans. A further

PHYTOREMEDIATION

3

threat is from radionuclides such as those of U, 137Cs, 90Sr, and 3H in soil and water (Negri and Hinchman, 2000). Of course, these elements may occur at elevated concentrations quite naturally in soils and waters. In these cases there may be “effects” on biodiversity and on animal and human health. Examples would be metal-tolerant vegetation that has evolved on metal-mineralized soils (Baker and Proctor, 1990), the effects on human health due to excess Se (Yang et al., 1983), and Cd accumulation in tissues of white-tailed ptarmigan (Lagopus leucurus) in the Colorado Rocky Mountains, resulting in toxicity (Larison et al., 2000). In these cases, it may not be possible or desirable to clean up the soils, but there may be a role for plants in reducing the exposure of biota to these elements, for example, by reduced uptake and exclusion from tissues, or removing elements like Se in geogenically laden water (Ohlendorf et al., 1986; Wu et al., 1995). Indeed, where these natural hot spots occur, there may be specialized fauna and ﬂora, like metallophyte vegetation, which may be in need of preservation (Reeves and Baker, 2000). Metals and metalloids enter soils and waters due to many processes including atmospheric deposition from industrial activities or power generation; disposal of wastes such as sewage sludge, animal manures, ash, domestic and industrial wastes or by-products; irrigation and ﬂood or seepage waters and the utilization of fertilizers, lime, or agrochemicals. Radionuclides may build up in some areas due to deliberate or accidental releases related to their use for energy production or for military purposes. Unlike nitrate or chloride, many of these elements are relatively strongly retained in the surface of soils and do not readily leach, causing the accumulation that may ultimately pose a threat to humans and biota. However, under some conditions, small amounts of these elements do leach and can be an issue in waters, particularly those used for irrigation or drinking. Key examples here would be radionuclides, As, Se, and Cr (Chiou et al., 1995; Kimbrough et al., 1999; Negri and Hinchman, 2000; Ohlendorf et al., 1986). Under these conditions, phytoremediation is an important developing technology for removal of these elements from either soil or water. It has the potential to be low cost and to be applicable to large areas where other methods may be too expensive and where the concentrations of contaminants are too small for other methods to be effective or economically viable.

B. THE NEED FOR CLEANUP OF CONTAMINATED SOILS There is a long history of contamination accumulating in soils due to the practices mentioned earlier. Public and political pressure to reverse this situation and clean up areas only occurs when critical levels are reached. Leaving aside the methods of deriving critical levels for microbes, animals, plants, and humans, once these exist, they provide a direct stimulus for cleanup.

4

S. P. McGRATH ET AL.

Using various ways of deﬁning “contaminated” land, it has been estimated that in the European Union alone, there are potentially 1,400,000 contaminated sites (ETCS, 1998). Not all of these will be contaminated with metals or metalloids, but this gives an indication of the scale of the problem as it may exist worldwide. For example, trace elements are present at high concentrations at 65% of the contaminated Superfund sites for which the US EPA has signed Records of Decision (US EPA, 1997). Indeed, some areas are not included in these assessments, such as those with low-level contamination due to atmospheric deposition or to the use of chemicals in agriculture. For example, the use of phosphate sources that are contaminated with Cd for agricultural fertilizers may result in crops that contain more than the allowed concentrations of Cd in foodstuffs (Commission of the European Communities, 2001). It is unlikely that these areas are included in the previously described estimates, as they focus more on urban and industrial land. However, for the sustained practice of agriculture with inputs of fertilizers, sewage sludge, and animal manures, there may be a role for plants in removing the small excess amounts of metals such as Cd, Zn, and Cu from soils, perhaps on a long rotational basis. Use of low-Cd phosphate is already taking place, while removal of Cd from phosphate rock is still not considered economically feasible (Oosterhuis et al., 2000). The average concentration in phosphate fertilizers in Europe is still 138 mg Cd kg−1 P (ERM, 1997). In comparison, the background level of cadmium is 0.3 mg Cd kg−1 soil or less in most agricultural soils in Europe. Concentrations in soils are increasing because the inputs are not balanced by the output in terms of removal by crops and leaching out of the ploughing layer (Eriksson et al., 1996; Kofoed and Klausen, 1983). Thus it is likely that phytoremediation will be needed for continued agriculture in the future.

C. PHYTOEXTRACTION, PHYTOMINING, AND REMOVAL TECHNOLOGIES Our focus in this review is on the methods that remove metals and metalloids from soil. This can be achieved by phytoextraction or phytovolatilization, depending on the element considered. A variant of phytoextraction, which applies when the extracted elements are of high value, is phytomining. In the latter case, the aims are to derive a “bio-ore” from the burning of the plant material and to proﬁt from the energy released by combustion of the biomass and the value of the ore itself. The recycling of elements that are bioconcentrated during phytoextraction will not be discussed, and the disposal options for plant biomass will depend on the market for the elements concerned. Related technologies exist or are under development, such as phytostabilization. This is when the plants are used essentially to stabilize contaminated land or the pollutants present in soil and in so doing prevent or reduce erosion, water

PHYTOREMEDIATION

5

ﬂow, and ﬂow of pollutants. In this case, metal-tolerant species that do not take up large quantities of metals are often used. In this review, for reasons of space, we chose to focus on remediation of contaminated soils, while preventing contamination of groundwater, and not remediation of contaminated water itself. That subject is covered elsewhere (Dushenkov and Kapulnik, 2000; Terry and Ba˜nuelos, 2000). The efﬁciency of phytoextraction is ultimately the product of a simple equation: biomass × element concentration in biomass. Both factors are important, but it is easy to show that high concentrations in the above-ground material are very important. Harvesting roots or other below-ground organs is difﬁcult and prevents regrowth if the “crop” is a perennial one. The increasing yield from 2 to 20 t ha−1, which is probably a biological maximum for an annual plant or harvestable from a perennial one, has little inﬂuence on the removal rate below about 1000 mg kg−1 of an element in the plant dry matter (Fig. 1). Therefore, maximizing concentrations in the plant seems to be the obvious strategy for increasing efﬁciency, while optimizing yields by agronomic means. However, it must be kept in mind that this thinking relates to very polluted soils that require hundreds of kilograms per hectare to be removed. For elements like Cd where relatively small removals (<1 kg ha−1)

Figure 1 Modeled removal of an element from soil by crops, showing the dependence on the concentration in the biomass and the effect of yield. All amounts relate to above-ground material.

6

S. P. McGRATH ET AL.

are important, or radionuclides where small quantities are of concern, this may not be so true. But the principle still applies to Cd, i.e., the scope for maximizing Cd concentration is in the magnitude of hundreds (hyperaccumulators, see following), whereas the scope for maximizing biomass is <10). Two strategies exist for obtaining plant biomass with high metal concentrations: (1) use of natural hyperaccumulators and (2) the enhancement of uptake of metals by normally non-accumulating species by applying various chemicals that increase uptake. These are discussed in the following sections.

II. PHYTOEXTRACTION USING HYPERACCUMULATOR PLANTS A. METAL HYPERACCUMULATORS 1. Definition Based on the relationship between metal concentrations in shoot and in soil, Baker (1981) proposed that plants growing on metalliferous soils can be grouped into three types: (1) excluders, where metal concentrations in the shoot are maintained at a low level across a wide range of soil concentration, up to a critical soil value above which the mechanism breaks down and relatively unrestricted root-to-shoot transport results; (2) accumulators, where metals are concentrated in above-ground plant parts from low to high soil concentrations; and (3) indicators, where uptake and transport of metals to the shoot are regulated so that internal concentration reﬂects external levels, at least until toxicity occurs. Exclusion of metals from the shoots is by far the most common strategy employed by many metal-tolerant species. On the other hand, metal accumulation can occur in some plant species that grow mainly on metalliferous soils. Reeves and Baker (2000) traced the earliest qualitative observation, Zn accumulation in Viola calaminaria in the Zn-rich soils in the Aachen area between Germany and Belgium, to A. Braun in 1855. Some 30 years later, Baumann (1885) showed that both Viola calaminaria and Thlaspi calaminare (later called Thlaspi caerulescens) growing over the calamine deposits contained over 1% Zn (10,000 μg g−1) in the shoot dry matter. Exceedingly high accumulations of Se in Astragalus plants and of Ni in Alyssum bertolonii were discovered in the 1930s and 1940s, respectively (see Brooks, 1998). Brooks et al. (1977) ﬁrst introduced the term “hyperaccumulators” to describe plants capable of accumulating more than 1000 μg Ni g−1 on a dry leaf basis in their natural habitats. The criterion for deﬁning Co, Cu, Pb, and Se hyperaccumulation is also 1000 μg g−1 in shoot dry matter, whereas for Zn and

PHYTOREMEDIATION

7

Mn the threshold is 10,000 μg g−1, and for Cd 100 μg g−1 (Baker et al., 2000; Brooks, 1998). Although these criteria are quite arbitrary, in general the concentrations of metals in hyperaccumulator plants are about 100- to 1000-fold higher than those in normal plants growing on soils with background metal concentrations, and about 10- to 100-fold higher than most other plants growing on metal-contaminated soils. In addition to the exceedingly high accumulation of metals in the shoots, hyperaccumulator plants are also characterized by a shoot-to-root metal concentration ratio of >1, whereas non-hyperaccumulator plants generally have higher metal concentrations in roots than in shoots (Baker, 1981; Baker et al., 1994a,b; Brown et al., 1995a; Gabbrielli et al., 1990; Homer et al., 1991a; Kr¨amer et al., 1996; Shen et al., 1997; Zhao et al., 2000). A highly efﬁcient transport of metals from roots to shoots is one of the key features associated with all hyperaccumulator plants. Metal hyperaccumulation is a rare phenomenon in terrestrial higher plants. To date, about 400 plant species have been identiﬁed as metal hyperaccumulators, representing <0.2% of all angiosperms (Baker et al., 2000; Brooks, 1998). It is foreseeable that the number of hyperaccumulator plants will increase as more geobotanical surveys are carried out worldwide. On the other hand, some of the hyperaccumulator species reported earlier may not be conﬁrmed as true hyperaccumulators, but may be incorrectly identiﬁed due to contamination or analytical errors (see following). Details of different metal hyperaccumulator species and their geographical distributions have been documented elsewhere (see Baker and Brooks, 1989; Baker et al., 2000; Brooks, 1998; Reeves and Baker, 2000). The following sections give only brief descriptions of the key features of different metal hyperaccumulators. 2. Nickel Hyperaccumulators Ni hyperaccumulators are the most numerous among hyperaccumulating species of plants, with a current total number of 318 taxa distributed mainly in the tropical to warm temperature regions of the world (Baker et al., 2000; Reeves and Baker, 2000). The richness of Ni hyperaccumulator species is probably due to the widespread occurrence of Ni-rich ultramaﬁc (serpentine) soils and the long history of geobotanical studies of ultramaﬁc ﬂoras. Some of the wellknown Ni hyperaccumulators are in the genus Alyssum L. (Brassicaceae), although the most remarkable example is perhaps Sebertia acuminata (Sapotaceae), a New Caledonian tree that can grow to a height of about 10 m. Jaffr´e et al. (1976) showed that this plant produces a blue-green latex containing 11.2% Ni on a fresh weight basis (25.7% on a dry weight basis). A mature tree of Sebertia acuminata was estimated to contain 37 kg Ni (Sagner et al., 1998). The other species that has recently attracted attention is Berkheya coddii, which can

8

S. P. McGRATH ET AL.

accumulate Ni to more than 1% and is tall, fast-growing, and productive (Morrey et al., 1989). These are the attributes that are ideal for phytoremediation or phytomining (see following). 3. Zinc and Cadmium Hyperaccumulators In comparison to Ni hyperaccumulators, far fewer plant species have been reported that are able to hyperaccumulate Zn and Cd. Baker et al. (2000) listed 11 taxa of Zn hyperaccumulator plants, whereas Reeves and Baker (2000) also considered two other species (Thlaspi ochroleucum and Polycarpaea synandra), which did not reach the criterion of 10,000 μg Zn g−1 to be hyperaccumulators. The best known examples of the Zn hyperaccumulators are Thlaspi caerulescens (formerly called T. alpestre) and Arabidopsis halleri (formerly named Cardaminopsis halleri), both belonging to the Brassicaceae family. In the case of Thlaspi ochroleucum, Shen et al. (1997) showed that its Zn accumulation and tolerance are considerably lower than those in T. caerulescens in hydroponic cultures. T. ochroleucum also accumulates more Zn in roots than in shoots, thus behaving in that sense rather like a non-hyperaccumulator. For Cd, T. caerulescens is the only known hyperaccumulator (Ernst, 1974; Reeves and Bakers, 2000), although recent hydroponic experiments showed that A. halleri is capable of accumulating >1000 μg Cd g−1 in the shoots without suffering from phytotoxicity (K¨upper et al., 2000), and for this reason may be classiﬁed as a Cd hyperaccumulator. Whether A. halleri accumulates over 100 μg Cd g−1 in any of its natural habitats (the criterion for deﬁning Cd hyperaccumulation) is probably determined by the Cd concentration in the soil. 4. Copper and Cobalt Hyperaccumulators Twenty-eight and 37 taxa of Co and Cu hyperaccumulator plants, respectively, have been reported (see Brooks, 1998; Reeves and Baker, 2000). These plants are mainly distributed in the Shaban Copper Arc of the Democratic Republic of Congo (formerly Za¨ıre). Some of these plants can hyperaccumulate both metals. However, there have been few experimental studies on the ability of these plants to accumulate metals, and therefore whether or not they can truly hyperaccumulate Cu and Co remains to be conﬁrmed. A recent study using hydroponic cultures showed that Haumaniastrum katangense and Aeollanthus biformifolius (Lamiaceae), both of which have been described as Cu and Co hyperaccumulators, did not hyperaccumulate Cu and Co in the shoots, but rather behaved as typical metal excluders (K¨ohl et al., 1997). Contamination of plant samples with dust is a possibility when sampling and analyzing wild plants from their natural habitats. This can cause large errors if the dust happens to be rich in the metals to be analyzed. Reeves and Baker (2000) gave an example in which 0.2 mg of malachite (a secondary

PHYTOREMEDIATION

9

Cu mineral) included as a dust with 100 mg of plant tissue genuinely containing 10 μg Cu g−1 is enough to raise the apparent Cu concentration to >1150 μg g−1. 5. Lead Pb hyperaccumulation is rare, primarily because Pb is very insoluble in soil. Non-accumulating plants such as Brassica juncea and Zea mays have been shown to hyperaccumulate Pb in the shoots once Pb solubility in the soil was greatly enhanced with synthetic chelates such as EDTA (see following). This chemically induced hyperaccumulation should not be confused with the natural hyperaccumulation discussed in this section. Fourteen taxa have been reported to be Pb hyperaccumulators with Pb concentration in the shoots varying from 1000 to 20,000 μg g−1 (Reeves and Baker, 2000). Similar to the situation with Cu/Co hyperaccumulators, Pb uptake and translocation in these reported hyperaccumulators have not been researched often under controlled conditions. Results from a ﬁeld survey showed that Thlaspi rotundifolium spp. cepaeifolium from a Pb/Zn mining area in northern Italy contained up to 8200 μg Pb g−1 in the shoots (Reeves and Brooks, 1983). However, our results using hydroponic and soil experiments (unpublished) and those of Huang and Cunningham (1996) indicate that this plant does not hyperaccumulate Pb in the shoots. In the roots, a large accumulation of Pb occurs in the apoplast, principally as lead phosphate deposits. This type of Pb accumulation does not represent a true uptake by roots. 6. Selenium and Arsenic The best known examples of Se hyperaccumulators are probably in the genus Astragalus (Leguminosae). In the 1930s, O. B. Beath and his colleagues found at least 13 taxa of Astragalus containing more than 1000 μg Se g−1 in the shoot dry matter in the Colorado Plateau in the United States (see Brooks, 1998; Reeves and Baker, 2000). The high Se concentrations in these plants caused serious disease in the grazing cattle and sheep. A number of other plants have been identiﬁed as Se hyperaccumulators in other regions of the world, including a Venezuelan tree, Lecythis ollaria (Lecythidaceae), which has a Se concentration in nuts of up to 18,200 μg g−1 and is therefore toxic to humans and animals (Aronow and Kerdel-Vegas, 1965). Arsenic accumulation by terrestrial plants is a very rare phenomenon, and no As hyperaccumulators have been reported until recently. Ma et al. (2001) discovered that Brake fern (Pteris vittata) growing on an As contaminated soil contained 3280–4980 μg As g−1 in the shoot (fronds) dry matter, compared to 2–23 μg As g−1 in 13 other species growing on the same soil. In a greenhouse experiment, Brake fern growing on a soil amended with 1500 μg As g−1 accumulated up to 22,630 μg As g−1 in the fronds in 6 weeks. It is also highly efﬁcient in transporting

10

S. P. McGRATH ET AL.

As from roots to shoots. This plant is hardy and fast growing, and thus could be potentially used in the phytoremediation of As. 7. Other Metals Reeves and Baker (2000) listed nine species from New Caledonia that had at least one specimen containing above 10,000 μg Mn g−1 in the shoots. These may be considered as Mn hyperaccumulators. In general, however, Mn hyperaccumulation has not been researched very much. Thallium (Tl) is extremely toxic to animals and humans, although soil contamination with Tl is rare. Unusual hyperaccumulation of Tl (>500 μg g−1 shoot dry weight) has been reported in two species from southern France, Iberis intermdia and Biscutella laevigata, both belonging to the Brassicaceae family (Anderson et al., 1999; Leblanc et al., 1999). Growing on soils with a total of Tl up to 40 μg g−1, Iberis intermdia and Biscutella laevigata contained up to 4000 and 14,000 μg Tl g−1 in the shoot dry weight. The plant-to-soil concentration quotient was mostly greater than 10 (Anderson et al., 1999), suggesting true hyperaccumulation rather than the possibility of soil contamination on plant specimens.

B. PHYTOEXTRACTION USING HYPERACCUMULATOR PLANTS 1. Zinc and Cadmium The concept of using hyperaccumulator plants to take up and remove heavy metals from contaminated soils was ﬁrst discussed by Chaney (1983). However, it was until the early 1990s that ﬁeld experiments were carried out to test the potential of phytoextraction of metals with hyperaccumulator plants. A ﬁeld-based experiment was conducted in 1991–1993 in sewage sludge-treated plots at Woburn, England, with the total Zn in soil varying from 124 to 444 mg kg−1 and the total Cd varying from 2.8 to 13.6 mg kg−1 (Baker et al., 1994a; McGrath et al., 1993, 2000). This experiment compared metal extraction efﬁciency of different hyperaccumulator plant species, including several populations of the Zn hyperaccumulator T. caerulescens. In 1991, two populations of T. caerulescens (Prayon from Belgium and Whitesike from the UK) produced shoot biomass yields of 3.6–4.5 t ha−1 (dry weight) and accumulated 2000–4300 μg Zn g−1 dry weight in the shoots (Fig. 2). Biomass increased to 7.5–7.8 t ha−1 in 1992, but shoot Zn concentration decreased to 500–2200 μg g−1. The concentrations of Zn in the shoots were considerably lower than the 10,000 μg g−1 value used to deﬁne hyperaccumulation, because the soil was only slightly or moderately contaminated with Zn. Nevertheless, these concentrations were still 10- to 20-fold higher than those in a number of normal crop species growing on the same plots (McGrath et al., 2000). Both the concentration

PHYTOREMEDIATION

11

Figure 2 Concentrations (a) and total uptake (b) of Zn in the shoots of T. caerulescens (the Prayon population) grown on different plots of a long-term sewage sludge experiment at Woburn, England, in 1991 and 1992.

of Zn and the total Zn removal in the shoots of T. caerulescens increased with increasing soil Zn (Fig. 2). In the plots with total soil Zn >300 mg kg−1, total Zn removals by a single crop of T. caerulescens were between 10 and 25 kg ha−1. Being a wild plant, T. caerulescens is not easy to grow under ﬁeld conditions, and a substantial variation in individual yields exists. It is possible to improve biomass production of hyperaccumulator plants through optimization of agronomic inputs (Bennett et al., 1998). To estimate the maximum potential removal of Zn by an optimized crop of T. caerulescens, McGrath et al. (1993, 2000) made model calculations based on the uptake by the largest two rows of the plants observed in the ﬁeld experiment. The calculated maximum potential removals of Zn were 25–50 kg ha−1, two to three times higher than the average values shown in Fig. 2. With the optimized removal rate, it would take 7–14 crops of T. caerulescens to reduce total soil Zn from 440 to 300 μg g−1. This compares favorably with over 800 croppings with Brassica napus (oilseed rape) and more than 2000 croppings with Raphanus sativus (raddish) (McGrath et al., 1993). In the ﬁeld experiment at Woburn described previously, the Cd concentration in the shoots of two populations of T. caerulescens (Prayon and Whitesike) was <40 μg g−1, and the Cd removal varied from a few to about 300 g ha−1 (McGrath et al., 2000). The Whitesike population was more efﬁcient than Prayon in Cd uptake. Although the rate of Cd removal by the Whitesike population is small compared to the inputs of Cd from sludge on this experimental site, it is far greater than inputs of Cd from the use of phosphate fertilizers and from the atmospheric deposition. Later, samples of a population of T. caerulescens from southern France (Ganges) was also grown in the Woburn experiment. This population was found

12

S. P. McGRATH ET AL.

Figure 3 Cadmium concentrations in three different ecotypes of T. caerulescens grown on different plots of a long-term sewage sludge experiment at Woburn, England.

to be far superior to the other two in Cd uptake (Fig. 3), with Cd concentration in the shoots reaching 500 μg g−1 in the highest Cd plot (13.6 μg Cd g−1 in the soil; Lombi et al., 2000a). This work shows that there is a substantial potential to screen for more efﬁcient ecotypes for metal uptake. Also in 1991–1992, Brown et al. (1995b) compared Zn and Cd uptake by T. caerulescens (Prayon) and two other non-hyperaccumulators on long-term sewage sludge plots in Beltsville, Maryland, which have total soil Zn and Cd up to 181 and 5.5 μg g−1, respectively. They found the concentrations of Zn in T. caerulescens varying from about 1000 up to 4000 μg g−1, about 10-fold greater than those in Silene vulgaris and lettuce. Similar to the results obtained from the Woburn experiment, the Prayon population of T. caerulescens did not take up signiﬁcantly more Cd than the other two plants. Brown et al. (1995b) did not present biomass yields on a unit area basis; therefore it was not possible to calculate metal extraction rates. Robinson et al. (1998) grew T. caerulescens (the Ganges population from southern France) both in pots and in mine waste in ﬁelds. They found that a single fertilized crop could remove 60 kg Zn ha−1 and 8.4 kg Cd ha−1. Because bioaccumulation coefﬁcients (plant/soil metal concentration quotients) were in general higher for Cd than for Zn, phytoremediation using T. caerulescens would be feasible for low levels of soil Cd, but not feasible to remediate the extremely high Zn concentrations (40,000 μg g−1) found in the mine wastes. The main constraints for using T. caerulescens to phytoextract Zn and Cd from contaminated soils are the slow growth rate, its rosette growth habit, and its generally small biomass production. Several approaches may be taken to tackle these constraints. Signiﬁcant differences in zinc concentration and plant size were found among sib families within the same populations of T. caerulescens, but the

PHYTOREMEDIATION

13

characters of the zinc concentration in shoots and plant size were not correlated (Pollard and Baker, 1996). In one population, the variations in these characters appeared to be heritable. The results indicate that selecting large-sized plants of T. caerulescens does not lead to reduced Zn concentration in the shoots as a result of a growth dilution, and also that good attributes for phytoextraction found in individual plants may be heritable. The other approach is to transfer the metal hyperaccumulation traits from hyperaccumulators to other high biomass plants, through either hybridization or genetic engineering. Recently, Brewer et al. (1999) produced several somatic hybrids between T. caerulescens and the high biomass crop oilseed rape (Brassica napus). These hybrids produced a larger biomass than T. caerulescens and had an erect growth habit that is suitable for mechanical harvesting. The hybrids were able to accumulate and tolerate Zn and Cd at levels that are toxic to Brassica napus, although their ability to accumulate metals appeared to be lower than T. caerulescens. In the future, it may be possible to introduce genes responsible for metal hyperaccumulation and internal tolerance to high biomass, and preferably non-food, crops. Genetic engineering has been successfully applied to transform poplar trees for the volatilization of Hg (see following). 2. Nickel Compared to Thlaspi caerulescens, many Ni hyperaccumulators yield a much higher biomass. Nicks and Chambers (1995) found that a natural crop of the Californian Ni hyperaccumulator Streptanthus polygaloides was capable of producing up to 100 kg Ni ha−1 (worth $550 ha−1 at the time). Incineration of the crop could produce combustion energy for electricity generation, yielding additional return. They concluded that the return to a farmer growing a “crop of Ni” would be roughly comparable, or superior to that obtained for a crop of wheat. Such an operation of growing hyperaccumulators over a low-grade ore body or mineralized soil for the prodcution of bio-ore has been termed phytomining (Brooks et al., 1998). Robinson et al. (1997a,b) investigated the Ni extraction potential of two other Ni hyperaccumulators. Dry biomass yields of Alyssum bertolonii and Berkheya coddii reached 9 and 22 t ha−1, respectively. The Ni yields were in the range of 70–100 kg ha−1. They concluded that the phytomining of Ni could be economically feasible at many sites worldwide. Their studies also showed that both hyperaccumulators responded markedly to fertilizer additions without a reduction in the Ni concentration in shoot tissues. The concentration of Ni in the shoots of B. coddii was found to correlate with extractable Ni in the soil. Unless the extractable concentration is perfectly buffered by soil solid phases, this means that the rate of Ni extraction will decrease as the number of cropping increases. For phytoremediation of Ni-contaminated soils, Robinson et al. (1997b) estimated that four croppings with B. coddii would be required to bring the total soil Ni from 250 μg g−1 to below the EU guideline value of 75 μg g−1.

14

S. P. McGRATH ET AL.

3. Other Metals Phytomining for Tl using Tl hyperaccumulators, such as Iberis intermdia and Biscutella laevigata, has also been suggested (Anderson et al., 1999).

C. MECHANISMS OF METAL HYPERACCUMULATION Hyperaccumulation of heavy metals is likely to involve several steps, including metal transport across root cell plasma membranes, xylem loading and translocation, and sequestration of metals in specialized leaf cells or vacuoles. Sequestration may involve both physical compartmentation and complexation with ligands. Because of the multiple processes involved, metal hyperaccumulation is likely to be controlled by multiple genes. Indeed, a recent study on crosses between the Zn hyperaccumulator Arabidopsis halleri and the non-accumulating, non-tolerant species Arabidopsis petraea showed that metal accumulation and tolerance in hyperaccumulators are likely to be genetically independent traits (Macnair et al., 1999). A true metal hyperaccumulator must posses both genetic traits for its hyperaccumulation potential to be realized. Another interesting observation is that metal hyperaccumulation appears to be a constitutive property; i.e., populations of the same hyperaccumulator species collected from both metalliferous and nonmetalliferous sites are able to hyperaccumulate the metal similarly (Bert et al., 2000; Boyd and Martens, 1998a; Reeves and Baker, 1984). This raises an interesting question regarding the evolution of metal hyperaccumulation, which still awaits convincing answers. A number of hypotheses have been advanced to explain the ecological functions of metal hyperaccumulation. Among them the hypothesis that metal hyperaccumulation plays a role in defence against pathogens and herbivores has gained support from recent studies (see reviews by Boyd (1998) and Pollard et al. (2000)). There has been rapid progress in our understanding of the processes involved in metal hyperaccumulation in recent years. However, the whole picture is still far from complete. 1. Metal Uptake a. Zinc and Cadmium A number of studies using hydroponic culture have demonstrated the extraordinary ability of T. caerulescens (Baker et al., 1994b; Brown et al., 1995a; Shen et al., 1997) and Arabidopsis halleri (Zhao et al., 2000) to take up Zn. Even with a low concentration of Zn in the nutrient solution (e.g., 1 μM), T. caerulescens was able to accumulate much higher concentrations of Zn in the shoots than in other

PHYTOREMEDIATION

15

non-hyperaccumulating plants (Shen et al., 1997). Zn uptake by T. caerulescens from non-contaminated or slightly contaminated soils was always greater than that by non-hyperaccumulator plants such as Brassica napus or Raphanus sativus (Knight et al., 1997; McGrath et al., 1993). These results suggest that T. caerulescens possesses a highly efﬁcient system in Zn uptake and/or translocation. Lasat et al. (1996) studied the kinetics of 65 Zn inﬂux in the roots of T. caerulescens and a non-accumulator Thlaspi arvense. They found that the maximum inﬂux velocity, Vmax , was 4.5-fold higher in T. caerulescens than in T. arvense. Recently, Pence et al. (2000) cloned a gene in T. caerulescens, ZNT1, which encodes a high-afﬁnity Zn transporter that is a member of the ZIP (ZRT, IRT-like Protein) family of metal transporters. They found that ZNT1 was highly expressed in the roots of T. caerulescens even when the plants had a high Zn status, whereas in T. arvense ZNT1 was expressed at far lower levels, and the expression was stimulated by Zn deﬁciency. Both physiological and molecular studies show that the Zn hyperaccumulator T. caerulescens has a constitutively high density of Zn transporter(s) on root plasma membranes. Thus, at least for Zn and Cd (see following), an uptake rate higher than that in non-accumulating plants is one of the reasons for metal hyperaccumulation. In the case of Cd, it is often assumed that its uptake is unspeciﬁc and inadvertent via transporters for other essential nutrients. Members of the ZIP gene family are capable of transporting metals including Fe, Zn, Mn, and Cd (Guerinot, 2000). The Fe transporters such as IRT1 (a member of ZIP) and Nramp have been shown to be capable of transporting several metals including Cd in Arabidopsis thaliana (Eide et al., 1996; Korshunova et al., 1999; Thomine et al., 2000). Using point mutation to substitute a single amino acid residue of IRT1, Rogers et al. (2000) demonstrated altered selectivity of the transporter for different metals. In addition, there is some evidence that Ca channels may also mediate transport of Cd across the plasma membrane in wheat, albeit at a low level of afﬁnity (Clemens et al., 1998). In the Zn hyperaccumulator T. caerulescens (Prayon), Pence et al. (2000) found that the Zn transporter ZNT1 can also mediate a low-afﬁnity uptake of Cd. The high expression of ZNT1 may explain the accumulation of Cd in the Prayon ecotype of T. caerulescens. However, the level of Cd accumulation in the Prayon population has been shown to be much lower than that in the Ganges population (Lombi et al., 2000). This difference cannot be explained by the difference in Zn uptake, because the two populations show similar characteristics of Zn uptake both in soil experiments and in short-term hydroponic studies of 65Zn inﬂux (Lombi et al., 2000, 2001a). We found that the Vmax for 109Cd and the Cd concentration in xylem sap were approximately ﬁvefold higher in Ganges than in Prayon (Fig. 4). In addition, at an equimolar concentration, Zn inhibited Cd uptake by Prayon but not by Ganges (Lombi et al., 2001a). Furthermore, Ca and its channel blocker La appeared to inhibit Cd uptake in the Prayon ecotype but not in the Ganges ecotype (Zhao et al., unpublished results). These results provide physiological

16

S. P. McGRATH ET AL.

Figure 4 Concentration-dependent T. caerulescens.

109

Cd uptake kinetics in two contrasting populations of

evidence that a high-afﬁnity Cd transporting system exists in the Ganges population of T. caerulescens. Iron deﬁciency further enhanced Cd uptake by the Ganges population, but not by the Prayon population, and had little effect on Zn uptake in both populations (Lombi et al., unpublished). However, whether or not the highafﬁnity Cd transporting system observed in the Ganges population belongs to one of the metal transporters in the ZIP family (Guerinot, 2000) remains to be elucidated. b. Nickel Although Ni hyperaccumulators are the most numerous among plant species known to show hyperaccumulation, the mechanisms of Ni uptake by these plants are still poorly understood. The molecular mechanisms of Ni transport in Ni hyperaccumulators are largely unknown. By comparing the Ni hyperaccumulator Thlaspi goesingense with the non-accumulator Thlaspi arvense, Kr¨amer et al. (1997a) found that the rates of Ni uptake and root-to-shoot translocation were similar between the two species, as long as the concentration of Ni in the uptake solution was below the toxicity threshold. However, because T. goesingense was much more tolerant to Ni than T. arvense, the former species could continue to take up Ni when the concentration of Ni in the solution was high and above the toxicity threshold for T. arvense. These authors suggested that Ni tolerance alone was sufﬁcient to explain the Ni hyperaccumulator phenotype observed in hydroponically grown T. goesingense when compared with the Ni-sensitive non-hyperaccumulator T. arvense. This conclusion is clearly very different from that for the Zn and Cd hyperaccumulation in T. caerulescens. It also conﬂicts with the observation that hyperaccumulation and tolerance are separate genetic traits (Macnair et al., 1999). It is not known whether the conclusion drawn by Kr¨amer et al. (1997a) is

PHYTOREMEDIATION

17

also applicable to other Ni hyperaccumulators, particularly those in the Alyssum genus. Although T. goesingense is a known Ni hyperaccumulator, the level of Ni accumulation observed in this plant is lower than that observed in Alyssum. For example, both Alyssum bertolonii and Alyssum lesbiacum readily accumulate over 10,000 μg Ni g−1 in the shoots in hydroponic experiments (Gabbrielli et al., 1990; Kr¨amer et al., 1996), whereas T. goesingense tend to accumulate <10,000 μg Ni g−1 under similar conditions (Kr¨amer et al., 1997a). In a pot experiment, we also found that T. goesingense was less tolerant to Ni than A. bertolonii and A. lesbiacum (K¨upper et al., 2001). 2. Metal Translocation from Root to Shoot One of the key features distinguishing metal hyperaccumulators from nonhyperaccumulators is the extremely efﬁcient translocation of metals from roots to shoots. For example, translocation of 65Zn from roots to shoots was approximately 10-fold greater in T. caerulescens compared with the non-acumulator T. arvense, over a 96-h uptake period (Lasat et al., 1996; Fig. 5). In contrast, the roots of T. arvense retained more 65Zn than the roots of T. caerulescens. This may be partly explained by a smaller sequestration of metals in the root vacuoles of hyperaccumulators than in non-hyperaccumulators (Lasat et al., 1998a). It is also possible that hyperaccumulators have a more efﬁcient xylem loading. Translocation of Ni from roots to shoots may involve speciﬁc ligands in some hyperaccumulator species. Kr¨amer et al. (1996) found that exposing several Ni hyperaccumulator species of Alyssum to Ni elicited a large and proportional increase in the levels of histidine in the xylem sap. Histidine in the xylem sap was shown to be coordinated with Ni, although the concentration of histidine was enough to complex only about 25% of the Ni in the xylem sap. A similar response was observed in two Ni hyperaccumulator species outside the genus Alyssum, Streptanthus polygaloides and Berkheya coddii (Smith et al., 1999). In contrast, exposing the Zn hyperaccumulator Arabidopsis halleri to Zn, or the Mn hyperaccumulator Grevillea exul var. exul to Mn, did not result in increased histidine in the xylem saps. The histidine response may not be universal in all Ni hyperaccumulator species. Persans et al. (1999) did not observe any Ni-inducible responses in terms of histidine concentrations in the roots, shoots, and xylem sap of T. goesingense, nor did they ﬁnd any regulation by Ni of three cDNAs encoding the enzymes involved in the histidine biosynthetic pathway. 3. Mechanisms of Metal Tolerance in Hyperaccumulator Plants Thlaspi caerulescens and Arabidopsis halleri have been shown to accumulate 25,000–30,000 μg Zn g−1 in the shoot dry matter without growth reduction or showing phytotoxicity (Brown et al., 1995a; Shen et al., 1997; Zhao et al., 2000).

18

S. P. McGRATH ET AL.

Figure 5 Time course of 65Zn accumulation in roots (a) and shoots (b) of T. caerulescens and T. arvense. (From Lasat et al., 1996, with permission from the American Society of Plant Physiologists).

Similarly, Alyssum lesbiacum accumulated up to 20,000 μg Ni g−1 in the shoots without reduction in plant biomass (Kr¨amer et al., 1996). These concentrations contrast dramatically with toxicity thresholds of about 500 μg Zn g−1 and 10–50 μg Ni g−1 in many crop species (Marschner, 1995). Clearly, metal hyperaccumulators must also be hypertolerant to the metals they accumulate. Furthermore, whereas metal excluders may be able to tolerate metals present in the substrate, metal hyperaccumulators must be tolerant to metals present both externally and internally. Tolerance mechanisms employed by metal hyperaccumulators most likely involve internal detoxiﬁcation, which may be achieved through (1) cellular and subcellular compartmentation and (2) complexation with ligands. a. Cellular and Subcellular Compartmentation Compartmentation of heavy metals may be achieved at both cellular and subcellular levels. Studies using energy dispersive X-ray microanalysis (EDXMA)

PHYTOREMEDIATION

19

Figure 6 EDXMA line-scan of a cross section of a T. caerulescens leaf showing preferential accumulation of Zn in the epidermal cells (top panel) and concentration of Zn in the single-cell saps extracted from epidermal and mesophyll cells of leaves of T. caerulescens (bottom panel). (Adapted from K¨upper et al., 1999, with permission from the American Society of Plant Physiologists).

clearly show that Zn is preferentially accumulated in the leaf epidermal cells in Thlaspi caerulescens (K¨upper et al., 1999; V´azquez et al., 1992, 1994) (Fig. 6), except for the guard and subsidiary cells in the stomatal complex (Frey et al., 2000). In contrast, P is preferentially distributed in the leaf mesophyll cells (K¨upper et al., 1999), thus avoiding precipitation of zinc phosphate in leaves which would cause

20

S. P. McGRATH ET AL.

P deﬁciency (Zhao et al., 1998). Further studies using single-cell sap sampling (K¨upper et al., 1999; Fig. 6) and ultrathin cryosectioning (Frey et al., 2000) indicate that Zn is sequestered in the vacuoles in the epidermal cells. K¨upper et al. (1999) also observed a strong correlation between the Zn concentration and the size of leaf epidermal cells. Because larger cells tend to be more vacuolated, this suggests that the size of vacuoles may be an important factor determining Zn distribution in leaf tissues. Preferential accumulation of Ni in the leaf epidermal cells has also been observed in several Ni hyperaccumulators including Senecio coronatus (MesjaszPrzybylowicz et al., 1994), Alyssum lesbiacum (Kr¨amer et al., 1997b), four other species of Alyssum, and three other Ni hyperaccumulators from Greece (Psaras et al., 2000), Alyssum bertolonii, Alyssum lesbiacum, and Thlaspi goesingense (K¨upper et al., 2001). All Alyssum species examined have dense nonglandular stellate trichomes on the leaf surfaces. Kr¨amer et al. (1997b) found high concentrations of Ni in the trichomes of A. lesbiacum. However, this was not conﬁrmed in a recent study, which showed that Ni is excluded from the trichomes in all Alyssum species examined (Psaras et al., 2000). These differences may be caused by differences in methodology, particularly sample preparation. Methods used to dehydrate fresh plant tissues (e.g., freeze-substitution or freeze-drying) prior to X-ray microanalysis (V´azquez et al., 1992, 1994) or proton microproble analysis (Kr¨amer et al., 1997b; Mesjasz-Przybylowicz et al., 1994) may cause redistribution or losses of elements, and thus artifacts. In contrast, X-ray microanalysis conducted on frozen hydrated tissues (Frey et al., 2000; K¨upper et al., 1999) is more likely to provide a true picture of the patterns of elemental distribution in plant cells. However, one drawback of the latter method is a lack of resolution for observation at the subcellular level. Preferential distribution of Zn and Ni in leaf epidermal cells and reduced accumulation of metals in the mesophyll cells and those in the stomatal complexes may be useful mechanisms to protect photosynthesis, which takes place in the mesophyll cells, and the function of stomata. However, this pattern of metal distribution is not observed in the Zn hyperaccumulator Arabidopsis halleri. A. halleri has trichomes on the leaf surface, and the basal compartments of the trichomes are highly enriched with Zn (and Cd, if present) (K¨upper et al., 2000; Zhao et al., 2000). But the epidermal cells other than trichomes in A. halleri leaves are very small and do not accumulate Zn or Cd. Instead, with increasing Zn/Cd accumulation, mesophyll cells (most likely the vacuoles) become an increasingly important sink of the metals (K¨upper et al., 2000). A common feature of the hyperaccumulators studied so far is the sequestration of metals in the vacuoles (Frey et al., 2000; K¨upper et al., 1999, 2000; V´azquez et al., 1992, 1994). The Ni hyperaccumulator T. goesingense was found to sequester more Ni in the vacuoles in leaf cells than the non-accumulator T. arvense (Kr¨amer et al.,

PHYTOREMEDIATION

21

2000). The difference in vacuolar sequestration has been proposed as an important reason for differences in tolerance to Ni between T. goesingense and T. arvense (Kr¨amer et al., 2000). In the non-hyperaccumulator Silene vulgaris, Chardonnens et al. (1999) also found that the naturally selected Zn-tolerant ecotype had an enhanced Zn uptake across tonoplasts compared to the Zn-sensitive ecotype. This suggests that metal transport across tonoplasts, and hence vacuolar sequestration, may be a key component of metal tolerance mechanisms. b. Complexation with Ligands Complexation of metals with ligands results in decreased free ion activity and thus decreased toxicity. There is evidence for the role of ligands in detoxifying metals in hyperaccumulators. For example, histidine has been shown to be involved in the tolerance of Ni by several Alyssum species, possibly through the formation of the Ni–histidine complex in root cells and in xylem saps (Kr¨amer et al., 1996). In many Ni hyperaccumulators, there is an apparent association of Ni and organic acids such as citrate, malate, and malonate (Brooks, 1998). For example, Lee et al. (1977, 1978) showed a strong correlation between Ni and citrate in the leaves of 17 plant species from New Caledonian which varied in their ability to accumulate Ni, and that the puriﬁed aqueous extracts of these plants contained Ni as a Ni–citrate complex. In the latex collected from Sebertia acuminata, citrate was found to complex about 40% of the Ni present (Sagner et al., 1998). Ni–citrate complexes also existed in the aqueous extracts of the leaves of the Philippine hyperaccumulators Dichapetalum gelonioides, Phyllanthus palawanensis, and Walsura monophylla (Homer et al., 1991b). In several Alyssum species from Europe, however, the concentrations of citrate were low, and malate and malonate appeared to be the main counter ions for Ni in the aqueous extracts (Brooks, 1998; Brooks et al., 1981; Lee et al., 1978). As pointed out by Brooks (1998), an apparent association of Ni and organic acids in the aqueous extracts of plant leaves does not necessarily mean that they were originally complexed together in the plant materials. It is also important to remember that malate and malonate have relatively low afﬁnity for Ni and other heavy metals, and therefore the complexes of Ni and these organic anions are not very stable. In comparison, citrate has a higher afﬁnity for Ni and other heavy metals, and the metal citrate complexes are expected to be stable in the vacuoles which have a pH of about 5.5. In the Zn hyperaccumulator Thlaspi caerulescens, Mathys (1977) proposed that malate plays an important role in shuttling Zn from the cytoplasm to vacuoles. This hypothesis is mostly likely incorrect, because malate has a rather low afﬁnity for Zn and the Zn–malate complex would be unstable in cytoplasm. Analysis of leaves of Thlaspi caerulescens and Arabidopsis halleri showed that malate and citrate concentrations did not vary greatly (Shen et al., 1997; Zhao et al., 2000). High concentrations of malate in Thlaspi caerulescens leaves appear to be a constitutive property, and may only explain the charge

22

S. P. McGRATH ET AL.

balance between cations and anions rather than a mechanism for metal complexation (Tolr`a et al., 1996). In the roots of Arabidopsis halleri, the concentration of citrate was found to increase with increasing Zn (Zhao et al., 2000). Recently, Salt et al. (1999) used X-ray absorption spectroscopy to investigate Zn binding ligands in T. caerulescens. The X-ray absorption spectra of liquid N2 frozen tissues were compared to the spectra of a number of model compounds. They found that 38, 9, 16, and 12% of the total Zn (5050 μg g−1 dry weight) in the shoots was complexed with citrate, oxalate, histidine, and cell wall, respectively, with the remaining 26% as aqueous free cation. The citrate–Zn complex also accounted for 21% of the total Zn in the xylem sap, with the remaining amount apparently present as the free cation. In the roots of T. caerulescens that contained 1320 μg Zn g−1 dry weight, histidine complexed 70% of the total Zn. Malate was found not to be involved in Zn complexation in the roots, shoots, or xylem sap. These results suggest that histidine may be important in detoxifying Zn in the root cells; however, organic acids, particularly citrate, may be involved in the root-to-shoot transport and storage of Zn in the leaf vacuoles. It is important to point out that although organic acids such as citrate may play a role in metal chelation and storage in vacuoles, they account for neither the metal speciﬁcity nor the species speciﬁcity of metal hyperaccumulation (Kr¨amer et al., 1996). Little is known about the ligands responsible for chelating Cd, Cu, and Co in metal hyperaccumulators. In the non-accumulator Arabidopsis thaliana, phytochelatins play an important role in the binding of Cd and the tolerance to Cd (Cobbett, 2000). It is not known whether phytochelatins are also involved in the detoxiﬁcation of Cd in Cd hyperaccumulators. There is no evidence for a role of phytochelatins in Zn or Ni tolerance and hyperaccumulation (Sagner et al., 1998; Salt and Kr¨amer, 2000). Kr¨amer et al. (1996) also observed an increase in the concentration of histidine in the xylem sap of Alyssum lesbiacum on exposure to Co, similar to the response to Ni. 4. Rhizosphere Aspects of Metal Acquisition by Hyperaccumulator Plants Rhizosphere processes are important aspects of plant nutrient acquisition, for ions that primarily reach the root surface by diffusion (e.g., Fe, Zn, and P) and particularly under nutrient-limiting conditions (Hinsinger, 1998; Marschner, 1995). In some cases, differences between plant species or between cultivars within the same species in nutrient acquisition from soils have been attributed to their different abilities to modify the rhizosphere microenvironment. Furthermore, the rooting pattern and associations between roots and microorganisms may also be important factors in determining nutrient acquisition by plants. Do heavy metal hyperaccumulators employ rhizosphere-related mechanisms to achieve or enhance metal accumulation?

PHYTOREMEDIATION

23

Figure 7 Root proliferation of T. caerulescens grown in rhizoboxes of previously uncontaminated soil with (+) or without (−) ZnO added at 1000 mg Zn kg−1 soil. (Photos courtesy of S. N. Whiting, The University of Melbourne Australia).

a. Rooting Pattern Two recent studies (Schwartz et al., 1999; Whiting et al., 2000) provided an important insight into the rooting pattern of the Zn/Cd hyperaccumulator Thlaspi caerulescens. These studies showed that T. caerulescens responded positively to Zn in soil. The roots of this plant predominantly colonized a zone of Zn-polluted soil that was embedded in an uncontaminated agricultural soil (Schwartz et al., 1999). The addition of sparingly soluble ZnO to one half of a rhizobox containing an uncontaminated soil stimulated proliferation of lateral roots in the metalamended half, whereas root biomass in the unamended half was reduced (Fig. 7) (Whiting et al., 2000). The plants consistently allocated about 70% of their total root biomass and root length, and about 70% of the current assimilate (14 C) into the metal-enriched soil. When both halves of soil were either unamended or amended with ZnO, even distribution of root biomass occurred. In contrast, the non-hyperaccumulator T. arvense tended to restrict root growth in the metalenriched soil and proliferate in the unamended soil. Addition of CdS appeared to have a similar stimulating effect on the root growth of a Cd hyperaccumulating population of T. caerulescens (Whiting et al., 2000), whereas Pb had no clear effect (Schwartz et al., 1999). These results suggest that roots of T. caerulescens are able to sense and actively forage in the Zn- or Cd-rich patches in soil. Because

24

S. P. McGRATH ET AL.

of the usual heterogeneity of metal concentrations in soil, the positive response of the rooting pattern of this hyperaccumulator to localized patches of soil containing high Zn and Cd may be an important reason for its hyperaccumulation ability compared to non-hyperaccumulators. However, the mechanisms of signal transduction that lead to the sensing of metal-rich patches remain to be elucidated. b. Rhizosphere Microbes Mycorrhizal associations increase the area of nutrient exploitation in soil by roots, and this in turn increases acquisition of nutrients like P and Zn under nutrientlimiting conditions (Marschner, 1995). However, in metal-contaminated soils, mycorrhizal associations, particularly those involving ectomycorrhizas, often result in decreased metal accumulation in plant shoots and enhanced plant tolerance to metals (Leyval and Joner, 2001; Marschner, 1995). Many hyperaccumulators belong to the family Brassicaceae and do not have mycorrhizal associations. The microbial community in the rhizosphere of hyperaccumulator plants may have its own characteristics, e.g., enhanced tolerance to heavy metals (Ghaderian et al., 2000). But this does not mean that rhizosphere microorganisms are directly involved in the acquisition of metals by hyperaccumulator plants. In the case of Se volatilization, rhizosphere microbes are known to play a role (see following). c. Rhizosphere pH Acidiﬁcation in the rhizosphere is expected to increase the solubility of heavy metals in soil. Both the Ni hyperaccumulator Alyssum murale and the Zn hyperaccumulator Thlaspi caerulescens appeared to thrive under neutral to slightly alkaline conditions, and hyperaccumulate metals at these pH’s (Bernal and McGrath, 1994a; Brown et al., 1994). Decreasing the pH from the neutral to the acidic range depresses the growth of the two species. The soil solution pH did not decrease after growing T. caerulescens (Knight et al., 1997). Also, there were no signiﬁcant differences between hyperaccumulators and non-hyperaccumulator species in the changes in the rhizosphere pH (Bernal et al., 1994b; McGrath et al., 1997). These studies demonstrate that Ni and Zn hyperaccumulation probably does not involve rhizosphere acidiﬁcation. d. Root Exudates Graminaceous plant species release speciﬁc metal-chelating compounds (phytosiderophores) into the rhizosphere to mobilize Fe, and possibly Zn (Marschner, 1995). Does metal hyperaccumulation involve exudation of speciﬁc metalmobilizing compounds? In a recent study (Zhao et al., 2001), we compared the ability of root exudates collected from Thlaspi caerulescens and the nonaccumulators Brassica napus and wheat to mobilize heavy metals from a soil or from metal loaded resin. The results showed that the root exudates of wheat

PHYTOREMEDIATION

25

mobilized substantial amounts of Zn, Cd, Cu, and Fe, and the mobilizing ability was greatly enhanced by the deﬁciency of Fe and Zn in the plants. In contrast, the root exudates from T. caerulescens and B. napus showed little ability to mobilize heavy metals under either Fe/Zn deﬁcienct or sufﬁcient conditions. In a study investigating root exudates from the Ni hyperaccumulator Thlaspi goesingense, Salt et al. (2000) found no evidence of the presence of any high-afﬁnity Ni-chelating compounds. In contrast, upon exposure to Ni, the non-hyperaccumulator Thlaspi arvense exuded much more histidine and citrate than T. goesingense. The enhanced release of histidine and citrate by T. arvense roots may be a strategy to reduce Ni uptake and toxicity. So far, the evidence appears to suggest that root exudates are not directly involved in Zn and Ni hyperaccumulation in T. caerulescens and T. goesingense. More research is needed to examine root exudates of other metal hyperaccumulators. e. Utilization of Metal Pools in Soil It has been speculated that hyperaccumulator plants may be able to access and utilize heavy metals in the less soluble pools in soil that are largely unavailable to non-accumulator species. The answer to this question is relevant to the assessment of the phytoremediation potential of hyperaccumulator plants. Studies by Knight et al. (1997) and McGrath et al. (1997) showed that T. caerulescens was able to substantially lower the concentrations of Zn and Cd in the soil solution or in the 1 M NH4NO3 extractable fraction of the soil. In addition, the uptake of Zn was far more than the amounts of Zn present initially in the soil solution or in the 1 M NH4NO3 extractable fraction. After growing T. caerulescens, the decreases in the soil solution pool or in the 1 M NH4NO3 extractable fraction explained <1 and <10% of the total Zn uptake, respectively. These results suggest that a large proportion of Zn taken up by the hyperaccumulator plant must have been derived from other pools, possibly exchangeable or adsorbed pools. Two possibilities exist to account for the previous observations. First, the metals in the soil solution or in the 1 M NH4NO3 extractable fraction are highly buffered, so that a decrease due to the root uptake is largely replenished from other pools. Second, the roots of T. caerulescens are able to mobilize metals in the other pools directly. In light of recent evidence that root exudates appear not to be responsible for metal hyperaccumulation (see previous), the second possibility has become less plausible. One way to assess the availability of metals in soil to plants is to determine the size of the isotopically exchangeable pools, the so-called L-value which represents the labile fraction of a metal in soil (Hamon et al., 1997). Recently, Hutchinson et al. (2000) determined the L-values of T. caerulescens and the non-hyperaccumulator Lepidium heterophyllum grown on contaminated soils, using 109 Cd as a radiotracer. The L-values were similar between the two contrasting plant species and also

26

S. P. McGRATH ET AL.

similar between different populations of T. caerulescens which have been shown to differ in Cd uptake (Lombi et al., 2000a). These results suggest that both species and different populations of T. caerulescens accessed the same isotopically exchangeable pool in the soil. A similar conclusion was also drawn by G´erard et al. (2000), who compared Cd uptake by T. caerulescens and ryegrass (Lolium perenne) and lettuce (Lactuca sativa). These results imply that T. caerulescens does not possess an enhanced ability to utilize the non-labile pool of Cd in soils. It is important to remember that the L-values for Cd often represent a large proportion of the total Cd in soils and certainly exceed the Cd uptake by a single crop. Therefore, before the isotopically exchangeable pool of metals is depleted, it may not be possible to detect signiﬁcant differences between plants in the utilization of the isotopically non-exchangeable (non-labile) pool. If indeed hyperaccumulator plants are no more efﬁcient than non-accumulating plants in removing metals from the non-labile pool, then the size of the labile pool, which can vary from 5 to >95% of the total metal in soils, may present the limit of phytoextraction.

III. CHEMICALLY ENHANCED PHYTOEXTRACTION Chemically induced or enhanced phytoextraction has been proposed as an alternative to the use of natural hyperaccumulators for the cleaning up of polluted soils. As discussed earlier, hyperaccumulator plants generally are, with the exception of some of the Ni hyperaccumulator species, low biomass plants. Hyperaccumulators are also slow growing and can have difﬁculty competing with other plants growing in soils with low levels of contamination. Furthermore, some of the most promising hyperaccumulators, such as species belonging to the Thlaspi genus, form small rosettes that could be difﬁcult to harvest mechanically. However, when extraction potential is considered the high biomass of some crops cannot offset the extremely large differences in terms of metal concentrations between these plants and the hyperaccumulator species. Chaney et al. (2000) showed that maize, even when grown in polluted soil, can hardly extract more than 0.05 kg ha−1 of Cd in a single crop, whereas T. caerulescens could extract 1.25 kg ha−1. This calculation could be even more in favor of the use of hyperaccumulator plants if we consider that the “super-Cd Thlaspi” mentioned by Chaney et al. (2000) or the Ganges population of T. caerulescens (Lombi et al., 2000a) can accumulate much larger concentrations of Cd under ﬁeld conditions. In non-hyperaccumulator plants, factors limiting their potential for phytoextraction include small root uptake and little root-to-shoot translocation of metals. Methods making use of metal-mobilizing agents have been proposed speciﬁcally to overcome these limitations. Following this approach, a high biomass crop is grown on the soil requiring remediation. When the crop has reached its

PHYTOREMEDIATION

Figure 8

27

Conceptual representation of chemically enhanced phytoextraction.

maximum biomass a metal-mobilizing agent, generally a metal chelating compound, is applied to the soil. This causes an immediate increase in metal mobility in the soil and uptake by the plants. The metal is then translocated into the shoot and at this point the plants can be harvested and consequently the metal can be removed (Fig. 8). Several studies have focused on optimizing the combination of plants and mobilizing agents for phytoremediation (Cooper et al., 1999; Ebbs and Kochian, 1998; Wu et al., 1999). From the results obtained it appears that B. juncea and EDTA (ethylenediaminotetraacetic acid) are the most promising. However, the ideal combination of plant and mobilizing agent depends on the speciﬁc pattern of pollution and the climatic conditions.

A. POTENTIAL APPLICATIONS Chemically enhanced phytoextraction offers the advantage of being applicable in situations where the mobility and phytoavailability of the pollutant are extremely low. For instance, in the case of Pb or U, plant uptake is limited by the extremely low rate of diffusion of these elements from the soil-to-root surface. In these situations, the application of a mobilizing agent that increases metal solubility and mobility can overcome the diffusion limitation in the plant uptake process. Another scenario where chemically enhanced phytoextraction may be necessary is represented by soil contaminated with elements for which there are no known hyperaccumulator plants as is the case for radionuclides. An additional beneﬁt

28

S. P. McGRATH ET AL.

could be that this approach reduces the potential risk related to the presence of metal-rich plants in the ﬁeld for long periods of time (Huang et al., 1997). Chemically enhanced phytoextraction has been primarily developed for remediation of Pb-contaminated soils but successive studies have focused on other elements or mixtures of elements.

B. CHEMICALLY ENHANCED PHYTOEXTRACTION OF LEAD Lead is one of the most important and common inorganic pollutants present in the environment. Public concern has risen due to several studies conducted in the United States in the late 1980s and early 1990s. These studies identiﬁed the exposure of children to Pb as the cause of adverse effects including deﬁcits in growth and stature, subtle learning disabilities, and hearing impairments (ATSDR, 1988; Needleman et al., 1990). Lead is present at high levels in 47% of the contaminated Superfund sites for which the Environmental Protection Agency of the United States has signed the Records of Decisions (USEPA, 1997). Lead in soil is strongly retained by organic matter, oxides, and clays, and forms fairly insoluble carbonate and phosphate compounds. As a consequence, its solubility and mobility in the soil are generally low. Lindsay (1979) reports a maximum activity of Pb2+ in the soil solution of approximately 10−8.5 M. The uptake and accumulation of Pb by plants are low even in polluted soils, generally less than 50 mg kg−1 (Cunningham et al., 1995). Moreover, Pb is usually conﬁned to the roots with minimal transport to the shoot (Dushenkov et al., 1995). Several studies have shown that mobilizing agents can effectively increase the mobility of Pb in the soil. Several chelating agents have been tested, the most effective having a log of the binding constant higher than 18. Huang et al. (1997) tested ﬁve chelating agents for their effectiveness in increasing Pb desorption from a Pb-contaminated soil collected from an industrial site. EDTA was the most efﬁcient mobilizing agent followed by HEDTA and DTPA. They also reported Pb desorption to be closely correlated with shoot Pb concentration in corn. Similarly, Blaylock et al. (1997) compared ﬁve chelating agents for their ability to enhance Pb mobilization and accumulation in B. juncea. EDTA produced both the largest increase in soluble Pb and the highest accumulation of Pb in the shoot of B. juncea. In the investigation by Cooper et al. (1999) HEDTA was able to desorb more Pb than any of the other six chelating compounds investigated, although EDTA was not tested. The relationship between Pb solubility in soil and plant uptake is shown for different combinations of chelating agents and plants in Fig. 9. In all the papers quoted previously, Pb desorption increased with application rate. However, to maximize Pb accumulation by plants and reduce the environmental risk of leaching, an application rate should be selected which maximizes the concentration of the Pb–ligand complex based on the extractability of Pb by the chelating agent chosen (Epstein et al., 1999).

Figure 9 Relationship between extractable Pb after application of chelating agents and concentration of Pb in the shoot of various plants.

30

S. P. McGRATH ET AL.

Wu et al. (1999) pointed out that EDTA is not the ideal chelating agent for a number of reasons. First, Pb–EDTA is a highly soluble complex and can therefore be leached to groundwater. Furthermore, this complex is not ideally suited in terms of plant uptake and translocation. Finally, EDTA is not speciﬁc for Pb binding and other ions present in the soil in large concentrations; for example, Fe3+ can compete with Pb for chelation sites on the molecule. The ideal chelating agent should be speciﬁc for Pb, water soluble in its free form allowing easy application, and be able to form a more lipophilic Pb complex to decrease the risk of leaching and to be easily absorbed by plants. Wu et al. (1999) compared several natural or synthetic chelates, some of which were rather speciﬁc for Pb, for their ability to solubilize Pb in soil. These chelates included compounds with sulfur-containing structures, macrocyclic structures, and small acid molecules. However, none of these compounds produced a large increase in Pb solubility compared with the untreated control. In a second experiment they compared six synthetic chelating agents containing at least two carboxymethyl groups. Both EDTA and HBED were able to increase Pb solubility, but only EDTA maintained a high, soluble Pb concentration during the 5-day experiment. This could be due to the fact that HBED has a much greater afﬁnity for Fe than for Pb and eventually forms a stable complex with Fe over time. HBED forms a complex with Pb that is less hydrophilic than Pb–EDTA. In a hydroponic experiment, more HBED–Pb was taken up by maize plants in comparison to Pb–EDTA, but it was not so efﬁciently translocated to the shoot. This could be explained by the fact that only EDTA increased the transpiration rate in maize compared to the control, possibly due to toxicity. However, Wu et al. (1999) concluded that it may be possible to design a more efﬁcient ligand for the phytoremediation of Pb by optimizing its speciﬁcity for Pb and the lipophilicity of the Pb–chelate complex. Once Pb is chelated by EDTA, or any other chelating agent, it has to move from the external part of the root to the xylem vessels. Two pathways are possible: the solute can move via the apoplast or enter the symplast by crossing the cell membranes. The Casparian strip represents a barrier to the movement in the apoplast. However, the presence of large concentrations of chelating agent and Pb could disrupt the physiological barriers that control root uptake under normal conditions. Vassil et al. (1998) reported that the kinetics of Pb and EDTA accumulation were biphasic, indicating that a threshold concentration of the chelating agent is required to obtain accumulation of Pb in the shoot. They suggested that synthetic chelates could disrupt the normal function of cell membranes by removing Zn and Ca ions that are involved in the stabilization of plasma membranes. This would explain results obtained in both hydroponic and soil systems that show that Pb is taken up as a Pb–EDTA complex (Epstein et al., 1999; Vassil et al., 1998). Several plants have been used in combination with chelating agents for Pb phytoextraction. The ideal plants should be fast growing and produce a large biomass while accumulating high concentrations of Pb. Furthermore, they should

PHYTOREMEDIATION

31

be tolerant enough to grow in contaminated soils and have well-established agronomic systems. B. juncea possesses several of these characteristics and is the most commonly used plant species in this approach. Efforts have been directed to select Pb-accumulating mutants of this plant (Schulman et al., 1999). However, dicotyledon species are usually less tolerant to metals than monocotyledons (Marschner, 1995) and Brassica could suffer metal toxicity in severely contaminated soils. This problem is less severe in Pb-only contaminated soils due to the low phytoavailability of this metal, but may represent a serious limitation in the more frequent case of soils that are co-polluted with more phytotoxic elements such as Cu and Zn. Several authors have reported reduction in both transpiration rate and biomass produced by B. juncea as a consequence of large concentrations of Pb and chelating agents (Vassil et al., 1998; Epstein et al., 1999). Other plant species such as maize and pea have also been used (Huang and Cunningham, 1996; Huang et al., 1997). To be economically viable, plants used for Pb extraction should be induced to accumulate at least 10,000 mg kg−1 of Pb in their shoots and achieve a biomass of 20 t ha−1 (Huang and Cunningham, 1996). Blaylock et al. (1997) reported the concentration of Pb in shoots of B. juncea of up to 15,000 mg kg−1 in a pot experiment. In this trial EDTA was applied to a soil amended with Pb carbonate to obtain a Pb concentration of 600 mg kg−1 in the medium. In the same paper they conducted a ﬁeld study at a former industrial site contaminated with 1200 mg Pb kg−1. In this more realistic situation, B. juncea accumulated only 785 mg kg−1 when EDTA was applied. The combined application of EDTA and acetic acid to reduce soil pH increased Pb accumulation in plants to 1475 mg kg−1. Huang et al. (1997), in a pot experiment, showed that when EDTA was applied to a contaminated soil, it could increase the Pb concentration in maize and pea shoots from less than 500 mg/kg to more than 10,000 mg kg−1. The variable results reported in different studies could be interpreted by taking into account that the characteristics of polluted soil and the form of lead present may play important roles in the potential for Pb removal (Cooper et al., 1999). Studies conducted with soils artiﬁcially contaminated with Pb may lead to an overestimation of the potential Pb removal due to the high solubilization of some forms of lead, e.g., carbonate, by EDTA. More recalcitrant forms of lead could be dissolved with more difﬁculty by chelating agents and therefore higher amounts of mobilizing agents may be required to achieve the same plant uptake (Epstein et al., 1999). Two ﬁeld demonstration trials were reported by Blaylock and Huang (2000). At both sites three crops of B. juncea were grown and treated with 2 mmol EDTA kg−1 to induce Pb uptake. The ﬁrst trial was conducted at an industrial site in Bayonne, New Jersey. Due to the shallow water table and potential ﬂooding this study was conducted in a large lysimeter. The soil treated was alkaline (pH 7.9) and contained an average of 2055 mg Pb kg−1 mainly in the carbonate form. After phytoextraction, the Pb concentration in the top soil decreased to 960 mg kg−1. The second trial

32

S. P. McGRATH ET AL.

was conducted at Dorchester, Massachusetts, in an urban residential area. In this case the soil was acidic (pH 5.1–5.9) and Pb was mainly associated with organic matter. After three croppings of B. juncea, the concentration of Pb in the top soil was reduced from 984 to 644 mg kg−1 while it slightly increased at 15 to 30 cm depth. Even though the regulatory limit of 400 mg Pb kg−1 was not met at either of the two sites, the apparent decreases in Pb concentration obtained were substantial. However, in these studies no attempts were made to obtain mass balances for the metals.

C. CHEMICALLY ENHANCED PHYTOEXTRACTION OF OTHER HEAVY METALS The same approach used for Pb has been also tested for other heavy metals. Phytoextraction of Cd was investigated using B. juncea in combination with different chelating agents (Blaylock et al., 1997). A soil was amended with carbonate forms of the Cd and B. juncea grown. Three weeks after seedlings emerged CDTA, DTPA, EDTA, and EGTA were applied to the soil. EGTA was the most effective in inducing Cd accumulation in plant shoots, at concentrations up to 2800 mg kg−1 from the soil spiked with Cd (100 mg Cd kg−1). Simultaneous extraction of Cd, Cu, Ni, Pb, and Zn was tested in a second experiment using B. juncea. EDTA application induced accumulation of 1000 mg kg−1 or more of Cu, Pb, and Zn, whereas Cd and Ni uptake was signiﬁcantly smaller. Ebbs and Kochian (1998) compared B. juncea, oat, and barley for their potential to phytoextract Zn from a contaminated soil. Application of EDTA caused a large increase in Zn solubility and signiﬁcantly increased the Zn uptake in B. juncea. This increase was much smaller than that in the study conducted by Blaylock et al. (1997) even though the Zn concentration in the soil was much higher. These results indicate that real contaminated soils may contain metals in forms more difﬁcult to mobilize and extract than soil amended with metals in forms of carbonates or other salts. Surprisingly, EDTA had no effect in terms of Zn accumulation by oat and barley. However, barley accumulated two to four times more Zn than B. juncea in combination with EDTA. Deram et al. (2000) studied the use of the metal-tolerant grass Arrhenatherum elatius, in combination with EDTA, to extract Cu, Co, and Ni from a modiﬁed copper ore. EDTA induced accumulation of all the metals in comparison with an untreated control. As a result the plants showed necrosis most likely due to the large concentrations of Cu (7500 mg kg−1), Co (175 mg kg−1), and Ni (1276 mg kg−1). However, the bioaccumulation factor was signiﬁcantly larger than 1 only for Cu (metal in plant/ore concentration = 40). A ﬁeld trial to examine the potential of simultaneously phytoextracting metals from a contaminated calcareous soil was conducted at Dornach in Switzerland

PHYTOREMEDIATION

33

(Kayser et al., 2000). Elemental sulfur and nitrotriacetate (NTA) were used as metal-mobilizing agents. Both amendments increased the solubility Zn, Cd, and Cu in the soil by factors of 21, 58, and 9, respectively. However, this was not translated into a correspondingly large increase in metal uptake by the plant species grown (A. murale, T. caerulescens, four agricultural crops and Salix viminalis). None of the species used in this experiment accumulated more than 10–30% of the maximum metal concentrations reported in the literature. At present, it is difﬁcult to judge whether or not chemically enhanced phytoextraction can be an economically and technically feasible option to remediate soils that are polluted with several heavy metals. The information available is much more limited than in the case of Pb phytoextraction, and more potential problems have to be considered. For instance, metals such as Cu and Zn are more phytotoxic than Pb, and, in order to grow, the plant species used will have to be tolerant to all of the contaminants present. It also seems that more research is needed to identify mobilizing agents that can simultaneously induce accumulation of several metals in plants.

D. CHEMICALLY ENHANCED PHYTOEXTRACTION OF RADIONUCLIDES Soils contaminated with radionuclides often contain minute concentrations of these contaminants but these translate into activity levels that can cause concern (Negri and Hinchman, 2000). The characteristically low concentration of pollutants and the general immobility of radionuclides in the soil are two factors that make the use of chemically induced phytoextraction an attractive option to remediate large areas of contaminated land. Several studies,which have been summarized by Negri and Hinchman (2000) and Shahandeh et al. (2000), at both greenhouse and ﬁeld scale have been conducted to evaluate the removal of U, 137Cs, 90Sr, and 3H from soil and water. Radiocesium is one of the radionuclides of major concern. This element is fairly insoluble in soil due to its strong retention by clay minerals (Cornell, 1993). In these minerals 137Cs ions are selectively sorbed, similarly to K+ and NH4+, in the interlayer spaces (wedge zones) where they lose their hydration and become speciﬁcally sorbed. However, once taken up, Cs is highly mobile within the plant and can be transferred to the shoot (Resnik et al., 1969). Screening of potential mobilizing agents for 137Cs have been conducted to investigate the efﬁciency of chelating agents, reducing agents, strong mineral acids, ammonium and potassium salts (Blaylock and Huang, 2000; Cornish et al., 1995; Dushenkov et al., 1999; Lasat et al., 1998a,b). Ammonium and K salts have been shown to be the most effective mobilizing agents. For instance, Ebbs et al. (2000) reported a solubilization of 10–25% of the total 137Cs (total in soil 400 pCi g−1) from a contaminated

34

S. P. McGRATH ET AL.

soil obtained from Brookhaven National Laboratory (BNL). In another study, the 137 Cs in the soil solution increased from 0.4 to 4.5 pCi g−1 upon application of 20 mmol kg−1 ammonium sulfate to a soil containing 370 pCi g−1 (Blaylock and Huang, 2000). In a pot study Lasat et al. (1997) showed that the application of 40 or 80 mmol NH4+ per kg of soil resulted in a 2- to 12-fold increase in the accumulation of 137 Cs in different plants. The best results were obtained with cabbage (B. oleracea) for which a bioaccumulation ratio of 2–3 was observed. A ﬁeld trial was conducted at BNL on a soil contaminated with 137Cs in the range of 80 to 900 pCi g−1 (Lasat et al., 1998). Three plant species (B. juncea, Amaranthus retroflexus, and Phaseolus acutifolius) and three additions of NH4NO3 (0, 0.1, and 0.2 M) were compared. The best results were obtained with A. retroflexus which, in 3 months, removed approximately 3% of the total 137Cs from the top 15 cm of the soil proﬁle. Interestingly, no signiﬁcant increase in 137Cs extraction was observed upon application of the NH4+ salt. The authors suggested either that NH4+ or the mobilized 137Cs may have been ﬂushed through the soil proﬁle by the combined effect of irrigation and rainfall occurring after the application of the amendment. Alternatively, nitriﬁcation of the NH4 may have occurred before it was effective in the soil proﬁle. However, the results obtained do indicate that with two crops grown every year on the contaminated soil, nearly 75% of the total radiocesium could be removed in approximately 15 years. Another ﬁeld trial is ongoing at a second site at BNL. In this case the soil is less polluted (approximately 100 pCi g−1), and three species of Amaranthus are being compared with or without combinations of ammonium and manure applications (Ebbs et al., 2000). Uranium is another radionuclide of environmental signiﬁcance. In the soil it is mainly present as a uranyl cation (UO2 2+ ) and is generally adsorbed by clay minerals, Fe–Mn oxides, and organic matter. Several studies have investigated the ability of different plant species to accumulate U. Hossner et al. (1999) reported that plant species from the Brassicaceae family can accumulate more U than grasses. Among the species screened B. juncea and Helianthus annuus accumulated the highest concentration of U in their shoot. However, most of the U remained in the roots, and the concentration ratio between U in the soil and in the shoots was <1. Similarly, Saric et al. (1995) found that sunﬂower and leafy vegetables accumulated more U than other plants. The application of mobilizing agents to increase U uptake by plants and translocation to the shoot has been investigated. However, in the case of U and Pu the mechanism of action of the mobilizing agents is not well understood (Shahandeh et al., 2000). Huang et al. (1998) compared chelating agents, inorganic and organic acids for their ability to dissolve U in contaminated soils. Citric acid (at an application rate of 20 mmol kg−1) provided the best results with a 200-fold increase in the U concentration in the soil solution

PHYTOREMEDIATION

35

after application of the amendment. The application of citric acid decreased the soil pH by 0.5–1.0 units. However, the mobilization of U was larger than that in the case of acidiﬁcation by nitric or sulfuric acid indicating that chelation of U by citric acid is the predominant mechanism in mobilization. Citric acid also induced an increase in U uptake by plants. Among the species screened B. juncea, B. chinensis, Amaranthus cruentus, and Beta vulgaris were induced to accumulate the largest concentrations of U with an increase of up to 1000 times in response to application of citric acid (Ebbs et al., 1998; Huang et al., 1998). Hossner et al. (1999) compared citric acid, EDTA, and oxalic acid for their ability to increase U uptake by sunﬂower grown in two calcareous and two acidic contaminated soils. In the case of the two calcareous soils, citric acid (at an application rate of 16 mmol kg−1) induced the largest increase in U uptake by sunﬂower. However, they observed only a two to three times increase after application of citric acid which is much lower than the increase observed by Huang et al. (1998). In contrast, in the acidic soils, EDTA was the most effective amendment. These results indicate that the speciation of U in the soil may not only inﬂuence its mobility but also the complexation by different chelating agents and the process of plant uptake. Because engineering technologies used to clean up U-contaminated soils cost in the order of $2500 m−3, induced phytoextraction could be a viable alternative (Huang, 2000). Citric acid represents an ideal chelating agent. This organic acid induces very rapid accumulation of U in plants, and the period during which U-enriched plants remain in the ﬁeld is limited. Furthermore, its rapid degradation (within a few days) and the resulting readsorption of U in the soil reduce the environmental hazard related to the potential leaching of U to groundwater. Although no studies have been conducted to directly investigate the potential of phytoextraction for remediating Pu-contaminated soils, several results indicate that the application of chelating agents could enhance accumulation of this radionuclide in plants. For instance, Lipton and Goldin (1976) reported an increase in the Pu concentration in pea plants of up to 1000 times as a result of DTPA application to a contaminated soil. The ability of DTPA to enhance Pu uptake is also conﬁrmed by the work of Ballou et al. (1978) and Vyas and Mistry (1983).

E. CHEMICALLY ENHANCED PHYTOMINING Phytomining involves the use of plants to concentrate valuable metals that can be subsequently recovered through hydrometallurgical or pyrometallurgical processes. This approach was ﬁrst described for Ni due to the availability of numerous Ni hyperaccumulator plants and large areas of serpentine soils worldwide (see earlier). A similar approach was also proposed for precious metals such as Au.

36

S. P. McGRATH ET AL. Table I Metal Concentration in Plants (in mg kg−1) Required to Obtain a Crop with a Value of $US 500 ha−1 as Influenced by Value of the Metal and Biomass Produceda Biomass t ha−1

Palladium Platinum Gold Thallium Silver Nickel Tin Copper Zinc Cadmium Lead

$US t−1

1

5

10

30736000 19612000 8775500 1322400 154000 7135 5125 1804 1028 551 471

16.2 25.5 57.0 378 3246 70,077 97,561 277,162 486,381 907,184 106,1571

3.2 5.1 11.4 75.6 649 14,015 19,512 55,432 97,276 181,436 212,314

1.6 2.5 5.7 37.8 324 7007 9756 27,716 48,638 90,718 106,157

20 0.8 1.3 2.8 18.9 162 3503 4878 13,858 24,319 45,359 53,078

a The values of the metals refer to December 2000. The potential sale of energy recovered from biomass combustion to offset the cropping costs is not included.

The economic feasibility of this technique depends on the value of the metal, the biomass produced, and the concentration of the metal in the biomass (Table I). However, no hyperaccumulator plants are reported for these elements, with the exception of Tl. For this reason Anderson et al. (1998) proposed the use of mobilizing agents for the phytomining of precious metals and in particular of Au. Gold concentrations in plants are generally below 10 ng g−1. In a series of experiments Anderson et al. (1998) applied ammonium thiocyanate to different types of ore or to sand amended with colloidal gold and measured the concentration of Au in different plant species grown in these substrates. All the plant species tested showed hyperaccumulation of Au (Au concentration >1 μg g−1. One plant of B. juncea had 57 μg Au g−1 of dry weight when grown in sand amended with 5 μg Au g−1. It was calculated that a concentration of about 3 μg Au g−1 in the plant and a crop yield of 20 t ha−1 would be necessary to obtain a return of $500 ha−1 (Anderson et al., 1999). This process could be applied to extract Au from substrates that contain metal concentrations too low to be extracted economically using conventional techniques. It could also be a part of programs aiming to revegetate mine tailings or piles of low-grade ore (Anderson et al., 1999). However, even though the toxicity of ammonium thiocyanate is lower in comparison to cyanide, there are environmental concerns related to the application of this compound in situ. Anderson et al. (1998) concluded that this approach could be extended to other noble metals such as Pt, Pd, and Ag.

PHYTOREMEDIATION

37

F. CHEMICALLY ENHANCED PHYTOEXTRACTION VERSUS NATURAL HYPERACCUMULATION Only a few studies have directly compared phytoextraction of metals using hyperaccumulators and high biomass plants in combination with a chemical treatment to induce metal accumulation. Kayser et al. (2000) compared, in a ﬁeld experiment (Dornach, NW Switzerland), the performance in terms of metal extraction of two hyperaccumulator plants (A. murale and T. caerulescens), four agricultural crops, and Salix viminalis. Elemental S and NTA were used to enhance metal accumulation. These treatments increased the solubility of Zn and Cd by factors of 21 and 58, respectively, but the uptake of these metals by plants was increased only by a factor of 2–3. The two hyperaccumulator species produced a low biomass, and the concentration of metals in their shoots, although larger than that in non-hyperaccumulator plants, was smaller than the concentrations reported in the literature for these plants. The authors suggested that the compacted soil conditions and the hot periods at the beginning of the growing season could have impaired the performance of the hyperaccumulators. Also, these results could be due to the presence of large concentrations of Cu in the contaminated soil. T. caerulescens is not Cu tolerant (McLaughlin and Henderson, 1999) and may have suffered from Cu toxicity. The largest amounts of Cu and Zn were removed by sunﬂower (H. annuus), whereas tobacco (Nicotiana tabacum) removed the largest amount of Cd. However, the removal rate achieved in this study was insufﬁcient to make phytoextraction practicable under those particular ﬁeld conditions. We conducted a pot experiment to compare natural phytoextraction using the Zn/Cd hyperaccumulator T. caerulescens versus chemically enhanced phytoextraction using maize (Z. mays) treated with EDTA (Lombi et al., 2001b). The study used an industrially contaminated soil and an agricultural soil contaminated by sewage sludge. Three croppings of T. caerulescens removed more than 8 mg kg−1 Cd and 200 mg kg−1 Zn, representing 43 and 7% of the two metals in the soil. In contrast, high concentrations of Cu in the agricultural soil severely reduced the growth of T. caerulescens, thus limiting its phytoextraction potential. The EDTA treatment greatly increased the solubility of heavy metals in soils (Fig. 10), but this did not result in a large increase in metal concentration in the maize shoots. This is in agreement with the results obtained by Kayser et al. (2000). Phytoextraction of Cd and Zn by maize + EDTA was much smaller than that by T. caerulescens from the industrially contaminated soil, and was either smaller than (Cd) or similar to (Zn) in the agricultural soil. The results reported previously indicate that either the chemically enhanced phytoextraction or the use of hyperaccumulator plants can be the most appropriate approach depending on the soil conditions and the pattern of pollution present in different sites. The use of natural phytoextraction is limited to situations where all

38

S. P. McGRATH ET AL.

Figure 10 Concentration of Cu and Cd in soil pore water extracted from two contaminated soils. Arrows indicate application of EDTA treatments. Soil A was contaminated by sewage sludge applications; Soil B was contaminated by a nearby Zn smelter.

of the pollutants present can be tolerated by the hyperaccumulator plants. On the other hand, soils contaminated with highly insoluble metals such as Pb or U may require chemical enhancement of accumulation.

G. POSSIBLE CONCERNS RELATING TO THE USE OF CHELATING AGENTS The marked increase in soluble metals in soil pore water following the application of chelating agents may pose a serious threat in terms of the leaching of heavy metals to ground water. Cooper et al. (1999) reported that the application of 20 mmol kg−1 of CDTA or DTPA greatly increased the leachability of Pb as measured by the Toxicity Characteristic Leaching Procedure (TCLP) to values above the USEPA regulatory limit (5 mg liter−1). Furthermore, they also observed a shift in the distribution of Pb in soil from more recalcitrant to more soluble forms. In a study we recently conducted (Lombi, 2001b), an application of EDTA (2.7 mmol kg−1) to two metal-contaminated soils planted with maize caused a large increase in metal concentrations in the soil pore water that persisted for several weeks after the application of the chelating agent. In fact, 5 months after EDTA application (which was the time interval between EDTA applications) metal–EDTA complexes were still found in the soil pore water (Fig. 10). This is in agreement with the work of Hong et al. (1999), who reported that EDTA is relatively biologically stable even under conditions favorable to biodegradation. Satroutdinov et al. (2000) showed that metal–EDTA complexes with high-stability constants (i.e., chelates of Cu, Fe, Pb, Zn) are degraded more slowly than complexes with low-stability constants (i.e., chelates of Ca, Mg, Mn).

PHYTOREMEDIATION

39

The results discussed here suggest that the application of EDTA may, in the future, be limited to ex situ conditions where control of the metal-enriched leachates can be safely achieved. However, mobilizing agents that are effective in terms of inducing hyperaccumulation of metals and that are not harmful to the environment may be discovered and tested in the future. For instance, compounds that are easily degradable in soil, such as citric acid used for U phytoextraction, could represent a useful alternative if they are able to enhance accumulation of pollutants in plants. Another option is represented by the use of plants genetically modiﬁed or naturally able to exude mobilizing compounds in their rhizosphere.

IV. PHYTOVOLATILIZATION Volatilization is based on different biological processes including reduction to volatile elemental forms and synthesis of methylated compounds of some metals and metalloids (Wenzel et al., 1999). In particular, this approach has been proposed and investigated to remediate soil polluted with As, Hg, and Se. Methylated volatile compounds of As and Se are relatively non-toxic, and therefore most of the research efforts have been focused on these compounds. In contrast, methylated forms of Hg are extremely toxic and the only acceptable volatilization of Hg is through reduction of the ion Hg2+ to gaseous elemental Hg. All these elements can be transformed by soil microorganisms into volatile compounds, but only Se volatilization by plants has been proved to occur, even though it is difﬁcult to estimate the relative importance of this pathway under natural conditions. However, plants have been genetically transformed by insertion of bacterial genes that allow the volatilization of Hg0.

A. SELENIUM Most of the studies by far have been dedicated to Se volatilization. However, the volatilization of Se in the soil–plant system is very complex and microorganisms play an important role in it. Soil organisms can directly produce volatile Se compounds such as dimethylselenide (DMSe) and dimethyldiselenide (DMDSe) but seem also to facilitate the uptake and volatilization of Se by plants (de Souza et al., 1999a,b). Selenium in soil undergoes a series of microbial transformations among which methylation is thought to be a protective mechanism (Losi and Frankenberger, 1997). This process is carried out by both bacteria and fungi. Bacterial methylation seems to be favored by anaerobic conditions and may be employed in constructed wetlands. Azaizeh et al. (1997) showed that rhizosphere bacteria are predominant in volatilizing Se in wetlands. They reported that rates of Se volatilization were higher in rhizosphere sediment samples than in

40

S. P. McGRATH ET AL.

non-vegetated soil samples. Similarly, rhizosphere bacteria (and not fungi) seem to be able to enhance selenate volatilization (Zayed and Terry, 1994). Experiments with the antibiotic ampicillin demonstrated that rhizosphere bacteria facilitated 35% of plant Se volatilization and 70% of Se accumulation in plants (de Souza et al., 1999a). In another experiment comparing plants, either grown under axenic conditions or inoculated with rhizosphere bacteria, the authors showed that inoculation increased Se concentrations in the plants and Se volatilization by a factor of 5 and four-fold, respectively. The bacteria may facilitate the maintenance of a high Se concentration in the roots, a factor that has been indicated to be a limiting factor in the process of volatilization. A heat-labile proteinaceous compound produced by the plant–bacteria interaction may be responsible for enhanced Se uptake and volatilization. This compound has not yet been identiﬁed. It can be produced by either the plant or the bacteria and may act by converting selenite to SeMet which is readily volatilized by plants (Zayed et al., 1998). Alternatively, bacteria may stimulate selenate uptake by enhancing the concentration of O-acetylserine in the rhizosphere (Terry et al., 2000). This compound regulates, together with the S status of the plant and glutathione (GSH), the expression of S transporter genes (Davidian et al., 2000). It has been proposed that selenate is taken up by plants through a sulfate transporter (e.g., Arvy, 1993; Breton and Surdin-Kerjan, 1977). Once selenate has been taken up it is translocated to the shoot much more easily than selenite or organic Se compounds (Arvy, 1993; Zayed et al., 1998). Then selenate enters the leaf chloroplasts where it undergoes a series of metabolic transformations mediated by sulfate assimilation enzymes. The metabolic pathways that lead to the formation of the principal volatile Se compound are shown in Fig. 11. This is derived from Terry et al. (2000) where more detailed information is available. Large differences exist in the rate of Se volatilization between both terrestrial and aquatic plant species. Pilon-Smits et al. (1999) screened 20 aquatic plant species and found that the best volatilizing species were also the best performers in terms of Se accumulation. In particular, Typha latifolia and Scirpus robustus were able to volatilize both selenate and selenite at high rates per unit of (water) surface area. Among terrestrial plants, several members of the Brassicaceae family, which accumulate S, can also accumulate and volatilize Se very efﬁciently. In particular, Indian mustard (Brassica juncea) is considered to be the best terrestrial plant for Se phytoremediation (Zayed et al., 1999). Lin et al. (2000) measured the annual Se volatilization in a ﬁeld trial with Salicornia bigelovii in the San Joaquin Valley, California. Biological volatilization removed 62 mg Se m−2 y−1 representing 6.5% of the annual Se input (958 mg Se m−2 y−1. This removal is smaller than estimates from other studies (Ba˜nuelos et al., 1997). However, it should be considered that these are data obtained under ﬁeld conditions and that the annual input at that particular site is quite large. Probably the best example of Se volatilization by wetlands is the Chevron Oil Water Enhancement Wetland at Point Richmont, California. Hansen et al. (1998)

PHYTOREMEDIATION

41

Figure 11 Biochemical pathway of Se volatilization in plants (modiﬁed from Terry et al., 2000, with permission from Annual Reviews).

reported that almost 90% of the Se entering the wetlands was removed from water. This was due to both accumulation in plants and sediments and biovolatilization. The latter process accounted for 10–30% of the Se removed.

B. MERCURY In addition to the study of Se, much current research also concentrates upon the potential for using the volatilization process to extract or eliminate Hg from contaminated soils. This requires the conversion of toxic organic or ionic mercury compounds to metallic mercury (Hg0) which is relatively inert and far less toxic to organisms.

42

S. P. McGRATH ET AL.

Bacteria containing the mercury resistance (mer) operon of genes have the ability to carry out this detoxiﬁcation, allowing them to colonize mercury-contaminated soils. The mer operon encodes a battery of genes responsible for a number of processes including the detection of mercury, its transport through membranes, and sequestration in the periplasmic space outside the cell. It is merA that is of most importance, encoding the enzyme, mercuric reductase, which enables Mer+ bacteria to reduce Hg2+ to Hg0 (Heaton et al., 1998). The resulting Hg0 rapidly evaporates through the bacterial cell membranes. A subset of the mer operon that contains the gene merB has also been identiﬁed. The enzyme encoded by this gene (organomercurial lyase) catalyzes the breakdown of organic mercury to Hg2+ and a reduced carbon compound (Heaton et al., 1998). It is thought that by genetically engineering plants to contain the two mer genes (merA and merB) that give bacteria this broad spectrum resistance to a variety of mercury compounds, the biochemical detoxiﬁcation pathway will be expressed in plants enabling them to remove mercury and, in turn, detoxify it. The resulting Hg0 would be expelled into the atmosphere and/or translocated to the shoots. If successful, this would provide an environmentally friendly, in situ detoxiﬁcation strategy. Transgenic Arabidopsis thaliana seedlings expressing merA9 (a modiﬁed merA sequence compatible with plant expression) were shown to be viable on media containing HgCl2 concentrations of 25–100 μM, levels considered to be toxic to the majority of plant species (Rugh et al., 1996). For transgenic plants to be suitable for phytoremediation, they must be able to grow on polluted soils. Consequently research has been extended to larger plants that will be more suitable for remediating mercury-contaminated sites. Tobacco (merA18) and Brassica (merA9) have been engineered to express the merA gene and have shown resistance to Hg2+ levels that killed wildtype control plants. In one experiment using tobacco (Heaton et al., 1998), merA and wildtype seedlings were exposed to a hydroponic solution containing 5 μM Hg2+. The merA plants were found to reduce almost 40% of the mercury after 24 h and removed it from the system without indicating stress. Between 75 and 80% of the Hg2+ had been removed from the system after a week. The concentration of mercury in the plant tissue was also found to decline over the experimental time course. In comparison, the non-transformed plants had reduced approximately 20% of the mercury by the end of the time course, having initially exhibited signs of stress. Yellow poplar (Liriodendron tulipifera) plants have been the basis of further experiments, and mer18 plantlets of this species released Hg0 at approximately 10 times the rate of their non-transformed counterparts (Rugh et al., 1998). Further studies have been undertaken in which model plants have been engineered to contain the merB gene. The resulting transgenic plants are capable of detoxifying a variety of organic mercury compounds. Most importantly, these investigations indicate that genetically engineered plants containing both merA and merB genes should have the ability to remove and detoxify a range of mercury compounds (Bizily et al., 1999). For example, A. thaliana plants expressing both genes

PHYTOREMEDIATION

43

are able to survive when exposed to 50-fold higher methylmercury concentrations than equivalent wildtype plants (Bizily et al., 2000). In time, plants transformed with mer genes may aid in removing mercuric compounds from contaminated soils by either translocating Hg2+ to their shoots (especially appropriate where release into the atmosphere is not acceptable) or by releasing Hg0 by phytovolatilization. Future concerns will focus upon engineering appropriate plant species and establishing an acceptable rate of Hg0 release into the environment, keeping atmospheric levels below regulatory guidelines (Meagher and Rugh, 1996). Experiments involving the removal of mercury have mainly been based on hydroponic systems. For transgenic plants to be most beneﬁcial they must be able to remove mercury compounds from the soil. Uptake, reduction, and volatilization from soil are likely to be slower than the rates of removal from hydroponic media, and this requires further research. Phytovolatilization has not yet been developed for As. This is surprising because of the similarities among Hg, Se, and As (McGrath, 1998; Wenzel et al., 1999). Phytovolatilization implies the ability of plants to volatilize As. Yet, this process has not been shown to occur in plants. Plants may assist and stimulate microbial volatilization via rhizospheric interactions with microorganisms. Microbial methylation of As has been known for a long time and is common to both bacteria and fungi. Bacterial methylation seems to be favored by anaerobic conditions and may be employed only in ex situ bioreactor systems (McBride and Wolfe, 1971). Fungal methylation seems to be important in the volatilization of As compounds used in agriculture (Baker et al., 1983; Cullen et al., 1984). It may be possible to adopt a strategy similar for As to the one adopted for Hg. Most likely, As is taken up by plants through their phosphate transporter (Lee, 1982), and methylation produces compounds that are less toxic than As (III) (Hindmarsh and McCurdy, 1986). Phytovolatilization of potentially toxic metals and metalloids has given rise to some concerns related to the release of these elements in the atmosphere. However, methylated forms of As and Se are much less toxic than the inorganic forms generally present in contaminated soils. Frankenberger and Karlson (1988) have calculated that if a Se volatilization rate of 250 μg m−2 h−1 is achieved, the maximum exposure would be 837 ng m−3 air which is much lower than acceptable levels. Studies conducted in the San Joaquin Valley in California have shown that during the 9.6-day lifetime of DMSe in the atmosphere, this volatile compound is diluted in the atmosphere and possibly deposited in Se-deﬁcient areas (Atkinson et al., 1990; Lin et al., 2000). The same applies for Hg0 which is less toxic than Hg2+ or CH3Hg+. Elemental Hg, with a half-life in the atmosphere in the order of years, is also much more stable in the atmosphere than all the other Hg species (Keating et al., 1997). Therefore, metallic Hg0 can be largely diluted into a large atmospheric pool, whereas any other form of Hg would be deposited near their sources (Meagher et al., 2000; Nriagu, 1979).

44

S. P. McGRATH ET AL.

V. SUMMARY AND FUTURE DIRECTIONS As mentioned earlier, it is clear that phytoremediation is a relatively young area in terms of both science and technology. Early in the 1990s there was a rush to try natural or chemically enhanced phytoremediation in the ﬁeld. It is now obvious that the success of those attempts was hindered by the absolute lack of basic knowledge of the growth requirements of hyperaccumulators and of which chemicals and combinations are best for the enhancement of metal concentrations in non-accumulator crops. Added to this, little knowledge existed concerning the basic mechanisms of uptake of these elements by plants and of the transport and sequestration within the tissues. In particular Cu and Cr appear to be difﬁcult for phytoextraction. In the case of Cu, there are doubts about the ability of any species to hyperaccumulate this metal, and so only the chemical enhancement route is showing any potential for development at present (Blaylock et al., 1997). Phytotoxicity of Cu in highly contaminated soils may present one of the main barriers to phytoextraction, but for Cr almost the opposite is true. This element is usually very insoluble in soils and present in waters at low concentrations. Hyperaccumulation of Cr has not been reported and inducing plants to take up Cr from soils is very difﬁcult, so the phytoextraction potential appears to be low. For chemically enhanced accumulation, we now have more knowledge regarding which chelates are best for the release of metals such as Pb from contaminated soils (Wu et al., 1999), but the exact mechanism of the induced accumulation in plant shoots is shrouded in mystery. Practitioners use cocktails of herbicides and other compounds during the process, and it is not clear which ones, including the chelates, are responsible for overcoming the usual mechanisms that restrict metal uptake by plants. Therefore, more recent research has been focused on gaining detailed knowledge of how plants obtain large concentrations of metals from soils at the rhizosphere and root level, along with studies about the way plants move the metals around inside the tissues and of the storage and detoxiﬁcation mechanisms. This research is not yet complete; however, great progress has been made in the understanding of metal hyperaccumulation in recent years, particularly for Zn and Ni. The real question is how this information can be used to improve the practical application of phytoextraction in future. So far, only the genes for Zn transporters in the hyperaccumulator T. caerulescens have been isolated (Pence et al., 2000), but it is worth exploring how this knowledge may be used. From the previously described, it is clear that at least for Zn the transporters may not be very different from those in non-accumulators. What is more important is a package of genes that will include those responsible for regulation of Zn transporters and for the internal transport, detoxiﬁcation, and sequestration of Zn. Several strategies are available to increase the efﬁciency of

PHYTOREMEDIATION

45

hyperaccumulation in future, and these are not mutually exclusive: (a) agronomic means (b) conventional breeding techniques, and (c) genetic modiﬁcation. Agronomic methods include the correct use of fertilizers, crop protection chemicals, growth regulators, methods of sowing, harvesting, and all the other factors that can be manipulated in order to grow a successful crop while maximizing the metal concentration and biomass. We know more about this now than we did 10 years ago, but there is still some way to go with natural hyperaccumulators—for example— no systematic studies appear to have been made of their exact requirements for crop protection. This is not very difﬁcult to do, but it is easy to underestimate the resources needed when half a dozen species that have never been cultivated before are to be tested. Also, our knowledge of plant–soil relations for hyperaccumulators has not reached a predictive stage. We know that in different soils with different types of metal contamination the concentrations of metals in hyperaccumulators and other types of plants do not relate to the total concentrations of metals in soil. The available evidence points to little importance of chelates in the rhizosphere mobilization of metals by natural metal hyperaccumulator plants, and rhizosphere acidiﬁcation is not involved. We now know that hyperaccumulator roots “hunt” for metals in soils and that this contributes to their high uptake (Schwartz et al., 1999; Whiting et al., 2000). However, when grown on different soils, the same hyperaccumulator “sees” different supplies of metals (Knight et al., 1997). So although the roots of hyperaccumulators seem to extract metals from the same isotopically exchangeable pools as “normal” plants (Hutchinson et al., 2000), the ability of the soil to resupply the metals removed at a high rate by hyperaccumulators is also very important, and this has not been addressed. This affects our ability to predict removal in a particular soil and pollutant situation. Methods are just becoming available to help us quantify the resupply of metals from the soil matrix (Davison et al., 2000; Zhang et al., 2001). Conventional crossing to improve the efﬁciency has not been feasible between hyperaccumulators and crop plants because of incompatibility problems. However, crosses between hyperaccumulator species of Arabidopsis and closely related nonhyperaccumulators have been successfully made for genetic studies (Macnair et al., 1999). Within a species, it appears that the use of the selection of individuals with greater efﬁciency has not been achieved, except that different natural populations of T. caerulescens have been found to possess distinct differences in Cd uptake, which confers greater efﬁciency of phytoextraction (Lombi et al., 2000, 2001a). Mutagenesis is another approach which appears not to have been tested. Other non-GM methods for improving hyperaccumulation by combining these traits with those of higher biomass crop species have been attempted already, for example, by somatic hybridization (Brewer et al., 1999). However, the traits are complex and it is well known that all of the techniques for combining genomes and selection of progeny take time. One advantage the relatively conventional methods have

46

S. P. McGRATH ET AL.

over genetic manipulation of crops is that they enjoy high public acceptance, and GM technology is presently poorly accepted. Modiﬁcation of the plant genome for phytoextraction of Hg has been achieved (Meagher et al., 2000). This and additional examples of transgenic plants for phytoremediation may soon present an interesting test of public opinion because it represents a use of GM techniques for environmentally beneﬁcial cleanup. However, it is difﬁcult to predict the public acceptability of this type of application. A ﬁnal point is that the GM route requires detailed knowledge of the mechanisms of uptake, translocation, sequestration, and tolerance of metals or metalloids. This information itself is not complete for Zn and can only be described as “under way” for Ni and Cd. Returning to chemically enhanced accumulation, the environmental impacts of the added chemicals are of course of concern. It is now clear from recent research that the chelates often persist for long periods in soil, and may leach downwards towards water courses. This is not acceptable environmentally, as the metals and the chelate molecules can be toxic. Because of this, it has now become clear that chelate-enhanced phytoremediation must be carried out in closed systems. Essentially, this means using this technique with appropriate containment off site; or, if on site, the soil has to be excavated and placed on a liner which allows any drainage to be collected and recycled. This is a young group of technologies, with, for example, natural hyperaccumulation of As only recently discovered (Ma et al., 2001). It will therefore take time to gain the understanding needed to optimize the biology of the various technologies, not forgetting that there will also be a need for the development of techniques and equipment for the disposal or the recycling of the metal-rich biomass.

ACKNOWLEDGMENTS IACR-Rothamsted receives grant-aided support from the Biotechnology and Biological Sciences Research Council of the United Kingdom.

REFERENCES Anderson, C. W. N., Brooks, R. R., Chiarucci, A., LaCoste, C. J., Leblanc, M., Robinson, B. H., Simocock, R., and Stewart, R. B. (1999). Phytomining for nickel, thallium and gold. J. Geochem. Explor. 67, 407–415. Anderson, C. W. N., Brooks, R. R., Stewart, R. B., and Simcock, R. (1998). Harvesting a crop of gold in plants. Nature 395, 553–554. Aronow, L., and Kerdel-Vegas, F. (1965). Seleno-cystathionine, a pharmacologically active factor in the seeds of Lecythis ollaria. Nature 205, 1185–1186. Arvy, M. P. (1993). Selenate and selenite uptake and translocation in bean plants (Phaseolus vulgaris). J. Exper. Bot. 44, 1083–1087.

PHYTOREMEDIATION

47

Atkinson, R., Aschmann, S. M., Hasegawa, D., Thompson-Aeagle, E. T., and Frankenberger, W. T. (1990). Kinetics of the atmospherically important reactions of dimethylselenide. Environ. Sci. Technol. 24, 1326–1337. ATSDR (1988). “The nature and extent of lead poisoning in the United States.” A report to Congress, U.S. Department of Health and Human Services, Atlanta, Georgia. Azaizeh, H., Gowthaman, S., and Terry, N. (1997). Microbial selenium volatilisation in rhizosphere and bulk soils from a constructed wetland. J. Environ. Qual. 26, 666–672. Baker, A. J. M. (1981). Accumulators and excluders—Strategies in the response of plants to heavy metals. J. Plant Nutr. 3, 643–654. Baker, A. J. M., and Brooks, R. R. (1989). Terrestrial higher plants which hyperaccumulate metallic elements—A review of their distribution, ecology and phytochemistry. Biorecovery 1, 81–126. Baker, M. D., Inniss, W. E., Mayﬁeld, C. I., Wong, P. T. S., and Chau, Y. K. (1983). Effect of pH on the methylation of mercury and arsenic by sediment microorganisms. Environ. Technol. Lett. 4, 89–100. Baker, A. J. M., McGrath, S. P., Reeves, R. D., and Smith, J. A. C. (2000). Metal hyperaccumulator plants: A review of the ecology and physiology of a biological resource for phytoremediation of metal-polluted soils. In “Phytoremediation of Contaminated Soil and Water” (N. Terry, G. Banuelos, and J. Vangronsveld, Eds.), pp. 85–107. Lewis Publisher, Boca Raton, FL. Baker, A. J. M., McGrath, S. P., Sidoli, C. M. D., and Reeves, R. D. (1994a). The possibility of in situ heavy metal decontamination of polluted soils using crops of metal-accumulating plants. Resour. Conser. Recycl. 11, 41–49. Baker, A. J. M., and Proctor, J. (1990). The inﬂuence of cadmium, copper, lead, and zinc on the distribution and evolution of metallophytes in the British Isles. Plant Syst. Evol. 173, 91–108. Baker, A. J. M., Reeves, R. D., and Hajar, A. S. M. (1994b). Heavy metal accumulation and tolerance in British populations of the metallophyte Thlaspi caerulescens J. & C. Presl (Brassicaceae). New Phytol. 127, 61–68. Ballou, J. E., Price, K. R., Gies, R. A., and Doctor, P. G. (1978). Inﬂuence of DTPA on the biological availability of transuranics. Health Phys. 34, 445–455. Ba˜nuelos, G. S., Ajwa, H. A., Mackey, B., Wu, L., Cook, C., Akohoue, S., and Zambruzuski, S. (1997). Evaluation of different plant species used for phytoremediation of high soil selenium. J. Environ. Qual. 26, 639–649. Ba˜nuelos, G. S., Cardon, G., Mackey, B., Ben-Asher, J., Wu, L., Beuselinck, P., Akohoue, S., and Zambrzuski, S. (1993). Boron and selenium removal in boron laden soils by four sprinkler irrigated plant species. J. Environ. Qual. 22, 786–792. Baumann, A. (1885). Das verhalten von zinksalzen gegen pﬂanzen und im boden. Die Landwirtsch. Versuchs-Stat. 31, 1–53. Bennett, F. A., Tyler, E. K., Borrks, R. R., Gregg, P. E. H., and Stewart, R. B. (1998). Fertilisation of hyperaccumulators to enhance their potential for phytoremediation and phytomining. In “Plants that Hyperaccumulate Heavy Metals” (R. R. Brooks, Ed.), pp. 249–259. CAB International, Wallingford, UK. Bert, V., Macnair, M. R., De Laguerie, P., Saumitou-Laprade, P., and Petit, D. (2000). Zinc tolerance and accumulation in metallicolous and nonmetallicolous populations of Arabidopsis halleri (Brassicaceae). New Phytol. 146, 225–233. Bernal, M. P., and McGrath, S. P. (1994a). Effects of pH and heavy metal concentrations in solution culture on the proton release, growth and elemental composition of Alyssum murale and Raphanus sativus L. Plant Soil 166, 83–92. Bernal, M. P., McGrath, S. P., Miller, A. J., and Baker, A. J. M. (1994b). Comparison of the chemical changes in the rhizosphere of nickel hyperaccumulator Alyssum murale and non-hyperaccumulator Raphanus sativus. Plant Soil 164, 251–259.

48

S. P. McGRATH ET AL.

Bizily, S. P., Rugh, C. L., and Meagher, R. B. (2000). Phytodetoxification of hazardous organomercurials by genetically engineered plants. Nature Biotechnol. 18, 213–217. Bizily, S. P., Rugh, C. L., Summers, A. O., and Meagher, R. B. (1999). Phytoremediation of methylmercury pollution: merB expression in Arabidopsis thaliana confers resistance to organomercurials. Proc. Nat. Acad. Sci. USA. 96, 6808–6813. Blaylock, M. J. (2000). Field demonstration of phytoremediation of lead contaminated soils. In “Phytoremediation of Contaminated Soil and Water” (N. Terry and G. S. Banuelos, Eds.), pp. 1–12. Lewis Publisher, Boca Raton, FL. Blaylock, M. J., and Huang, J. W. (2000). Phytoextraction of metals. In “Phytoremediation of Toxic Metals” (I. Raskin and B. D. Ensley, Eds.), pp.71–88. Wiley, New York. Blaylock, M. J., Salt, D. E., Dushenkov, S., Zakharova, O., Gussman, C., Kapulnik, Y., Ensley, B. D., and Raskin, I. (1997). Enhanced accumulation of Pb in Indian mustard by soil-applied chelating agents. Environ. Sci. Technol. 31, 860–865. Boyd, R. S. (1998). Hyperaccumulation as a plant defensive strategy. In “Plants that Hyperaccumulate Heavy Metals” (R. R. Brooks, Ed.), pp. 181–201. CAB International, Wallingford, UK. Boyd, R. S., and Martens, S. N. (1998). Nickel hyperaccumulation by Thlaspi montanum var. montanum (Brassicaceae): A constitutive trait. Amer. J. Bot. 85, 259–265. Breton, A., and Surdin-Kerjan, Y. (1977). Sulphate uptake in Saccharomyces cerevisiae: biochemical and genetic study. J. Bacteriol. 132, 224–232. Brewer, E. P., Saunders, J. A., Angle, J. S., Chaney, R. L., and McIntosh, M. S. (1999). Somatic hybridization between the zinc accumulator Thlaspi caerulescens and Brassica napus. Theore. Appl. Gen. 9, 761–771. Brooks, R. R. (1998). “Plants That Hyperaccumulate Heavy Metals.” CAB International, Wallingford, UK. Brooks, R. R., Chambers, M. F., Nicks, L. J., and Robinson, B. H. (1998). Phytomining. Trends Plant Sci. 3, 359–362. Brooks, R. R., Lee, J., Reeves, R. D., and Jaffr´e, T. (1977). Detection of nickeliferous rocks by analysis of herbarium species of indicator plants. J. Geochem. Explor. 7, 49–57. Brooks, R. R., Shaw, S., and Marﬁl, A. A. (1981). The chemical form and physiological function of nickel in some Iberian Alyssum species. Physiol. Plant. 51, 167–170. Brown, S. L., Chaney, R. L., Angle, J. S., and Baker, A. J. M. (1994). Phytoremediation potential of Thlaspi caerulescens and bladder campion for zinc- and cadmium-contaminated soil. J. Environ. Qual. 23, 1151–1157. Brown, S. L., Chaney, R. L., Angle, J. S., and Baker, A. J. M. (1995a). Zinc and cadmium uptake by hyperaccumulator Thlaspi caerulescens grown in nutrient solution. Soil Sci. Soc. Am. J. 59, 125–133. Brown, S. L., Chaney, R. L., Angle, J. S., and Baker, A. J. M. (1995b). Zinc and cadmium uptake by hyperaccumulator Thlaspi caerulescens and metal tolerant Silene vulgaris grown on sludgeamended soils. Environ. Sci. Technol. 29, 1581–1585. Chaney, R. L. (1983). Plant uptake of inorganic waste constituents. In “Land Treatment of Hazardous Wastes” (J. F. Parr, P. B. Marsh, and J. S. Kla, Eds.), pp. 50–76. Noyes Data Corp, Park Ridge, NJ. Chaney, R. L., Li, Y.-M., Brown, S. L., Homer, F. A., Malik, M., Angle, J. S., Baker, A. J. M., Reeves, R. D., and Chin, M. (2000). Improving metal hyperaccumulator wild plants to develop commercial phytoextraction systems: Approaches and progress. In “Phytoremediation of Contaminated Soil and Water” (N. Terry, G. Ba˜nuelos, and J. Vangronsveld, Eds.), pp.129–158. Lewis Publishers, Boca Raton, FL. Chardonnens, A. N., Koevoets, P. L. M., van Zanten, A., Schat, H., and Verkleij, J. A. C. (1999). Properties of enhanced tonoplast zinc transport in naturally selected zinc-tolerant Silene vulgaris. Plant Physiol. 120, 779–785.

PHYTOREMEDIATION

49

Chiou, H. Y., Hueh, Y. M., Liaw, K. F., Horng, S. F., Chiang, M. H., Pu, Y. S., Lin, J. S. N., Huang, C. H., and Chen, C. J. (1995). Incidence of internal cancers and ingested organic arsenic: A seven years follow up study in Taiwan. Cancer Res. 55, 1296–1300. Clemens, S., Antosiewicz, D. M., Ward, J. M., Schachtman, D. P., and Schroeder, J. I. (1998). The plant cDNA LCT1 mediates the uptake of calcium and cadmium in yeast. Proc. Natl. Acad. Sci. USA 95, 12,043–12,048. Cobbett, C. S. (2000). Phytochelatins and their roles in heavy metal detoxiﬁcation. Plant Physiol. 123, 825–832. Commission of the European Communities (2001). Commission Regulation No. 466/2001 of 8 March 2001 setting maximum levels for certain contaminants in foodstuffs. Ofﬁcial Journal of the European Communities, L77/1-13, Brussels. Cooper, E. M., Sims, J. T., Cunningham, S. D., Huang, J. W., and Berti, W. R. (1999). Chelate-assisted phytoextraction of lead from contaminated soils. J. Environ. Qual. 28, 1709–1719. Cornell, R. M. (1993). Adsorption of caesium on minerals, a review. J. Radioanal. Nucl. Chem. 171, 483–500. Cornish, J. E., Goldberg, W. C., Levine, R. S., and Benemann, J. R. (1995). Phytoremediation of soils contaminated with toxic elements and radionuclides. In “Bioremediation of Inorganics” (R. E. Hinchee, J. L. Means, and D. R. Burris, Eds.), pp. 55–63. Battelle Press, Columbus-Richland. Cullen, W. R., McBride, B. C., Pickett, A. W., and Regalinski, J. (1984). The wood preservative chromated copper arsenate is a substance for trimethylarsine biosynthesis. Applied Environ. Microbiol. 47, 443–444. Cunningham, S. D., Berti, W. R., and Huang, J. W. (1995). Remediation of contaminated soils and sludges by green plants. In “Bioremediation of Inorganics” (R. E. Hinchee, J. L. Means, and D. R. Burris, Eds.), pp. 33–54. Batelle Press, Columbus-Richland. Davidian, J. C., Hatzﬁeld, Y., Cathala, N., Tagmount, A., and Vidmar, J. J. (2000). Sulphate uptake and transport in plants. In “Sulphur Nutrition and Sulfur Assimilation in Higher Plants: Molecular, Biochemical and Physiological Aspects” (C. Brunold, H. Rennenberg, L. De Kok, I. Stuhlen, and J. Davidian, Eds.), pp. 1–19. Paul Haupt, Bern. Davison, W., Hooda, P. S., Zhang, H., and Edwards, A. C. (2000). DGT measured ﬂuxes as surrogates for uptake of metals by plants. Adv. Environ. Res. 3, 550–555. Deram, A., Petit, D., Robinson, B., Brooks, R. R., Gregg, P., and Van Halluwyn, C. (2000). Natural and induced heavy-metal accumulation by Arrhenatherum elatius: Implications for phytoremediation. Comm. Soil Sci. Plant Anal. 31, 413–421. de Souza, M. P., Chu, D., Zhao, M., Zayed, A. M., Ruzin, S. E., Schichnes, D., and Terry, N. (1999a). Rhizosphere bacteria enhance selenium accumulation and volatilisation by Indian Mustard. Plant Physiol. 119, 563–573. de Souza, M. P., Huang, C. P. A., Chee, N., and Terry, N. (1999b). Rhizosphere bacteria enhance the accumulation of selenium and mercury in wetland plants. Planta 209, 259–263. Dushenkov, S., and Kapulnik, Y. (2000). Phytoﬁltration of metals. In “Phytoremediation of Toxic Metals: Using Plants To Clean Up the Environment” (R. Raskin and B. D. Ensley, Eds.), pp. 89–106. Wiley, New York. Dushenkov, V., Nanda Kumar, P. B. A., Motto, H., and Raskin, I. (1995). Rhizoﬁltration—the use of plants to remove heavy metals from aqueous streams. Environ. Sci. Technol. 29, 1239–1245. Dushenkov, S., Mikheev, A., Prokhnevsky, A., Rucho, M., and Sorochinsky, B. (1999). Phytoremediation of radiocesium-contaminated soil in the vicinity of Chernobyl, Ukraine. Environ. Sci. Technol. 33, 469–475. Ebbs, S. D., and Kochian, L. V. (1997). Toxicity of zinc and copper to Brassica species: Implications for phytoremediation. J. Environ. Qual. 26, 776–781. Ebbs, S. D., and Kochian, L. V. (1998). Phytoextraction of zinc by oat (Avena sativa), barley (Hordeum vulgare), and Indian mustard (Brassica juncea). Environ. Sci. Technol. 32, 802–806.

50

S. P. McGRATH ET AL.

Ebbs, S., Kochian, L., Lasat, L., Pence, N., and Jiang, T. (2000). An integrated investigation of the phytoremediation of heavy metal and radionuclide contaminated soils: from the laboratory to the ﬁeld. In “Bioremediation of Contaminated Soils” (D. L. Wise, D. J. Trantolo, E. J. Cichon, H. I., Inyang, and U. Stottmeister, Eds.), pp. 745–769. Marcel Dekker, New York. Eide, D., Broderius, M., Fett, J., and Guerinot, M. L. (1996). A novel iron-regulated metal transporter from plants identiﬁed by functional expression in yeast. Proc. Natl. Acad. Sci. USA 93, 5624–5628. Epstein, A. L., Gussman, C. D., Blaylock, M. J., Yermiyahu, U., Huang, J. W., Kapulnik, Y., and Orser, C. S. (1993). EDTA and Pb-EDTA accumulation in Brassica juncea grown in Pb-amended soil. Plant Soil 208, 87–94. ¨ Eriksson, E., Oborn, I., Jansson, G., and Andersson, A. (1996). Factors inﬂuencing Cd-content in crops—Results from swedish ﬁeld investigations. Swed. J. Agric. Res. 26, 125–133. ERM (1997). “Cadmium in Fertilisers,” Vols. I and II. Final report for the Commission of the European Communities, DG III. Environmental Resources Management, London. Ernst, W. H. O. (1974). “Schwermetallvegetation der Erde.” Fischer, Stuttgart. ETCS (European Topic Centre Soil) (1998). Topic report—Contaminated sites. European Environment Agency, Copenhagen. Frankenberger, W. T., and Karlson, U. (1998). Dissipation of soil selenium by microbial volatilization at Kersterson Reservoir. Final report submitted to U.S. Department of Interior. Project 7-EC-2005240, November. U.S. Department of Interior, Bureau of Reclamation, Sacramento, CA. Frey, B., Keller, C., Zierold, K., and Schulin, R. (2000). Distribution of Zn in functionally different leaf epidermal cells of the hyperaccumulator. Thlaspi caerulescens. Plant Cell Environ. 23, 675–687. Gabbrielli, R., Mattioni, C., and Vergnano, O. (1991). Accumulation mechanisms and heavy metal tolerance of a nickel hyperaccumulator. J. Plant Nutr. 14, 1067–1080. Gabbrielli, R., Pandolﬁni, T., Vergnano, O., and Palandri, M. R. (1990). Comparison of two serpentine species with different nickel tolerance strategies. Plant Soil 122, 271–277. G´erard, E., Echevarria, G., Sterckeman, T., and Morel, J. L. (2000). Cadmium availability to three plant species varying in cadmium accumulation pattern. J. Environ. Qual. 29, 1117–1123. Ghaderian, Y. S., Lyon, A. J. E., and Baker, A. J. M. (2000). Seedling mortality of metal hyperaccumulator plants resulting from damping off by Pythium spp. New Phytol. 146, 219–224. Guerinot, M. L. (2000). The ZIP family of metal transporters. Biochim. Biophys. Acta 1465, 190–198. Hamon, R. E., Wundke, J., McLaughlin, M., and Naidu, R. (1997). Availability of zinc and cadmium to different plant species. Aust. J. Soil Res. 35, 1267–1277. Hansen, D., Duda, P. J., Zayed, A. M., and Terry, N. (1998). Selenium removal by constructed wetlands: Role of biological volatilisation. Environ. Sci. Technol. 32, 591–597. Heaton, A. C. P., Rugh, C. L., Wang, N. J., and Meagher, R. B. (1998). Phytoremediation of mercuryand methylmercury-polluted soils using genetically engineered plants. J. Soil Contamin. 7, 497–509. Hindmarsh, J. T., and McCurdy, R. F. (1986). Environmental aspects of arsenic toxicity. CRC Crit. Rev. Clini. Lab. Sci. 33, 315–347. Hinsinger, P. (1998). How do plant roots acquire mineral nutrients? Chemical processes involved in the rhizosphere. Adv. Agron. 64, 225–265. Homer, F. A., Morrison, R. S., Brooks, R. R., Clement, J., and Reeves, R. D. (1991a). Comparative studies of nickel, cobalt, and copper uptake by some nickel hyperaccumulators of the genus Alyssum. Plant Soil 138, 195–205. Homer, F. A., Reeves, R. D., Brooks, R. R., and Baker, A. J. M. (1991b). Characterization of the nickel-rich extract from the nickel hyperaccumulator Dichapetalum gelonioides. Phytochem. 30, 2141–2145. Hong, P. K. A., Li, C., Banerji, S. K., and Regmi, T. (1999). Extraction, recovery, and biostability of EDTA for remediation of heavy metal-contaminated soil. J. Soil Contam. 8, 81–103.

PHYTOREMEDIATION

51

Hossner, L. R., Loeppert, R. H., Newton, R. J., Szaniszlo, P. J., and Shahandeh, H. (1999). Phytoaccumulation of chromium, uranium, and plutonium in plant systems. Amarillo National Plutonium Research Center, Amarillo, TX, Project UTA96-0043, Final Report. Huang, J. W. (2000). Phytoremediation of radionuclides. In “Proceedings of SoilRem 2000. International Conference of Soil Remediation” (Y. M. Luo, S. P. McGrath, Z. H. Cao, F. J. Zhao, Y. X. Chen, and J. M. Xu, Eds.), pp. 257–261. October 15–19, Hangzhou, China. Huang, J. W., Blaylock, M. J., Kapulnik, Y., and Ensley, B. D. (1998). Phytoremediation of uraniumcontaminated soils: role of organic acids in triggering uranium hyperaccumulation in plants. Environ. Sci. Technol. 32, 2004–2008. Huang, J. W., Chen, J., Berti, W. B., and Cunningham, S. D. (1997). Phytoremediation of leadcontaminated soils: Role of synthetic chelates in lead phytoextraction. Environ. Sci. Technol. 31, 800–805. Huang, J. W., and Cunningham, S. D. (1996). Lead phytoextraction: Species variation in lead uptake and translocation. New Phytol. 134, 75–84. Hutchinson, J. J., Young, S. D., McGrath, S. P., West, H. M., Black, C. R., and Baker, A. J. M. (2000). Determining uptake of ‘non-labile’ soil cadmium by Thlaspi caerulescens using isotopic dilution techniques. New Phytol. 146, 453–460. Jaffr´e, T., Brooks, R. R., Lee, J., and Reeves, R. D. (1976). Sebertia acuminata: A hyperaccumulator of nickel from New Caledonia. Science 193, 579–580. Kayser, A., Wenger, K., Keller, A., Attinger, W., Felix, H. R., Gupta, S. K., and Schulin, R. (2000). Enhancement of phytoextraction of Zn, Cd and Cu from calcareous soil: The use of NTA and sulfur amendments. Environ. Sci. Technol. 34, 1778–1783. Keating, M. H., Mahaffey, K. R., and Schoeny, R. (1997). “Mercury Study Report to Congress.” EPA-4521, H-97-003, U.S. Environment Protection Agency, Washington DC, 1, 2–11. Kimbrough, D. E., Cohen, Y., Winer, A. M., Creelman, L., and Mabuni, C. (1999). A critical assessment of chromium in the environment. Crit. Rev. Env. Sci. Technol. 29, 1–46. Knight, K., Zhao, F. J., McGrath, S. P., and Shen, Z. G. (1997). Zinc and cadmium uptake by the hyperaccumulator Thlaspi caerulescens in contaminated soils and its effects on the concentration and chemical speciation of metals in soil solution. Plant Soil 197, 71–78. Kofoed, A. D., and Klausen, P. S. (1983). Inﬂuence of fertiliser on the cadmium content of soils and crops. Investigations from long term ﬁeld experiments in State Research Station Askov 1894– 1981. Tidsskrift for Planteavl 87, 23–32. K¨ohl, K. I., and Smith, J. A. C. (1997). Metal-allocation patterns in cuprophytes from the Zairean copper belt contrast with those of nickel hyperaccumulation. J. Exp. Bot. (Suppl.) 48, 103. Korshunova, Y. O., Eide, D., Clark, W. G., Guerinot, M. L., and Pakrasi, H. B. (1999). The IRT1 protein from Arabidopsis thaliana is a metal transporter with a broad substrate range. Plant Mol. Biol. 40, 37–44. Kr¨amer, U., Cotter-Howells, J. D., Charnock, J. M., Baker, A. J. M., and Smith, J. A. C. (1996). Free histidine as a metal chelator in plants that accumulate nickel. Nature 379, 635–638. Kr¨amer, U., Grime, G. W., Smith, J. A. C., Hawes, C. R., and Baker, A. J. M. (1997b). Micro-PIXE as a technique for studying nickel localization in leaves of the hyperaccumulator plant Alyssum lesbiacum. Nuclear Instr. Meth. Physics Res. B 130, 346–350. Kr¨amer, U., Pickering, I. J., Prince, R. C., Raskin, I., and Salt, D. E. (2000). Subcellular localization and speciation of nickel in hyperaccumulator and non-accumulator Thlaspi species. Plant Physiol. 122, 1343–1353. Kr¨amer, U., Smith, R. D., Wenzel, W. W., Raskin, I., and Salt, D. E. (1997a). The role of metal transport and tolerance in nickel hyperaccumulation by Thlaspi goesingense Halacsy. Plant Physiol. 115, 1641–1650. Kruger, E. L., Anderson, T. A., and Coats, J. R. (1997). “Phytoremediation of Soils and Water Contaminants.” American Chemical Society Symposium Series 664, ACS, Washington DC.

52

S. P. McGRATH ET AL.

K¨upper, H., Lombi, E., Zhao, F. J., and McGrath, S. P. (2000). Cellular compartmentation of cadmium and zinc in relation to other elements in the hyperaccumulator Arabidopsis halleri. Planta 212, 75–84. K¨upper, H., Zhao, F. J., and McGrath, S. P. (1999). Cellular compartmentation of zinc in leaves of the hyperaccumulator Thlaspi caerulescens. Plant Physiol. 119, 305–311. K¨upper, H., Lombi, E., Zhao, F. J., Wieshammer, G., and McGrath, S. P. (2001). Cellular compartmentation of nickel in the hyperaccumulators Alyssum lesbiacum, Alyssum bortoilonii and Thlaspi goesingense. J. Exp. Bot. 52, (in press). Larison, J. R., Likens, G. E., Fitzpatrick, J. W., and Crock, J. G. (2000). Cadmium toxicity among wildlife in the Colorado Rocky Mountains. Nature 406, 181–183. Lasat, M. M., Baker, A. J. M., and Kochian, L. V. (1996). Physiological Characterisation of root Zn2+ absorption and translocation to shoots in Zn hyperaccumulator and nonaccumulator species of Thlaspi. Plant Physiol. 112, 1715–1722. Lasat, M. M., Baker, A. J. M., and Kochian, L. V. (1998a). Altered zinc compartmentation in the root symplasm and stimulated Zn2+ absorption into the leaf as mechanisms involved in zinc hyperaccumulation in Thlaspi caerulescens. Plant Physiol. 118, 875–883. Lasat, M. M., Fuhrmann, M., Ebbs, S. D., Cornish, J. E., and Kochian, L. V. (1998b). Phytoremediation of a radiocaesium-contaminated soil: Evaluation of caesium-137 bioaccumulation in the shoots of three plant species. J. Environ. Qual. 27, 165–169. Lasat, M. M., Norvell, W. A., and Kochian, L. V. (1997). Potential of phytoextraction of 137Cs from a contaminated soil. Plant Soil 195, 99–106. Leblanc, M., Robinson, B. H., Petit, D., Deram, A., and Brooks, R. R. (1999). The phytoming and environmental signiﬁcance of hyperaccumulation of thallium by Iberis intermedia from southern France. Econ. Geol. 94, 109–114. Lee, R. B. (1982). Selectivity and kinetics of ion uptake by barley plants following nutrient deﬁciency. Ann. Bot. 50, 429–449. Lee, J., Reeves, R. D., Brooks, R. R., and Jaffr´e, T. (1977). Isolation and identiﬁcation of a citratocomplex of nickel from nickel-accumulating plants. Phytochemistry 16, 1503–1505. Lee, J., Reeves, R. D., Brooks, R. R., and Jaffr´e, T. (1978). The relation between nickel and citric acid in some nickel-accumulating plants. Phytochemistry 17, 1033–1035. Leyval, C., and Joner, E. J. (2001). Bioavailability of heavy metals in the mycorrhizosphere. In “Trace Elements in the Rhizosphere” (G. R. Gobran, W. W. Wenzel, and E. Lombi, Eds.), pp. 165–185. CRC Press, Boca Raton, FL. Lin, Z.-Q., Schemenauer, R. S., Cervinka, V., Zayed, A., Lee, A., and Terry, N. (2000). Selenium volatilization from a soil-plant system for the remediation of contaminated water and soil in the San Joaquin Valley. J. Environ. Qual. 29, 1048–1056. Lindsay, W. L. (1979). “Chemical Equilibria in Soils,” p. 79. Wiley, New York. Lipton, W. V., and Goldin, A. S. (1976). Some factors inﬂuencing the uptake of plutonium-239 by pea plants, Health Phys. 31, 425–432. Lombi, E., Zhao, F. J., Dunham, S. J., and McGrath, S. P. (2000). Cadmium accumulation in populations of Thlaspi caerulescens and Thlaspi goesingense. New Phytologist 145, 11–20. Lombi, E., Zhao, F. J., McGrath, S. P., Young, S., and Sacchi, A. (2001a). Physiological evidence for a high afﬁnity cadmium transporter in a Thlaspi caerulescens ecotype. New Phytol. 149, 53–60. Lombi, E., Zhao, F. J., Dunham, S. J., and McGrath, S. P. (2001b). Phytoremediation of heavy metal contaminated soils: natural hyperaccumulation versus chemically-enhanced phytoextraction. J. Environ. Qual. (in press). Losi, M. E., and Frankenberger, W. T., Jr. (1997). Reduction of selenium oxyanions by Enterobacter cloacae SLD1a-1:isolation and growth of the bacterium and its expulsion of selenium particles. Appl. Environ. Microbiol. 63, 3079–3084. Ma, L. Q., Komar, K. M., Tu, C., Zhang, W., Cai, Y., and Kennelley, E. D. (2001). A fern that hyperaccumulates arsenic. Nature 409, 579–79.

PHYTOREMEDIATION

53

Macnair, M. R., Bert, V., Huitson, S. B., Saumitou-Laprade, P., and Petit, D. (1999). Zinc tolerance and hyperaccumulation are genetically independent characters. Proc. R. Soc. London 226, 2175– 2179. Marschner, H. (1995). “Mineral Nutrition of Higher Plants,” 2nd ed., Academic Press, London. Mathys, W. (1977). The role of malate, oxalate, and mustard oil glucosides in the evolution of zincresistance in herbage plants. Physiol. Plant. 40, 130–136. McBride, B. C., and Wolfe, R. S. (1971). Biosynthesis of dimethylarsine by Methanobacterium. Biochemistry 10, 4312–4317. McGrath, S. P. (1998). Phytoextraction for soil remediation. In “Plants that Hyperaccumulate Heavy Metals” (R. R. Brooks, Ed.), pp. 261–287. CAB International, Wallingford, UK. McGrath, S. P., Dunham, S. J., and Correll, R. L. (2000). Potential for phytoextraction of zinc and cadmium from soils using hyperaccumulator plants. In “Phytoremediation of Contaminated Soil and Water” (N. Terry and G. Ba˜nuelos, Eds.), pp. 109–128. Lewis Publishers, Boca Raton, FL. McGrath, S. P., Shen, Z. G., and Zhao, F. J. (1997). Heavy metal uptake and chemical changes in the rhizosphere of Thlaspi caerulescens and Thlaspi ochroleucum grown in contaminated soils. Plant Soil 180, 153–159. McGrath, S. P., Sidoli, C. M. D., Baker, A. J. M., and Reeves, R. D. (1993). The potential for the use of metal-accumulating plants for the in situ decontamination of metal-polluted soils. In “Integrated Soil and Sediment Research: A Basis for Proper Protection” (H. J. P. Eijsackers and T. Hamers, Eds.), pp. 673–676. Kluwer Academic, Dordrecht. McLaughlin, M. J., and Henderson, R. (1999). Effect of zinc and copper on cadmium uptake by Thlaspi caerulescens and Cardaminopsis halleri. In “Proceedings of the 5th International Conference on the Biogeochemistry of Trace Elements,” pp. 886–887. Vienna, Austria, 11–15 July 1999. Meagher, R. B., and Rugh, C. L. (1996). Phytoremediation of heavy metal pollution: Ionic and methyl mercury. Biotechnology for water use and conservation: the Mexico ’96 workshop, Morelos, Mexico, 20–23 October 1996. Organisation for Economic Cooperation and Development (OECD), Paris, France, 1997. pp. 305–321. Meagher, R. B., Rugh, C. L., Kandasamy, M. K., Gragson, G., and Wang, N. J. (2000). Engineered Phytoremediation of Mercury Pollution in Soil and Water Using Bacterial Genes. In “Phytoremediation of Contaminated Soil and Water” (N. Terry and G. Ba˜nuelos, Eds.), pp. 201–220. Lewis Publishers, Boca Raton, FL. Mesjasz-Przybylowicz, J., Balkwill, K., Przybylowicz, W. J., and Annegarn, H. J. (1994). Proton microprobe and X-ray ﬂuorescence investigations of nickel distribution in serpentine ﬂora from South Africa. Nucl. Instr. Meth. Phys. Res. B 89, 208–212. Morrey, D. R., Balkwill, K., and Balkwill, M. J. (1989). Studies on serpentine ﬂora, preliminary analyses of soils and vegetation associated with serpentine rock formations in the southeast Transvaal. S. Afri. J. Bot. 55, 171–177. Needleman, H. L., Schell, A., Bellinger, D., Leviton, A., and Allred, E. N. (1990). The long-term effects of exposure to low doses of lead in childhood. An 11-year follow-up report. New Engl. J. Med. 322, 83–88. Negri, M. C., and Hinchman, R. R. (2000). The use of plant for the treatment of radionuclides. In “Phytoremediation of Toxic Metals” (I. Raskin and B. D. Ensley, Eds.), pp. 107–132. Wiley, New York. Nicks, L., and Chambers, M. F. (1995). Farming for metals. Mining Eng. Manag. September 15–18. Nriagu, J. O. (1979). Production and uses of mercury. In “The Biogeochemistry of Mercury in the Environment,” pp. 23–41. Elsevier/North Holland Biomedical Press, Amsterdam. Ohlendorf, H. M., Hoffman, D. J., Saiki, M. K., and Aldrich, T. W. (1986). Embryonic mortality and abnormalities of birds: apparent impacts of selenium from irrigation drain water. Sci. Total Environ. 52, 49–63.

54

S. P. McGRATH ET AL.

Oosterhuis, F. H., Brouwer, F. M., and Wijnants, H. J. (2000). A possible EU wide charge on cadmium in phosphate fertilisers: Economic and environmental implications. Report E-00/02, European Commission, Brussels. Pence, N. S., Larsen, P. B., Ebbs, S. D., Letham, D. L. D., Lasat, M. M., Garvin, D. F., Eide, D., and Kochian, L. V. (2000). The molecular physiology of heavy metal transporter in the Zn/Cd hyperaccumulator Thlaspi caerulescens. Proc. Nat. Acad. Sci. USA 97, 4956– 4960. Persans, M. W., Yan, X. G., Patnoe, J., Kr¨amer, U., and Salt, D. E. (1999). Molecular dissection of the role of histidine in nickel hyperaccumulation in Thlaspi goesingense (Halacsy). Plant Physiol. 121, 1117–1126. Pilon-Smits, E. A. H., Hwang, S.-B., Lytle, C. M., Zhu, Y.-L., Tai, J. C., Bravo, R. C., Chen, Y.-C., Leustek, T., and Terry, N. (1999). Overexpression of ATP sulfurylase in Indian Mustard leads to increased selenate uptake, reduction and tolerance. Plant Physiol. 199, 123–132. Pollard, A. J., and Baker, A. J. M. (1996). Quantitative genetics of zinc hyperaccumulation in Thlaspi caerulescens. New Phytol. 132, 113–118. Pollard, A. J., Dandridge, K. L., and Jhee, E. M. (2000). Ecological genetics and the evolution of trace element hyperaccumulation in plants. In “Phytoremediation of Contaminated Soil and Water” (N. Terry and G. Ba˜nuelos, Eds.), pp. 251–264. Lewis Publishers, Boca Raton, FL. Psaras, G. K., Constantinidis, T., Cotsopoulos, B., and Manetas, Y. (2000). Relative abundance in the leaf epidermis of eight hyperaccumulators: evidence that the metal is excluded from both guard cells and trichomes. Ann. Bot. 86, 73–78. Reeves, R. D., and Baker, A. J. M. (1984). Studies on metal uptake by plants from serpentine and nonserpentine populations of Thlaspi goesingense H´alacsy (Cruciferae). New Phytol. 98, 191–204. Reeves, R. D., and Baker, A. J. M. (2000). Metal-accumulating plants. In “Phytoremediation of Toxic Metals: Using Plants To Clean Up the Environment” (I. Raskin and B. D. Ensley, Eds.), pp. 193–229. Wiley, New York. Reeves, R. D., and Brooks, R. R. (1983). Hyperaccumulation of lead and zinc by two metaophytes from a mining area in Central Europe. Environ. Pollut. 31, 277–287. Resnik, M. C., Lunt, O. R., and Wallace, A. (1969). Caesium, potassium, strontium and calcium transport in two different plant species. Soil Sci. 108, 64–73. Robinson, B. H., Brooks, R. R., Howes, A. W., Kirkman, J. H., and Gregg, P. E. H. (1997b). The potential of the high-biomass nickel hyperaccumulator Berkheya coddii for phytoremediation and phytomining. J. Geochem. Explor. 60, 115–126. Robinson, B. H., Chiarucci, A., Brooks, R. R., Petit, D., Kirkman, J. H., Gregg, P. E. H., and DeDominicis, V. (1997a). The nickel hyperaccumulator plant Alyssum bertolonii as a potential agent for phytoremediation and phytomining of nickel. J. Geochem. Explor. 59, 75–86. Robinson, B. H., Leblanc, A., Petit, D., Brooks, R. R., Kirkman, J. H., and Gregg, P. E. H. (1998) The potential of Thlaspi caerulescens for phytoremediation of contaminated soils. Plant Soil 203, 47–56. Rogers, E. E., Eide, D. J., and Guerinot, M. L. (2000). Altered selectivity in an Arabidopsis metal transporter. Proc. Nat. Acad. Sci. USA 97, 12,356–12,360. Rugh, C. L., Senecoff, J. F., Meagher, R. B., and Merkle, S. A. (1998). Development of transgenic yellow-poplar for mercury phytoremediation. Nature Biotechnol. 33, 616–621. Rugh, C. L., Wilde, D., Stack, N. M., Thompson, D. M., Summers, A. O., and Meagher, R. B. (1996). Mercuric ion reduction and resistance in transgenic Arabidopsis thaliana plants expressing a modiﬁed bacterial merA gene. Proc. Nat. Acad. Sci USA 93, 3182–3187. Sagner, S., Kneer, R., Wanner, G., Cosson, J. P., Deus-Neumann, B., and Zenk, M. H. (1998). Hyperaccumulation, complexation and distribution of nickel in Sebertia acuminata. Phytochemistry 47, 339–347.

PHYTOREMEDIATION

55

Salt, D. E., Kato, N., K¨amer, U., Smith, R. D., and Raskin, I. (2000). The role of root exudates in nickel hyperaccumulation and tolerance in accumulator and nonaccumulator species of Thlaspi. In “Phytoremediation of Contaminated Soil and Water” (N. Terry and G. Ba˜nuelos, Eds.), pp. 189–200. Lewis, Boca Raton, FL. Salt, D. E., and Kr¨amer, U. (2000). Mechanisms of metal hyperaccumulation in plants. In “Phytoremediation of Toxic Metals: Using Plants To Clean Up the Environment” (I. Raskin and B. D. Ensley, Eds.), pp. 231–246. Wiley, New York. Salt, D. E., Prince, R. C., Baker, A. J. M., Raskin, I., and Pickering, I. J. (1999). Zinc ligands in the metal hyperaccumulator Thlaspi caerulescens as determined using X-ray absorption spectroscopy. Environ. Sci. Technol. 33, 713–717. Salt, D. E., Smith, R. D., and Raskin, I. (1998). Phytoremediation. Annu. Rev. Plant Physiol. Plant Mol. Biol. 49, 643–648. Saric, M. R., Stojanovic, M., and Babic, M. (1995). Uranium in plant species grown on natural barren soil, J. Plant Nutr. 18, 1509–1518. Satroutdinov, A. D., Dedyukhina, E. G., Chistyakova, T. I., Witschel, M., Minkevich, I. G., Eroshin, V. K., and Egli, T. (2000). Degradation of metal-EDTA complexes by resting cells of the bacterial strain DSM 9103. Environ. Sci. Technol. 34 1715–1720. Schulman, R. N., Salt, D. E., and Raskin, I. (1999). Isolation and partial characterisation of a leadaccumulating Brassica juncea. Theor. Appl. Genet. 99, 398–404. Schwartz, C., Morel, J. L., Saumier, S., Whiting, S. N., and Baker, A. J. M. (1999). Root development of the zinc-hyperaccumulator plant Thlaspi caerulescens as affected by metal origin, content and localization in soil. Plant Soil 208, 103–115. Shahandeh, H., Lee, J.-H., Hossner, L. R., and Loeppert, R. H. (2000). Bioavailability of uranium and plutonium to plants in soil-water systems and the potential of phytoremediation. In “Trace Elements in the Rhizosphere” (G. R. Gobran, W. W. Wenzel, and E. Lombi, Eds.), pp. 93–126. CRC Press, Boca Raton. FL. Shen, Z. G., Zhao, F. J., and McGrath, S. P. (1997). Uptake and transport of zinc in the hyperaccumulator Thlaspi caerulescens and the non-hyperaccumulator Thlaspi ochroleucum. Plant Cell Environ. 20, 898–906. Smith, J. A. C., Harper, F. A., Leighton, R. S., Thompson, I. P., Vaughan, D. J., and Baker, A. J. M. (1999). Comparative analysis of metal uptake, transport and sequestration in hyperaccumulator plants. In “Proceedings of the 5th International Conference on the Biogeochemistry of Trace Elements” (W. W. Wenzel, D. C. Adriano, B. Alloway, H. E. Doner, C. Keller, N. W. Lepp, M. Mench, R. Naidu, and G. M. Pierzynski, Eds.), pp. 22–23. Vienna. Terry, N., and Ba˜nuelos, G. (2000). “Phytoremediation of Contaminated Soil and Water.” Lewis Publishers, Boca Raton, FL. Terry, N., Zayed, A. M., de Souza, M. P., and Tarun, A. S. (2000). Selenium in higher plants. Annu. Rev. Plant Physiol. Plant Mol. Biol. 51, 401–432. Thomine, S., Wang, R., Ward, J. M., Crawford, N. M., and Schroeder, J. I. (2000). Cadmium and iron transport by members of a plant metal transporter family in Arabidopsis with homology to Nramp genes. Proc. Nat. Acad. Sci. USA 97, 4991–4996. Tolr`a, R. P., Poschenrieder, C., and Barcel´o, J. (1996). Zinc hyperaccumulation in Thlaspi caerulescens. II. Inﬂuence on organic acids. J. Plant Nutr. 19, 1541–1550. US EPA (1997). Recent development for in-situ treatment of metal contaminated soils. Ofﬁce of Solid Waste and Emergency Response. EPA-542-R-97-004. p. 7. Vassil, A. D., Kapulnik, Y., Raskin, I., and Salt, D. E. (1998). The role of EDTA in lead transport and accumulation by Indian mustard. Plant Physiol. 117, 447–453. V´azquez, M. D., Barcel´o, J., Poschenrieder, C., M´adico, J., Hatton, P., Baker, A. J. M., and Cope, G. H. (1992). Localization of zinc and cadmium in Thlaspi caerulescens (Brassicaceae), a metalophyte that can hyperaccumulate both metals. J. Plant Physiol. 140, 350–355.

56

S. P. McGRATH ET AL.

V´azquez, M. D., Poschenrieder, C., Barcel´o, J., Baker, A. J. M., Hatton, P., and Cope, G. H. (1994). Compartmentation of zinc in roots and leaves of the zinc hyperaccumulator Thlaspi caerulescens J & C Presl. Bot. Acta 107, 243–250. Vyas, B. N., and Mistry, K. B. (1983). Inﬂuence of chelating agents on the uptake of 238 Pu and 241 Am by plants. Plant Soil 73, 345–353. Wenzel, W. W., Adriano, D. C., Salt, D., and Smith, R. (1999). Phytoremediation: A plant-microbebased remediation system. In “Bioremediation of Contaminated Soils” (D. C. Adriano, J. M. Bollag, W. T. Frankenberger, and R. C. Sims, Eds.), pp. 457–508. SSSA, Madison, WI, Monograph No. 37. Whiting, S. N., Leake, J. R., McGrath, S. P., and Baker, A. J. M. (2000). Positive responses to Zn and Cd by roots of the Zn and Cd hyperaccumulator. Thlaspi caerulescens. New Phytol. 145, 199–210. Wu, L., Chen, J., Tanji, K. K., and Ba˜nuelos, G. S. (1995). Distribution and biomagniﬁcation of selenium in a restored upland grassland contaminated by selenium from agricultural drain water. Environ. Tox. Chem. 14, 733–742. Wu, J., Hsu, F. C., and Cunningham, S. D. (1999). Chelate assisted Pb phytoextraction: Pb availability, uptake, and translocation constraints. Environ. Sci. Technol. 33, 1898–1904. Yang, G., Wang, S., Zhou, R., and Sun, S. (1983). Endemic selenium intoxication of humans in China. Am. J. Clin. Nutr. 37, 872–881. Zayed, A. M., Lytle, M., and Terry, N. (1998). Accumulation and volatilisation of different chemical species of selenium by plants. Planta 206, 284–292. Zayed, A. M., Pilon-Smits, E. A. H., de Souza, M. P., Lin, Z. Q., and Terry, N. (1999). Remediation of selenium-polluted soils and waters by phytovolatilisation. In “Phytoremediation of MetalContaminated Water and Soils” (N. Terry and G. Banuelos, Eds.), pp. 61–83. CRC Press, Boca Raton, FL. Zayed, A. M., and Terry, N. (1994). Selenium volatilisation in roots and shoots: Effects of shoot removal and sulfate level. J. Plant Physiol. 143, 8–14. Zhao, F. J., Hamon, R. E., and McLaughlin, M. J. (2001). Root exudates of the hyperaccumulator Thlaspi caerulescens do not enhance metal mobilization. New Phytol. 151, 613–620. Zhao, F. J., Lombi, E., Breedon, T., and McGrath, S. P. (2000). Zinc hyperaccumulation and cellular distribution in Arabidopsis halleri. Plant Cell Environ. 23, 507–514. Zhao, F. J., Shen, Z. G., and McGrath, S. P. (1998). Solubility of zinc and interactions between zinc and phosphorus in the hyperaccumulator Thlaspi caerulescens. Plant Cell Environ. 21, 108–114. Zhang, H., Zhao, F. J., Sun, B., Davison, W., and McGrath, S. P. (2001). The new concept of effective soil solution concentration predicts metal availability to plants. Environ. Sci. Technol. 35, 2602–2607.

THE SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES: A NOVEL UNDERSTANDING OF HUMUS CHEMISTRY AND IMPLICATIONS IN SOIL SCIENCE Alessandro Piccolo Dipartimento di Scienze Chimico-Agrarie Universit`a Degli Studi di Napoli Federico II Via Universit´a 100 80055 Portici, Italy

I. II. III. IV.

V.

VI. VII. VIII.

IX.

Introduction Paradigmatic View of Humus Chemistry The Dilemma of the Conformational Structure of Humic Substances Size-Exclusion Chromatography of Humic Substances A. Low-Pressure Chromatography B. High-Pressure Chromatography Supramolecular Associations of Self-Assembling Humic Molecules A. Analytical Evidence: Low-Pressure Size-Exclusion Chromatography B. Analytical Evidence: High-Pressure Size-Exclusion Chromatography C. Preparative HPSEC and Characterization of Fractions D. Solute–Gel–Eluent as an Interactive System in Size-Exclusion Chromatography E. Concepts of Supramolecular Association of Humic Substances Chemical and Spectroscopic Evidence of Supramolecular Associations Turning Loose Humic Superstructures into Stable Polymers Role of Hydrophobic Humic Superstructures in Soil A. Sequestration of Organic Carbon in Soil by Hydrophobic Protection B. Hydrophobic Humic Associations in the Stabilization of Soil Structure Future Perspectives in Research and Technology References

The scientific understanding of the molecular structure of humic substances is critically reviewed here. The traditional view that humic substances are polymers in soil is not substantiated by any direct evidence but only assumed on the basis of laboratory experiments with model molecules and unwarranted results by incorrectly applying either analytical procedures or mathematical treatments developed for purified and undisputed biopolymers. A large body of evidence instead shows 57 Advances in Agronomy, Volume 75 C 2002 by Academic Press. All rights of reproduction in any form reserved. Copyright 0065-2113/02 $35.00

58

ALESSANDRO PICCOLO an alternative understanding of the conformational nature of humic substances which should be regarded as supramolecular associations of self-assembling heterogeneous and relatively small molecules deriving from the degradation and decomposition of dead biological material. A major aspect of the humic supramolecular structure is that it is predominantly stabilized by weak dispersive forces instead of covalent linkages. Hydrophobic (van der Waals, π–π, CH–π ) and hydrogen bonds are responsible for the apparent large molecular size of humic substances, the former becoming more important with the increase of pH. Such novel description of humic substances structure better accounts for their essential role in providing and maintaining soil physical and chemical quality and their reactivity towards pesticides and other environmental soil contaminants. This innovative understanding of the nature of humic substances implies a further development of the science and technology for the control of the chemistry and dynamics of natural organic matter C 2002 Academic Press. in the soil and the environment.

I. INTRODUCTION Humic substances (HS) are natural organic substances that are ubiquitous in water, soil, and sediments. Because of the beneﬁcial effects which HS have on the physical properties of soil, their role in the soil environment is signiﬁcantly greater than that attributed to their contributions to sustaining plant growth. HS are recognized for their role in controlling both the fate of environmental pollutants and the biogeochemistry of organic carbon in the global ecosystem (Piccolo, 1996). Despite the prominent importance which the substances have in sustaining life, their basic chemical nature and reactivities are still poorly understood. It is hard to understand why detailed studies on the structure and reactivity of HS (other than percentage of organic carbon and cation exchange capacity) have not been of higher interest to soil and agronomic sciences. Possibly because in the realm of naturalistic and taxonomic descriptions of soils, statistically based agronomy, and traditional inorganic fertility, an in-depth chemical description of humic organic molecules would have multiplied rather than simpliﬁed the different practical issues at stake. On the other hand, the scientiﬁc community in charge of the chemical research on humic substances has failed up to date to reach a uniﬁed understanding of this ﬁeld of science that hence remained obscure for the enlarged operators. Nevertheless, the implications of humic acid structure should extend far beyond the interests of a few chemists; humic acid structure affects functionality of the soil ecosystem strongly as well as the bioavailabilty of organic substances (including potential pollutants) therein (Tate, 1999). Soil structure, biological activity, and chemical bioavailability (or sequestration) are only three of the major soil properties and processes affected directly by humic matter.

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

59

Most of the difﬁculties encountered in chemically deﬁning the structure and reactivity of humus derives from its extremely large chemical heterogeneity and geographical variability. Being undoubtedly a mixture that originates randomly from the decay of plant tissues or microbial metabolism–catabolism or both, the Chemistry of humus is not only of utmost complexity but also a function of the different general properties of the ecosystem in which it is formed: vegetation, climate, topography, etc. It is not surprising that, despite the efforts of many excellent scientists in the distant and recent past (see Kononova (1961) and Stevenson (1994) for an outline of the history of humus chemistry research), modern pure chemists have generally avoided the study of HS, preferring either chemical or biochemical issues of more recognized molecular regularity. Besides the problem of identifying causes and attributing responsibilities, the tremendous task of advancing the knowledge of humic chemistry still lies ahead of us. It should be obvious, to a world that appreciates the potentials of genetic engineering based on understanding of DNA structure, that accurate predictions of reactivities and development of related technologies can only be made when there is a basic knowledge of the chemical structure of the reacting molecules. It is the aim of this review to report on recent innovative ﬁndings related to the conformational structure of humic material and the profound implications that these may have on our understanding of soil organic matter functions and reactivity.

II. PARADIGMATIC VIEW OF HUMUS CHEMISTRY The amounts of humic substances in soils are several times greater than those in waters. Estimates of the global abundance of organic carbon (OC) in soil organic matter (SOM) vary greatly, although the most widely accepted values for this OC are in the range of 14 to 15 × 1017 g (Eswaran et al., 1993; Schlesinger, 1997). Up to 70–80% of the OC in mineral soils can be composed of humic material. Recognizable plant remains constitute a small percentage of the SOM of mineral soils. Thus the abundance of OC in HS is on the order of two to three times greater than the terrestrial biomass estimated to be on the order of 5.6 × 1017 g (Schlesinger, 1997). However, microbial biomass is an important constituent of the organic matter pool and greatly inﬂuences its dynamics in soils (Insam, 1996). Biological activity decomposes labile plant materials rapidly on entering aerobic soil environments with adequate water supplies, but more resistant components transform slowly in the same environment. Because of the compositional diversities and the differences in the transformation modes of the components, it is impossible to accurately deﬁne the gross mixtures that compose SOM, or the dissolved organic matter (DOM), or the particulate organic matter (POM) of waters.

60

ALESSANDRO PICCOLO

Simpliﬁcation and reductionism had to be adopted. Stevenson (1994), summarizing previous reports and deﬁnitions, stated that humus includes a broad spectrum of organic constituents, many of which have their counterparts in biological tissues, and distinguished between nonhumic substances and humic substances. The former consists of compounds belonging to the well-known classes of organic chemistry such as amino acids, carbohydrates, lipids, lignin, nucleic acids, etc., while the latter is an unspeciﬁed, transformed, dark-colored, heterogeneous, amorphous and “high-molecular-weight” material. It must be recalled that the classical deﬁnitions of HS are only operational and based on solubility properties in the aqueous solutions used as soil extractants. A number of textbooks and speciﬁc reviews give extended accounts of the different extracting solutions which have been applied to isolate HS from soils (Hayes, 1985; Schnitzer and Khan, 1972; Stevenson, 1994). The generalized terms humic acids (HAs), fulvic acids (FAs), and humins cover the major fractions still used to describe HS components, but the boundaries between these fractions have not been yet clariﬁed in chemical terms. HAs are the fraction of HS solubilized under neutral or more often alkaline conditions and precipitates when solution pH is reduced to 1 by acid addition. Fulvic acids (FAs) are the fraction of humic substances that remain soluble under all pH conditions or the fraction that stays in solution when alkaline soil extracts are adjusted to pH 1. Humin is the fraction of humic substances that is not soluble in water at any pH value and because of the difﬁculty in extraction has not been, untill recently, studied as extensively as HAs and FAs. The chemical properties of HAs and FAs have been generally investigated without attempting any further fractionation. Several studies showed that chemical features such as elemental analysis (C, H, O) and acidic functional groups appeared relatively constant for HAs and FAs from many different soils (Kononova, 1961). Though not based on any molecular understanding, such simple correlations encouraged scientists to consider humic fractions, rather than complex mixtures of nonspeciﬁc compounds, as chemical entities having speciﬁc properties which can be recognized in different environments. Much of the modern understanding of HS derives from the synthesis of existing theories illustrated in the book of Kononova (1961). Upon a review of hypotheses advanced by scientists from Russia and other countries, Kononova introduced the concept that HS are comprised of system of polymers based on the observation that elemental composition, optical properties, exchange acidities, electrophoretic properties, and molecular weight characteristics varied consistently with soil classes. Using this concept, the various fractions of HS obtained on the basis of solubility characteristics are imagined to be part of a heterogeneous mixture of molecules, which, in any given soil, range in molecular weight from as low as several hundreds to perhaps over 300,000 daltons (Da) and exhibit a continuum of any given chemical property (Stevenson, 1994).

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

61

It must be noted that despite the existence of data showing that molecular dimensions (as measured by osmometry, viscometry, and diffusion) of some HS were hardly beyond 1000 or 2000 Da (Scheffer and Ulrich, 1960; Schnitzer and Khan, 1972), more conﬁdence was placed on the early work of Flaig (1958), who showed, using the ultracentrifuge, that molecular weights were in the range of 30,000–50,000 Da for HAs and about 10,000 Da for FAs. A reason for such bias towards high-molecular-weight structures may be explained by the predominant role of historical hypotheses viewing HS as a product of a biologically assisted syntheses from compounds derived from degradation of lignin, polyphenols, cellulose, and amino acids. Though no direct evidence had ever been produced for the occurrence of such polymer build-up in a natural soil system, many classical laboratory experiments had shown the possibility for either abiotic or biotic condensation of simple molecules into humic-like material (Kononova, 1961). Some of these early laboratory studies have been proposed again under more or less conﬁned conditions in recent times (Flaig, 1988; Flaig et al., 1975; Haider and Martin, 1967; Hedges, 1988; Martin and Haider, 1969). The polymeric view of humic substances included the general idea of polydispersity (Dubach and Metha, 1963) by which HS are made of polymers having different molecular weights similar to other natural biological macromolecules such as proteins, polysaccharides, and lignin. The experimental difﬁculty in isolating chemically homogeneous fractions of HS was approximated to that observed for those biopolymers which are synthesized in the living cell with varying molecular dimensions. This unwarranted similarity supported the polydisperse polymeric understanding of HS and justiﬁed the undisputed observation that they are a mixture of compounds. The concept of high-molecular-weight and polydisperse polymers became a paradigmatic part of any descriptive deﬁnitions of humic substances proposed thereafter (Aiken et al., 1985; Malcolm, 1990; Schnitzer and Khan, 1972; Stevenson, 1994). The assumption that HS were polymers, without a sound molecular basis, has promulgated the use of simple physical–chemical measurements to characterize HS. An example is the E4/E6 index, the ratio of the absorbance of HS at 465 nm to that at 665 nm, introduced by Welte (1955) and reproposed by Kononova (1961) as supposedly indicating a reverse relationship with progressive humiﬁcation and increased condensation (large content of polycondensated aromatic-ring structures). Despite its popularity and continued application, the E4/E6 ratio has repeatedly been shown not to hold the predicted relationship with molecular weight. At variance with the hypotheses related to this ratio, Campbell et al. (1967) already indicated that humic material with the lowest mean residence time had the highest E4/E6 ratio. Handerson and Hepburn (1977), using gel permeation chromatography, showed that humic fractions with large molecular size and low E4/E6 ratio had mainly aliphatic content, whereas those with low molecular size had the highest aromatic content. Piccolo (1988) compared E4/E6 ratios of various HS with gel

62

ALESSANDRO PICCOLO

permeation chromatograms and showed that the two methods produced comparable results only after extreme puriﬁcation of HS. Summers et al. (1987) reported on the limitations of E4/E6 ratio and found that the ratio values varied considerably with the concentration of ultraﬁltered fractions of HS. By comparing HS extracted from soils and soil particles, before and after long-term amendments with organic wastes, Piccolo and Mbagwu (1990) found that the E4/E6 ratio did not account for the increase in molecular size observed in amended samples by gel ﬁltration. There may have been various reasons for which the scientiﬁc community thought that HS had a polymeric structure despite the lack of any sound evidence of their macromolecularity. One reason resides in the acceptance of Staudinger’s (1935) description of macromolecular polymers occurring in living cells. According to Staudinger’s view, it seemed convenient to assume that HS were polymers, notwithstanding the fact that HS were not a product of cellular synthesis as other biomolecules but rather that of cell death. The rapid degradation and decomposition in soil of biopolymers liberated from cell lysis are now well-accepted processes from both a biological and a thermodinamic perspective (Clapp and Hayes, 1999; Haider, 1987; Jenkinson, 1981; Spaccini, Piccolo, Haberhauer, Stemmer et al., 2001). Surprisingly, Staudinger’s view is currently still acritically advocated by defenders of the polymeric nature of HS (Swift, 1999). A second reason is that a stable polymeric structure accounted for the refractory characteristics of HS in soil, other than the physical protection conferred by inorganic soil particles. The organic carbon in stable humic materials is known (Stevenson, 1994) to possess long residence times (from 250 up to 3000 years) in soil. Moreover, the classical hypothesis of HS formation through a condensation between amino acids and components of degraded lignin spread the assumption that HS had a polymeric structure similar to that of lignin (Waksman, 1936). Lignin is known to be polydisperse in molecular weight with values ranging from <1000 to several million Da (Goring, 1971), and its resistance to microbial degradation in soil has been repeatedly attributed to the macromolecular structure (Amalﬁtano et al., 1992; Waksman, 1936). Similarly, other classical hypotheses of HS formation, such as the polyphenol theory (Flaig et al., 1975) or the melanoidin theory based on the Maillard reaction (Maillard, 1912), resided on the same logic that justiﬁes the polymeric paradigm of HS with polymerization processes occurring under laboratory or conﬁned conditions. Another aspect that contributed to the polymeric view of HS is their colloidal properties which may be assimilated to those of polyelectrolytes in aqueous media (Flaig et al., 1975; van Dijk, 1972). Most of the properties of polyelectrolytes such as processes of ﬂocculation and dispersion, responses to electrolytes, doublelayer behavior were apparently observed also for HS, thereby easily transferring to humic extracts the concept of high molecular weight. Despite the widely accepted view of the macromolecularity of HS, this has never been unambiguously demonstrated in chemical and physical–chemical terms

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

63

in soil-extracted humic fractions. If the instrumentation for organic and physical chemistry analysis had not been generally available in the past nor completely reliable, this situation has changed dramatically in the present world of sophisticated biophysical technology. The polymeric paradigm of HS should therefore be fully proved beyond any doubt or abandoned for other descriptions.

III. THE DILEMMA OF THE CONFORMATIONAL STRUCTURE OF HUMIC SUBSTANCES A determination of the true molecular weight of a humic fraction is the central problem of HS research. The molecular weight determines the conformational structure (i.e., size and shape) of HS and ultimately their reactivity in the soil and the environment. A number of reviews have repeatedly pointed out that there is no accordance among the values obtained by the different methods applied to measure molecular weight of HS (Clapp et al., 1989; De Nobili et al., 1989; Stevenson, 1994; Swift, 1989; Wershaw and Aiken, 1985). The differences are not slight but of several orders of magnitude. This undeniable fact has had (with some exception) little impact on those who view HS as polymers. Without questioning the polymeric paradigm, the great difference in values has been attributed to either the variability of HS or the intrinsic limitation of methods when applied to polydisperse systems. However, very few attempts have been made to seriously reduce such polydispersity by classical chemical methods. Much of the confusion on the molecular weight of HS had arisen by using sedimentation velocity and diffusion methods, performed with an ultracentrifuge, which became fashionable in the years between the 1950s and the 1970s with the upsurge of biochemistry research. Despite the clear indication that such semiempirical methods are not suitable for polydisperse systems because of the multiple diffusion coefﬁcients and sedimentation constants of different size particles (Wershaw and Aiken, 1985), studies on whole extracts of humic acids were conducted and molecular weight values ranging from 25,000 to more than 200,000 Da were reported (Flaig, 1958; Piret et al., 1960; Stevenson et al., 1953). Interestingly, Flaig and Beutelspacher (1968), also working with polydisperse humic solutions, found that molecular weight varied from 77,000 Da, when 0.2 M NaCl was added to depress repulsive negative charges, to only 2050 Da in the absence of the background electrolyte. Though the latter is commonly added to polyelectrolytes to reduce interferences from charge repulsion in monodisperse systems, a jump of about 2 orders of magnitude in molecular weight (MW) for the same humic polydisperse system could rather suggest a molecular association rather than an extreme interference in sedimentation-velocity of polyelectrolytes. In an attempt of reduce the polydispersity of HS, Cameron, Thornton et al. (1972) extensively fractionated soil-extracted humic acids by ultraﬁltration and gel

64

ALESSANDRO PICCOLO

permeation and subjected the fractions to sedimentation-velocity ultracentrifugation. The resulting fractions were more homogeneous than the unfractionated solution, but still were not monodisperse. In fact, they reported molecular weights ranging from 2600 to a surprising 1,360,000 Da. Instead of true polymers, they measured associations of smaller molecules randomly aggregated by different forces during the fractionations procedures. A more successful fractionation into homogeneous fractions may have been achieved if classical solid–liquid or liquid– liquid separations of mixtures of organic compounds had been employed, instead of gel permeation that notoriously separates HS by size exclusion only in speciﬁc circumstances (De Nobili et al., 1989; Lindqvist, 1967). Moreover, Cameron, Thornton et al. (1972) overlooked the difﬁculties in assessing sedimentation coefﬁcients for associations of molecules such as HS. The measurement of sedimentation coefﬁcients of polydisperse materials that include subunits invariably leads to erroneous values of molecular weight (Laue and Rodhes, 1990). Other studies using sedimentation-velocity ultracentrifuge studies (Ritchie and Posner, 1982) and even the more mathematically sound equilibrium centrifugation (Posner and Creeth, 1972; Reid et al., 1990) also showed polydispersity in HS, and, for this very reason, conﬁrmed the ambiguity of the MW values obtained by ultracentrifuge methods. There is more reliability in molecular weight values for fulvic acids or organic matter dissolved in waters. There is a general agreement that the molecular weight of these humic molecules is in the range of 400–1500 as determined by a variety of methods (Aiken and Gillam, 1989; Stevenson, 1994; Wershaw, 1989). As shall be seen later, the smaller discrepancy among fulvic acid MW values can be ascribed to their larger hydrophilicity (small very acidic molecules) that prevents strong molecular associations by hydrophobic forces. However, a process of molecular association in solution can produce a polydisperse system of apparently high molecular weight also for fulvic acids. In fact, fulvic acid molecular weights consistently higher than when measured with osmometry or cryoscopy were reported when analyzed by ultraﬁltration and gel ﬁltration methods (Thurman et al., 1982). If the molecular weight value of HS has been a source of confusion, similar contradictions exist about the shape to be attributed to humic polymeric macromolecules. Globular shapes (Visser, 1964), ﬂexible linear conﬁgurations (Mukherjee and Lahiri, 1959), ellipsoidal shapes (Orlov et al., 1975), spheroid polyelectrolytes (Ghosh and Mukherjee, 1971), and randomly folding long chains (Cameron, Thornton et al., 1972), have all been proposed to describe the ever elusive polydisperse humic system. Ghosh and Schnitzer (1980) reconciled the different views by measuring surface pressures and viscosities of HS at different pH’s and neutral salt concentrations and adapting the results to relationships (the Flory and Fox and the Staudinger equations) which had been developed for real polymers. They explained the observed behavior of HS (uncharged matter at low pH and polyelectrolytes at high pH) on the basis of the polymeric theory and stated that HS macromolecular conﬁgurations are not unique but vary with the pH and

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

65

ionic strength of the medium in which HS are dissolved. They proposed that HS are rigid spherocolloids at high sample concentration and ionic strength and at low pH, whereas HS behave as ﬂexible linear polymers at low sample concentration and ionic strength and at high pH. This understanding had two major ﬂaws: it was based on whole humic extracts with full polydispersity and, despite the lack of a direct knowledge of the real molecular structure of HS, data were arbitrarily used in equations speciﬁcally derived for polymers. Nevertheless, this reversible coiling model for humic conﬁgurations soon became the most widely used to describe HS, though it does not explain all the behavior of HS. In applications of small-angle X-ray scattering to determine the particle sizes of whole humic acids in solution or of fractions separated by adsorption chromatography on crosslinked dextran gels (see Wershaw, 1989), it was found that HS formed molecular aggregates in solution and their size was a function of pH. It was concluded that the various fractions were chemically different and that the differences in aggregation behavior were a reﬂection of the interaction of different bonding mechanisms. These ﬁndings, coupled to other results showing that humic fractions from different sources have surface-active properties (Hayase and Tsubota, 1983), led to the formulation of a description of HS that was alternative to the random polymeric coil structure. Wershaw (1986, 1993) proposed that HS consist of ordered aggregates of amphiphiles composed mainly of relatively unaltered plant polymer segments possessing acidic functionality. In this model, HS are aggregates held together by hydrophobic (π –π and charge–transfer bonds) and H-bonding interactions and the hydrophobic parts of the molecules are in the interiors, whereas the hydrophilic parts make up the exterior surfaces. Ordered aggregates of humus in soils were depicted to exist as bilayer membranes coating mineral grains and as micelles in solutions. Wershaw’s model represented a major breakthrough, because it introduced the concept of aggregation of different particle sizes of humic constituents in contrast to the traditional view of humic polydisperse linear polymers. Nevertheless, the issue of HS molecular weight was not yet solved by the micelle-like model. The spontaneous aggregation of humic molecules into micellar aggregates was advocated by other authors (Engebretson and von Wandruszka, 1994, 1997) to explain ﬂuorescence quenching of pyrene, but an explanation of the results was still based on the polymeric nature of the aggregating humic molecules. However, the classical concept of an ordered micelle is hardly applicable to the heterogeneity of HS. In fact, the Critical Micelle Concentration (CMC) reported in the literature for HS, in the range of 1 to 10 g L−1 (Hayase and Tsubota, 1983), is much higher than that found for surface-active compounds giving regular micellar structures (Tanford, 1980). Notwithstanding its limitations, the concept of aggregation of hydrophobic parts of HS could explain (1) results from light scattering which showed that addition of Cu ions to dilute soil FA increased the amount of light scattered by the solution

66

ALESSANDRO PICCOLO

(Ryan and Weber, 1982), (2) the increased solubility of nonpolar compounds in humic solution because of partition/adsorption in the hydrophobic interior of HS (Carter and Suffet, 1982), and (3) the further release of humic matter through dialysis bags from an already extensively dialyzed HS when this is treated with an amphiphilic compound such as acetic acid or an electrolyte (De Haan et al., 1987; Nardi et al., 1988). The further understanding in explaining the environmental behavior of HS that was given by a molecular aggregation model with respect to the polymeric paradigm did not prevent recent hypotheses, based on pyrolytic analysis of HS, of schematic macromolecular structures for HA of up to about 100,000 Da (Shulten et al., 1998). Despite the many limitations inherent to analytical pyrolysis of HS (Saiz-Jimenez 1995, 1996), compounds identiﬁed by pyrolysis–mass spectrometry techniques as having a molecular mass hardly higher than 500 Da were used in computer molecular models to be arbitrarily linked together by covalent bondings and yield pictured views of large size branched polymers. Such large macromolecules are being presently proposed as models of humic molecules and used to explain the behavior of HS in soil (Schulten and Leinweber, 2000).

IV. SIZE-EXCLUSION CHROMATOGRAPHY OF HUMIC SUBSTANCES A. LOW-PRESSURE CHROMATOGRAPHY Gel-permeation chromatography or low-pressure size-exclusion chromatography on Sephadex crosslinked dextran gel columns, a method devised to desalt and purify proteins, has been extensively applied to humic fractions to evaluate molecular sizes and to obtain more size-homogeneous materials (see reviews of De Nobili et al., 1989, and Wershaw and Aiken, 1985). It soon became clear that separation by gel permeation is not a pure size fractionation (Determann and Walter, 1968) and that a number of interferences may occur with HS, namely, ionic-exclusion and adsorption chromatography. In the ﬁrst process, electrostatic repulsion between the negative charges present on both the dissolved HS and the dextran gel enhances the chromatographic velocity; whereas in the second process, hydrophobic interactions between HS and the stationary phase retard the chromatographic elution of HS (Lindqvist, 1967). The ionic strength of mobile phases should be sufﬁciently high to prevent electrostatic interactions, but not too high (> 0.5 M) to drive hydrophobic interactions (Chicz and Regnier, 1990). Much of the inconsistencies found in HS behavior in gel permeation have been repeatedly ascribed either to ionic (electrostatic) exclusion or to hydrophobic gel– solute interactions. Swift and Posner (1971), by eluting a Sephadex G-100 with

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

67

distilled water, showed that an increasing concentration of HS produced shifts of peaks from high- to low-molecular-size ranges. They disregarded the possible hydrophobic retardation and qualitatively explained this behavior as due to the repulsion between the HS and the negatively charged Sephadex, and “assumed” that repulsion should become stronger when a decreasing HS concentration might reduce ionic strength. Using a classical polyelectrolytic (polymeric) view, they claimed that “charged double layers on the solute and the gel extend further into the solution, resulting in effectively larger solute molecules and smaller pore sizes. Charge repulsion effects therefore occur at great distance leading to increased exclusion with decreasing sample concentration.” However, no measurements of actual charge density in different sample concentrations or in the gel bed were conducted by the authors to corroborate these theoretical assumptions. Notwithstanding the theoretical and experimental inconclusiveness of the HS behavior in gel permeation, buffers of high ionic strength were repeatedly recommended as mobile phases to suppress interferences due to ionic exclusion (DeNobili et al., 1989; Swift, 1989; Swift and Posner, 1971). High-ionic-strength buffers reduce the molar volume of HS in solution and, by thermodynamically favoring the hydrophobic associations of humic molecules, invariably produce chromatograms with a bimodal distribution. However, they also enhance hydrophobic adsorption on the gel solid phase. Piccolo and Mirabella (1987) used a 1 M Tris(2-amino-2-hydroxymethyl-1,3-propandiol) hydrochloride buffer at pH 9 to gel-permeate HS and observed, as expected, a decrease in molecular size of different humic fractions passing from Sephacryl 200S to Sephadex G150 and G100. By comparing elemental analysis of HS before and after elution in the Tris buffer, it was found that the content of humic-N had increased after elution in the buffer. An unwanted reaction with the N-containing Tris molecule may have then changed the molecular composition and behavior of HS. Moreover, gels became regularly brown after each run showing that the use of such a concentrated buffer induced permanent adsorption of humic material on the gel. Concomitantly, Yonebayashi and Hattori (1987) showed that addition of 2 M urea to either phosphate or borate buffers eliminates the noted hydrophobic adsorption during gel permeation of HS. The reason for the urea effect will be explained later. Adsorption of HS on columns has been traditionally accounted for by either salinity or changes in ionic strength (De Nobili et al., 1989; Lindqvist, 1967). Handerson and Hepburn (1977) attributed the appearance of new HS elution bands to adsorption when sodium acetate was added to the mobile phase and regarded these bands as unspeciﬁed “artifacts” created by adsorption effects. However, De Haan et al. (1987) demonstrated for the ﬁrst time that humic acids in saline solution were able to diffuse more freely across a dialysis membrane than when they were in lower-ionic-strength solutions. After comparing dialysis and gelﬁltration experiments they suggested, at variance with classical explanations, that changes in elution proﬁles due to ionic strength variation had to be attributed to

68

ALESSANDRO PICCOLO

real alterations of humic conﬁgurations rather than to any interactions between the gel and the HS. Further understanding came from the work of Yonebayashi and Hattori (1987) who studied several HS by gel ﬁltration through Sephadex G-75 under elution in 0.1 M phosphate or borate buffers in 2 M urea, thereby avoiding interferences from both ionic exclusion and adsorption on the column. They ﬁrst found that by increasing the pH of the buffer eluent from 4.7 to 11.2 the molecular size of HS increased progressively, though the fraction excluded at high molecular size (V0) was larger at pH 4.7. Interestingly, the elution pattern at pH 4.7 showed adsorption of humic material beyond the total volume (Vt) of the column. They attributed the behavior at low pH to an aggregation process of humic molecules but still advocated ionic repulsion as the reason for the size increase at a high pH. However, their experimental conditions (large ionic strength and urea) should have excluded an electrostatic effect. Yonebayashi and Hattori (1987) also showed that for both phosphate and borate buffers and for all pH’s from 4.7 to 11.2, the molecular size of HS increased signiﬁcantly with time of standing (from 5 min to 24 h) in the buffer solution before injection into the gel column. To explain these ﬁndings they had to invoke again a process of molecular association into micelle-like aggregates. In order not to contradict the polymeric paradigm of HS, they assumed that association occurred between humic molecules of large and small molecular size. However, this explanation did not account for what they further observed when HS were treated with ethanol–benzene, 6 N HCl, 1 N H2SO4, or 5 N NaOH before dilution in the buffer and separation by gel permeation. After the ethanol–benzene or acid treatments, the gel chromatograms were almost the same as those for the original humic acid, but the excluded fraction increased with an increase of standing time in the buffer solution. Conversely, after the alkaline treatment the excluded fraction disappeared and the gel chromatogram remained unchanged regardless of the standing time in the buffer solution before injection. Though they only invoked micellar aggregation to explain their ﬁndings these results were clearly a sign of conformational changes due to establishment of hydrophobic interactions (see later discussion). The conformational rearrangement could take place in the buffer solution for all pretreated samples except in 5 N NaOH where a full humic dispersion (disrupture of all hydrogen bondings) and hydrophobic rearrangement had already occurred. In an attempt to clarify the formation of micellar aggregates, Yonebayashi and Hattori (1987) also measured the surface tension of HS. They found that the surface tension of whole HS in neutral phosphate buffer, containing urea, decreased with an increase in sample concentration. This suggested that molecular association increases the surface activity of humic material. In fact, four fractions isolated by gel permeation revealed that only the ﬁrst high-molecular-size fraction behaved as a surface-active agent, whereas the three smaller fractions showed low surface activity. A cluster analysis grouping gel behavior of several HS with their

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

69

chemical and physical–chemical properties showed that HS with high surface activity (and a large fraction excluded at V0 in gel ﬁltration) were rich in long aliphatic chains, whereas HS with low surface activity (and a small excluded fraction) were composed of aromatic rings, a high content of –COOH groups and a low content of –OCH3 groups. From the results of Yonebayashi and Hattori (1987), which were similar to those of Handerson and Hepburn (1977), it may have been concluded that the associative nature of HS was proved by gel-ﬁltration studies. Fractions had different chemical composition, and their association into apparently large and surface-active fractions is favored by hydrophobic attractive forces among aliphatic compounds. Piccolo, Zaccheo et al. (1992) extracted humic acids from soils in controlled experiments which have been amended for 4 years with cattle manure (25 t ha−1 year−1) and different amounts of sewage sludge (25, 50, 100 t ha−1 year−1). They studied the chemical characteristics of the humic samples as well as their size distribution by low-pressure gel ﬁltration. They found that despite the lower methoxyl content (a parameter directly related to lignin content) of humic matter from sludgetreated soils, these samples had, contrary to the lignin-based polymeric model, an apparently larger molecular size than the samples from manure-treated soils. Moreover, molecular size of humic extracts increased signiﬁcantly with additions of sludge. They also showed, using infrared spectroscopy, that soils amended with sludge rich in alkyl components produced humic material with a much larger content of methyl and methylene groups than those amended with cattle manure richer in carbohydratic constituents. These results could well be viewed with those cited earlier to show that humic associations rich in hydrophobic compounds not only have large apparent molecular sizes but also are microbially more stable with time. New evidence that HS behavior in gel permeation studies, where changes in ionic strength occurred, cannot be ascribed to simple electrostratic interactions (ionic exclusion), as for a polyelectrolyte (a polymer holding multiple charges), came from the work of Ceccanti et al. (1989). They fractionated HS by ultraﬁltration and gel ﬁltration using both water and salt solutions as mobile phases. They found that 100% of the humic carbon was in the lowest molecular size fraction (<nominal 10,000 Da) when HS were ultraﬁltered in 100 mM pyrophosphate/HCl buffer at pH 7.1. However, no humic carbon was in the same fraction when ultraﬁltration was only in water and 43% was in the highest molecular size fraction (>nominal 100,000 Da). Contrary to previous qualitative research, Ceccanti et al. (1989) did measure by isoelectric focusing the negative net charge of HS fractions isolated by gel ﬁltration. Despite the charge being very similar, refractionation of fractions in the Sephadex G-50 column in water gave distinctly different elution patterns. Moreover, fractionation by water elution on the negatively charged polydextran Sephadex gave a molecular size distribution in accordance with that obtained on the uncharged PM-10 (polysulfone) ultraﬁltration membrane. They concluded

70

ALESSANDRO PICCOLO

that gel ﬁltration operated by a true “size” fractionation rather than by a “charge”dependent elution. The work of Ceccanti et al. (1989) also showed that the strength and the shape of humic associations controlled the elution volumes of HS fractions. A major problem in using size-exclusion chromatography to determine the size of humic fractions is the lack of adequate standards to calibrate the gel column. A calibration of Sephadex columns with presumably size-controlled humic fractions differed substantially from that obtained with globular proteins (Cameron, Swift et al., 1972). This proved that the hydrodynamic behavior of HS is different from that of globular macromolecules, and their molecular size may be more apparent than real and could be ascribed to molecular aggregation phenomena. Hayes (1997) reported that the ability to obtain humic fractions of homogeneous size by repetitive gel fractionation is not effective because fraction reprocessing results in the separation of smaller sized components. Moreover, reprocessing gives concentrations of components that would appear to have similar molecular sizes. The difﬁculty of obtaining MW values of HS by gel ﬁltration was noted by Reuter and Perdue (1981), who measured by osmometry the number-average molecular weight of the humic acid fraction excluded from G-50. They determined a value of 1231 Da by osmometry, whereas the published globular-protein-exclusion limit for the gel is 30,000 Da. Again, the reason for the humic material exclusion at such a high-molecular-weight cutoff may be due to molecular aggregation of small molecules into an apparent large size. This interpretation is conﬁrmed by the lack of differences between the infrared spectrum of the original humic acid and the spectra of fractions isolated through gels of different molecular weight cutoffs (Reuter and Perdue, 1981). The association of small humic molecules into different aggregate sizes may have been also the reason why Cornel et al. (1986) could not ﬁnd sufﬁcient similarity between the diffusion behavior of HS and that of synthetic polymers of known composition such as polyethylene oxides and polystyrene sulfonates, and why Summers et al. (1987) did not ﬁnd differences in infrared spectra of HS size-fractions isolated by ultraﬁltration, a separating technique that is slower than the velocity by which humic molecules aggregate or disaggregate in solution.

B. HIGH-PRESSURE CHROMATOGRAPHY The advent of commercially available rigid stationary phases which separate materials by size-exclusion chromatography at high pressures has improved enormously the possibilities to study the conformational behavior of HS. Piccolo, Conte, Cozzolino, and Spaccini (2001) enumerated the advantages that high-pressure size-exclusion chromatography (HPSEC) of HS has over the traditional low-pressure gel-permeation chromatography: (1) high reproducibility of

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

71

chromatograms (±5%), (2) relatively higher rapidity of analysis (about 60 min per chromatogram), (3) longer column life (more than 1000 injections per HPSEC column) provided there is regular maintenance, and (4) higher sensitivity to chemical changes in injected samples because of much lower loading mass. Conversely, low-pressure size-exclusion chromatography is usually poorly reproducible (adsorption occurs on the polysaccharide gel matrix), time consuming (16–20 h for one run), and it requires careful handling (to avoid preferential ﬂow) and frequent recharging of the gel bed (at least every three or four runs). These shortcomings have prevented the low-pressure gel-permeation chromatography technique from providing a more deﬁnite contribution to our understanding of humus chemistry (Piccolo and Conte, 2000). The application of HPSEC has been progressing in more recent years. Saito and Hayano (1979) were among the ﬁrst to use high-pressure stationary phases such as TSK-3000SW (glycol–ether-coated high-pore-volume rigid spherical silica gel) in stainless-steel columns (60 cm × 0.75 cm i.d.) to study a humic and a fulvic acid. They employed either water or 0.1 M NaCl as mobile phases. They again attributed the water elution of most humic and fulvic acids at the V0 of the column to electrostatic repulsion, as in low-pressure chromatography. Conversely, the salt solution produced a bimodal separation of HS. Unlike low-pressure Sephadex gel, a retardation based on adsorption forces was not observed. Miles and Brezonik (1983) conﬁrmed these ﬁndings. However, they found a hydrophobic adsorption on their Waters μBondagel E125 column by comparing methylated and nonmethylated HS. The lack of adsorption by HS on TSK-G 3000SW was noted by Becher et al. (1985), who showed, by eluting with a 0.02 M KHPO4 buffer at pH 6.5, that the average recovery of aquatic humic material from the column was 99% for three determinations. However, they also observed a shift of HS to higher elution volumes when passing from elution with water to phosphate buffer and to phosphate buffer in which 0.1 M sodium sulfate was added. They attributed the shift of the latter mobile phase to adsorption, but failed to conﬁrm this interpretation by measuring recovery of material as they did in the case of phosphate buffer alone. Vartianen et al. (1987) showed that the HPSEC resolution of aquatic HS on TSK columns was by far better with 0.01 M sodium acetate solution at pH 7 than by Tris and phosphate buffer solutions at the same concentrations. They found that the correlation between the sum of peak heights for the chromatogram was not linear with the absorbance values of dissolved HS at different concentrations, thereby conﬁrming variations of HS molar absorptivity with molecular size and deviations from the Lambert and Beer law (Stevenson, 1994). By studying the molecular size distribution of a special humic acid from an oxidized coal, Piccolo, Rausa et al. (1990) reported that both HPSEC and ultraﬁltration gave larger molecular sizes in distilled water than in 0.05 M NaNO3 solution. If this

72

ALESSANDRO PICCOLO

result was expected for HPSEC, based on the traditional ionic exclusion interpretation, it hardly explained the elutions through uncharged ultraﬁltering membranes (Amicon). As previously observed (Ceccanti et al., 1989; De Haan et al., 1987) this phenomenon could only be ascribed to a conformational rearrangement occurring in the electrolytic medium and consequent disaggregation of apparently large aggregates into smaller humic associations. Piccolo, Rausa et al. (1990) also noted that the HPSEC elution patterns of ultraﬁltered fractions differed signiﬁcantly depending on whether detection was by a UV-spectrophotometer or a refractive index (RI) detector. This was taken as an evidence that weight-average molecular weights of HS based on UV measurements are not representative of the real humic mass. The authors investigated the ultraﬁltered fractions by infrared spectroscopy and observed that large-molecular-size material (>500K Da, 67% of total mass) was much less hydrophilic than low-molecular-size fractions, thereby implying an association into apparently large aggregates by weak hydrophobic interactions. Berden and Berggren (1990) eluted dissolved HS from soil in a TSK 2000SW using different salts and buffers and different pH’s and ionic strengths. All previous ﬁndings on ionic and hydrophobic interferences were conﬁrmed. However, they unexpectedly found that a very dilute sample (64 times dilution) showed irreversible adsorption at ionic strength of 0.1 M but not at that of 0.15 M. They also could not explain, using the classical polymeric model, that a dissolved HS sample rich in Al (1.9 × 10−4 M) signiﬁcantly decreased the elution volume after having passed through a cation-exchange column and that it resumed the large retention volume of the pretreated sample when it was added again with 1.9 × 10−4 M of Al in the form of Al(NO3). Rausa et al. (1991) conducted a systematic study on the chromatographic conditions to be applied in HPSEC analysis of HS. By using Shodex (crosslinked sulfonated polystyrene–divinylbenzene copolymers) columns, they did not ﬁnd it necessary to use buffers and reported a 0.05 M NaNO3 solution and a ﬂow rate of 0.6 mL min−1 as the most suitable conditions to eliminate ionic exclusion and to obtain highly reproducible elutions. By using both UV–Vis and Refractive Index (RI) detectors they showed that chromatograms were independent of HS concentration in the range of (0.3–1%, m/v) and no adsorption phenomena were detectable. Chin and Gschwend (1991) found that when the ionic strength of a humic solution was lower than that of the mobile phase, isolation and refractionation of only the second peak gave an elution pattern similar to the ﬁrst fractionation. To explain this result, which indicated a mass distribution of HS in HPSEC, they abandoned the explanation given by Swift and Posner (1971) based on changes of double-layer thickness in a polyelectrolyte and based their ﬁndings on a hypothetical conﬁgurational rearrangement of a linear polymer. They proposed that an increase in salt concentration would fold the humic “macromolecule” upon

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

73

itself to form a more compact shape that would diffuse through the smaller pores of the HPSEC stationary phase and increase elution volumes. Conversely, low ionic strength allows humic macromolecules to uncoil and increase their hydrodynamic radius. This phenomenon should prevent the humic species from entering the smaller pores and hence elute quickly through the column. Chin et al. (1994) noted that molecular weights of HS obtained by HPSEC methods are higher than those measured by other methods. They attributed this discrepancy in part to the use of a UV spectrophotometer as the detector in the HPSEC system. In fact, they showed that the spectrophotometric properties of dissolved organic carbon vary with molecular weight. In comparison to a numberaverage molar absorptivity (ε) based on the whole sample, the higher-molecularweight fractions have greater than the average ε and appear to be more abundant than they actually are, while the low-molecular-weight fractions (with lower molar absorptivity) appear to be lower in concentration. However, as already shown by Piccolo, Rausa et al. (1990) and as will be discussed later, none of the fractions separated by HPSEC in UV mode are representative of the fraction concentration but only of the ε of chromophores contained in them. Following the early attempts of Piccolo, Rausa et al. (1990) and Rausa et al. (1991) to go beyond the UV detection of HS separation by HPSEC, von Wandruszka et al. (1999) applied refractive index (RI) and multiangle light scattering (MALS) detectors together with UV detector. Though molecular artifacts may have been created by using Tris buffer as eluent (see previous discussion on Tris), they conﬁrmed that UV-detected elution patterns were signiﬁcantly different from those obtained by MALS and RI detectors. The RI detector even showed some low-molecular-weight peaks which were not visible with the other two detectors. Similarly to low-pressure size-exclusion chromatography, the HPSEC method has been used to assess molecular sizes of both aquatic and terrestrial HS. However, absolute measurements of molecular weights of HS by column calibration are complicated by the lack of understanding their chemical structures. For any possible calibration standard, the hydrodynamic radius and the interaction with the stationary phase are bound to be different from those of humic substances (Wershaw and Aiken, 1985). A number of investigators (Beckett et al., 1987; Chin and Gschwend, 1991; Miles and Brezonik, 1983) have used the globular protein standards despite their recognized overprediction of humic substances molecular weights. Polystyrene sulfonates (PSS) are also popular standards in exclusion studies of humic substances (Beckett et al., 1987; Berden and Berggren, 1990; Miles and Brezonik, 1983; Peuravuori and Pihlaja, 1997). These compounds were reported to have a coiled conﬁguration similar to a single fulvic acid from the Suwannee River at pH 6.8 and an ionic strength of 0.1 M (Chin et al., 1994). However, PSS have an aromatic C content much greater than FA molecules (Berden and Berggren, 1990),

74

ALESSANDRO PICCOLO

and a different branching and crosslinking degree (Chin and Gschwend, 1991). Furthermore, the average pKa of PSS (benzenesulfonate has a pKa of 0.7 according to the “Handbook of Physics and Chemistry,” 1976) is different from the average pKa interval, about 4–7 (Stevenson, 1994), usually attributed to humic substances. Thus, a similarity in charge density between PSS and humic substances cannot be generally accepted. Inconsistencies in the size-exclusion behavior of PSS with ionic strength and pH changes have been already described and only partially explained by hydrophobic gel–solute interactions (Berden and Berggren, 1990; Chin and Gschwend, 1991). Nonionic, hydrophilic biomolecules such as polysaccharides have been successfully used to evaluate molecular size distributions of dissolved humic substances from different sources when eluted in dilute salts such as a 0.05 M NaNO3 at pH 7 and constant ionic strength (Piccolo, Rausa et al., 1990; Rausa et al., 1991). This neutral mobile phase was suitable for rapid and efﬁcient HPSEC of humic substances and for avoiding solute–stationary phase interaction phenomena (Berden and Berggren, 1990; Rausa et al., 1991). Based on the random-coil polymeric model, the 0.05 M concentration was considered to be the best compromise between the sample solubility and the ability of humic molecules to form a fully coiled conformation in solution (Ghosh and Schnitzer, 1980). Ballarin et al. (1999) calibrated a TSK 4000 PW column with polysaccharides standard to evaluate molecular weights of four fractions of HS (correctly measured by RI detector) previously separated by ultraﬁltration. They attempted to apply the “universal calibration procedure” to obtain the molecular weight values on the conceptual basis of the hydrodynamic volume of polymer molecules. To this aim they measured viscosity values for four ultraﬁltered fractions to be utilized in equations relating viscosity to molecular weight (see Clapp et al. (1989) for a discussion of the method). They failed to obtain reliable results for the application of the Mark–Houwink relationship and could not derive molecular weights by the universal calibration method. This failure was attributed to the heterogeneity in chemical composition of the HS. Again, the lack of knowledge of the chemical composition and, even more important, of the conformational structure of HS prevented evaluation of molecular size. A range of model compounds of known structure and different molecular weights has been used together with a PSS standard to evaluate HS molecular size (Peuravuori and Pihlaja, 1997). However, the authors found it impossible to correlate independent vapor–pressure osmometry molecular weight values with those obtained by HPSEC, which were four to ﬁve times greater than those measured by osmometry. It thus seems unlikely that other approaches (Kudryavtsev et al., 2000; Perminova et al., 1998) using indirect chromatographic descriptors such as small model compounds, in which there is no good information on their conformational structure, could have any practical use for assessment of HS molecular weights.

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

75

V. SUPRAMOLECULAR ASSOCIATIONS OF SELF-ASSEMBLING HUMIC MOLECULES A. ANALYTICAL EVIDENCE: LOW-PRESSURE SIZE-EXCLUSION CHROMATOGRAPHY Both low-pressure and high-pressure size-exclusion chromatography cannot provide reliable values of molecular weights of HS, because no adequate standards are available to simulate the complex arrangement of humic molecules in solution. Nevertheless, several studies have pointed out that HS behavior reﬂects that of molecular associations rather than of polymers, and that some conformational changes can occur during chromatography (Ceccanti et al., 1989; De Haan et al., 1987; Piccolo, Rausa et al., 1990; Yonebayashi and Hattori, 1987). Sizeexclusion chromatography may be then used to compare relative conformational changes of HS when they interact with other solutes. Piccolo et al. (1996a,b) have employed low-pressure size-exclusion chromatography to monitor the chromatographic variation of alkaline humic solutions which were brought to pH 2 with addition of different organic acids and then placed on a Biorad P100 Biogel column (nominal molecular range 5–100 kDa) under elution with 0.02 M borate (Na2B4O7) buffered at a pH 9.2. This buffer was chosen because it reduced the HS adsorption on the gel column more than a Tris or a carbonate/bicarbonate buffer, though it produced exactly the same bimodal distribution as the other buffers. When the control alkaline solution was chromatographed through the gel in the borate buffer, most of the humic material eluted at the void volume (V0) of the column (>100 kDa). Conversely, when the humic solution added with organic acids (all the monocarboxylic and most of the bicarboxylic acids) the total peak absorbance was shifted to elution volumes near the total column volume (Vt) suggesting a nominal MW<25 kDa. Mineral acids did not have any apparent inﬂuence on the chromatographic performance of the HS. A progressive shift from low to high elution volumes of the band was also observed when the pH of the solution was progressively brought to lower pH’s with acetic acid prior to insertion on the column (Fig. 1, I). The phenomenon was reversible since the band shifted back to low elution volumes when the action of acetic acid was progressively neutralized by the addition of 0.5 M KOH (Fig. 1, II). Supporters (Swift, 1999) of the traditional polymeric model of HS criticized the previously cited results not on the basis of a replication of the same experiment but on theoretical and qualitative interpretations of gel–solute interactions. However, the results of Piccolo et al. (1996a,b) could not be attributed to a buffering action of the organic acid towards the alkaline eluent because the amount of the different organic acids varied by 2 orders of magnitude, but the shift to larger elution volumes remained the same for all acids. Nor could an elution delay due

76

ALESSANDRO PICCOLO

Figure 1 Low-pressure size-exclusion chromatograms of a humic acid treated with acetic acid and eluted with a 0.02 M Na4B2O7 solution at pH 9.2 (I) and with a 0.1 M Na4B2O7 solution at pH 9.2 (II). Humic acid was treated before elution as follows: (A) dissolved at pH 11.8, (B) titrated with acetic acid to pH 6, (C) to pH 4.5, (D) to pH 3.5, (E) to pH 2; (F) the material brought to pH 2 was further back-titrated with KOH to pH 3.5; (G) back-titrated to pH 4.5; (H) back-titrated to pH 6; (I) back-titrated to pH 8.5; (J) the latter material at pH 8.5 was further roto-evaporated to attempt the elimination of the residual acetic acid (Piccolo et al., 1996a).

to a solid deposition on the gel and subsequent resolubilization by the eluent have been the cause of the shift, since the treated samples remained soluble at a low pH and entered the chromatographic elution immediately after deposition on the column. Nevertheless, if this had been the case, a progressive neutralization of the acidic buffering capacity by the alkaline eluent would have caused a smearing out

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

77

in the column of the polydisperse humic mixture and a diffuse chromatographic band from lower to larger elution volumes, instead of the sharp peak at the total volume (Vt) of the column. Furthermore, the ionic strength effect (De Nobili et al., 1989) could not be invoked for the reversible peak shift shown in Fig. 1 because elution in an ionic strength quencher, such as a borate buffer 10 times more concentrated, gave the same chromatographic change upon acetic acid and KOH additions. Piccolo et al. (1996a,b) considered their ﬁndings as an expression of the associative nature of humic molecules which aggregate into only apparent high-molecularsize materials. This interpretation was in accordance with previous research which showed that HS behaved as molecular associations when studied by size-exclusion chromatography (Ceccanti et al., 1989; De Haan et al., 1987; Piccolo, Rausa et al., 1990; Yonebayashi and Hattori, 1987). Moreover, laboratory observations have indicated that when acetic acid is added to HS that have been already extensively dialyzed, further small-sized components are released during subsequent dialysis (Nardi et al., 1988). These low-molecular-size fractions are a product of a conformational rearrangement and a chemical composition different from the bulk HS. In fact, the separated fractions were found to stimulate speciﬁc biological properties in plants, and were more biologically active than the whole humic material from which they were separated (Piccolo, Nardi et al., 1992). Piccolo et al. (1996a,b) described HS as micellar associations which are stabilized by predominantly hydrophobic forces at pH 7. They reasoned that the organic acids would penetrate into the inner (hydrophobic) core of the micellar structure while neutralizing the HS acidic functions from pH 7 to 2. The association between the organic acids and HS would occur because of the amphiphilic properties of the acids which are able to interact with both the hydrophilic and the hydrophobic domains of humic aggregates. Such interactions are capable of disrupting the weak forces which stabilize the humic conformations, and the subsequent chromatographic elution separates the small subunits that form the aggregate and prevents the reaggregation that would have occurred under static conditions.

B. ANALYTICAL EVIDENCE: HIGH-PRESSURE SIZE-EXCLUSION CHROMATOGRAPHY As previously outlined, low-pressure size-exclusion chromatography has severe limitations in both time requirements and reproducibility in the same gel bed. Hence, Conte and Piccolo (1999a) have compared the capacities of two commercial HPSEC columns to measure accurately and precisely the molecular sizes of HS. They determined for each column the chromatographic parameters, such as peak asymmetry factors (As), the number of theoretical plates (N), the coefﬁcient of distribution (kd), and the column resolution (Rs). Both columns had been

78

ALESSANDRO PICCOLO Table I Weight-Averaged (Mw), Number-Averaged (Mn), and Polydispersity (P) of HAs Obtained with TSK and Biosep Columns and Their Relative Standard Deviations (RSD)a Based on Three Replicatesb

Sample HA1 HA2 HA3 HA4

Mw Biosep

Mw TSK

33773 ± 615 (1.8) 28319 ± 1161 (4.1) 27489 ± 324 (1.2) 33976 ± 653 (1.9)

39730 ± 1192 (3.0) 32242 ± 312 (1.0) 30102 ± 1485 (4.9) 38300 ± 696 (1.8)

Mn Biosep

Mn TSK

6271 ± 83 11151 ± 409 (1.8) (3.7) 5182 ± 223 9487 ± 42 (4.3) (0.44) 4945 ± 15 8826 ± 207 (0.30) (2.3) 5774 ± 28 10555 ± 104 (0.48) (1.1)

P TSK

P Biosep

3.5

5.3

3.4

5.4

3.4

5.5

3.6

5.8

a

Percentage. Conte and Piccolo (1999a).

b

calibrated with polysaccharides of known MW, and so it was possible to obtain very reproducible weight-average (Mw) and number-average (Mn) molecular weight values for various HAs (Table I) even though the chromatographic resolution differed according to the gel pore size of each column (Figs. 2 and 3). In fact, the molecular sizes of the HAs were more distributed in the TSK column (Fig. 2) over a wider volume range than in the Biosep column (Fig. 3). However, the shape of the chromatographic peak for a single HA sample differed in the two columns. In the case of the TSK column, HA2 and HA3 showed a higher intensity for the ﬁrst peak than for the second (diffused) peak; whereas in the case of the Biosep column, the same humic material gave a diffused peak that had a higher intensity than the ﬁrst peak. This difference in peak intensities of HAs between the columns was not justiﬁed by the higher resolution of the TSK column. Moreover, both columns showed a similar molecular-size distribution for HA1 and HA4 without the inversion in peak intensities noted for HA2 and HA3. This discrepancy in the chromatographic behavior of the four samples was considered as evidence that the different UV response between the efﬂuents from these columns was due to different stabilities of humic associations rather than to differences in column properties. If the latter was true, a consistent difference in elution behavior should have been observed for all humic materials. The differences between the HA2 and HA3 chromatograms from the two columns can be explained by considering the response of HS and the mixture of molecules to electron excitation. While it is known that humic substances do not strictly follow Beer’s law and that their molecular absorptivity varies with molecular size (Chin et al., 1994; Stevenson, 1994), a recognized way to assess molecule

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

79

Figure 2 HPSEC chromatograms of four different humic acids eluted through a TSK column (Conte and Piccolo, 1999a).

mixtures in close association is to vary the resulting molar absorptivity according to the reciprocal orientation of chromophores (Cantor and Shimmel, 1980). This is because the close interaction of the transition dipole moment of an absorbing chromophore and the induced dipoles of neighboring chromophores depends on their reciprocal orientation and may either increase (hyperchromism) or decrease (hypochromism) the molar absorptivity (Freifelder, 1982; Cantor and Shimmel, 1980). The observed decrease in peak absorbance in the efﬂuent from the TSK column indicates that the molecular absorptivity of the ﬁrst peak is lower than that for the Biosep column. A most probable reason for the decrease in the UV-reading (hypochromism) is that a separation of molecules (or chromophores) from the high-molecular-size arrangement may have taken place during elution from the TSK column, thereby causing an alteration of the reciprocal orientation

80

ALESSANDRO PICCOLO

Figure 3 HPSEC chromatograms of four different humic acids eluted through a Biosep column (Conte and Piccolo, 1999a).

of the dipole moments among chromophores. This can be attributed to the higherresolution capacity of the TSK column. These results were in agreement with the association model proposed by Piccolo et al. (1996a,b) and indicated that humic materials are loosely bound molecular clusters which may be disrupted during HPSEC elution. Weakly bound associations are disrupted during the diffusion through the column. When gel pores are smaller than the aggregate hydrodynamic radius, small humic molecules are separated from the larger aggregate. The aggregate molar absorptivity is thus decreased, as is the absorbance reading of the eluting fraction at that molecular size. Although shear degradation of water-soluble polymers during the passage of a sample through a HPSEC column had not been reported (Barth, 1980), high-molecular-weight polystyrene was shear degraded in nonaqueous HPSEC (Kirkland, 1976). Shear disruption of the weakly bound humic conformation and consequent hypochromism may thus explain the lower UV readings for the

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

81

diffused peaks of HA2 and HA3 from the TSK column. The Biosep column has a lower separation capacity, and it did not give the same molecular disaggregation and hypochromic effect for the humic materials diffusing through its gel pores. Thus the absorbance reading of its second (diffuse) peak remained higher than the ﬁrst peak. The ﬁndings of Conte and Piccolo (1999a) showed that the behavior of HS in both high- and low-pressure chromatography is consistent with the model for association of small heterogeneous molecules into apparently large high-molecularsize structures. Since the apparent macromolecular structure of HS reﬂects an assemblage of molecules associated by weak forces, HPSEC chromatograms might provide a measure of the conformational stabilities of humic materials from different sources and an indication of their molecular composition. A HPSEC experiment was conducted by Conte and Piccolo (1999b) with a TSK 3000SW column to verify the stabilities of various humic solutions when the mobile phase (0.05 M NaNO3, pH 7, not absorbing light at 280 nm) was modiﬁed by small additions of methanol (4.6 × 10−7 M to a pH of 6.97), HCl (≤2.0 × 10−6 M to pH 5.54), and acetic acid (4.6 × 10−7 M to a pH of 5.69) so that the ionic strength of the eluting solution was always kept constant. In fact, all mobile phases had the same ionic strength (I = 0.0504 M) since the low ionic changes introduced in solution modiﬁed with low amounts of HCl and acetic acid (≤2.0 × 10−6 M) did not signiﬁcantly vary the I values. Moreover, sodium humate and fulvate solutions (0.5 g L−1) were previously titrated to pH 7 so that their dissolution in the mobile phase at pH 7 would have prevented any random occurrence of negative charges. This is to avoid any uncontrolled formation of negative charges on the solute which have been ascribed to change in the ionic strength of humic solutions, thereby affecting the volume of sample elution (Swift and Posner, 1971). UV–Vis and RI detectors were used to obtain the chromatograms of the molecular-size distribution of the humic materials. The objective was to compare the chromatographic behavior of chromophores (UV at 280 nm) with that of the real humic mass (RI). Modiﬁcations of the mobile phase resulted in the progressive decreases in the molecular sizes of humic solutions when going from the control solution to those to which methanol, HCl, or acetic acid were added. The HPSEC chromatograms obtained by means of the UV–Vis and the RI detectors for a HA isolated from a Danish agricultural soil are shown in Figs. 4 and 5, respectively. The decrease in molecular size was revealed by the shift toward increasingly larger elution volumes for both UV and RI detectors, and by the concomitant reductions in peak absorbance at the UV–Vis detector. The variation in molecular size distribution in the different mobile phases were reﬂected in the weight-average molecular weights of the humic material calculated from the chromatograms (Table II). The observed changes were not attributable to variations in ionic strength because this was kept constant in the different mobile phases, but could be attributable to the interactions of the added chemicals with the weakly associated

82

ALESSANDRO PICCOLO

Figure 4 HPSEC chromatograms of a humic acid from a Danish soil recorded with the UV–Vis detector. A control mobile phase (0.05 M NaNO3, pH 7, I = 0.05); B, same as A but 4.6 × 10−7 M in methanol (ﬁnal pH 6.97); C, same as A but to pH 5.54 with HCl; D, same as A but 4.6 × 10−7 M in acetic acid (ﬁnal pH 5.69) (Conte and Piccolo, 1999b).

macromolecular structure of the humic matter. In the case of the methanol addition, no new ions were introduced, nor were there changes in pH. Hence, the alteration of molecular-size distribution of the humic materials can be attributed only to the capacity of the CH3OH to form both van der Waals bonds with the hydrophobic humic components and hydrogen bonds with the oxygen-containing functional groups of the HA. Thus a very small amount of methanol in the eluting solution could disrupt the weak forces which temporarily stabilized humic associations in aggregates. The result can be interpreted as a dispersion of large aggregates into smaller humic molecules, a diffusion of these through smaller gel pores, and an overall decrease in molecular size (Table II). This effect was conﬁrmed by the RI chromatogram (Fig. 5) which showed a shift of humic mass towards elution volumes typical of lower-molecular-weight materials. While the RI detector in the HPSEC studies of HS (Piccolo, Rausa et al., 1990; Rausa et al., 1991; von Wandruszka et al., 1999) is an essential means to evaluate

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

83

Figure 5 HPSEC chromatograms of a humic acid from a Danish soil recorded with the RI detector. A, control mobile phase (0.05 M NaNO3, pH 7, I = 0.05); B, same as A but 4.6 × 10−7 M in methanol (ﬁnal pH 6.97); C, same as A but to pH 5.54 with HCl; D, same as A but 4.6 × 10 −7 M in acetic acid (ﬁnal pH 5.69) (Conte and Piccolo, 1999b).

Table II Weight-Average Molecular Weight (Mw) and Polydispersity (P) of HA from a Danish Soil in Different Mobile Phases as Determined by UV and RI Detectorsa Ab

UV RI a

B

C

D

Mw

P

Mw

P

%

Mw

P

%

Mw

P

%

35000 51130

3.6 5.2

34000 49670

3.9 5.3

2.8 2.9

6500 3100

1.9 1.1

81.4 93.4

5300 10720

1.4 1.3

84.8 79.0

Mw changes (%) as compared to control mobile phase (A) are reported (Conte and Piccolo, 1999b). b A, NaNO3 (pH 7); B, CH3OH (pH 6.97); C, HCl (pH 5.54); D, AcOH (pH 5.69); P, Polydispersion: weight-average molecular weight/number-average molecular weight (Mw/Mn).

84

ALESSANDRO PICCOLO

mass rather than chromophoric distribution of HS, it may not exclude completely the polymeric structure of HS as proposed by the classical random-coil model. In fact, supporters of that model may argue that the RI results may be due to a dramatic coiling down of large macromolecules rather than dispersion of weakly associated small molecules. Such a hypothetical explanation based on a decrease of the HS hydrodynamic radius had been previously used, but never proved, to account for elution shifts to higher elution volumes (Chin and Gschwend, 1991; Berden and Berggren, 1991). However, the traditional polymeric model of HS should not explain the experimental observation of the decreased peak intensities (seen in UV chromatograms where the various treatments were applied) compared with those for the control solution (Fig. 4). In fact, the substantial reduction in peak intensity revealed by the UV detector should be regarded as evidence for the hypochromic effect described earlier, as a result of real separation of small molecules for diffusion through pores. Not only were the peaks shifted to higher elution volumes (lower molecular sizes), but also the decreases in molecular absorptivities of the humic fractions indicated that the chromophores were drawn apart from each other because of the disrupting effect of methanol on the loosely associated humic structures. If the humic samples had been composed of polymeric macromolecules which, despite the constant ionic strength, had coiled down to give the changes observed in the elution proﬁles, one would have instead expected an increase of molecular absorptivity and UV readings with respect to control solution. The greater changes produced by decreasing the pH of the control solution (with addition of HCl) to 5.54 were due to a larger disruption (than for methanol) of the humic molecular associations (Figs. 4 and 5). The additional hydrogen ions in this mobile phase protonated the humic carboxylic functionalities (which were in their dissociated forms at pH 7 with the control mobile phase). A number of negative charges were then neutralized and hydrogen bonds were concomitantly formed between the complementary functionalities of the humic molecules. In that way the conformational stability that existed in the control solution was disrupted. Due to the parallel observation that a reduction in peak absorbance (hypochromism) was seen in the UV chromatograms, and that a shift of humic mass to high elution volumes was visible in RI chromatograms, this change could not simply be due to a volume decrease of the random coil, as was previously suggested (Berden and Berggren, 1991; Chin and Gschwend, 1991; Swift and Posner, 1971; von Wandruszka et al., 1999). Again, a more plausible explanation is that the heterogeneous humic conformation collapsed into molecular associations of smaller dimensions, but of greater thermodynamic stabilities than for the control solution. The chemical rationale for this behavior lies in the energy gained in hydrogen bond formation (ranging from 10 to 20 kJ mol−1) compared with van der Waals bonding (Schwarzenbach et al., 1993). Humic molecules, protonated by HCl addition, abandoned the loose conformation assumed at the pH 7 of the

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

85

control solution and formed relatively strong intermolecular hydrogen bonds. The concomitant large decrease in molecular size suggests that the weak association of apparently high molecular size, as observed for the HA in the control solution, must therefore have been due predominantly to weak intermolecular hydrophobic forces such as van der Waals, π–π , and CH–π (Nishio et al., 1998) bonds which hold small molecules together. The further decrease in molecular size distribution with acetic acid addition (Fig. 4 and Table II) can be attributed to the methyl group of the acetic acid. As for HCl addition, acidiﬁcation of the solution favors hydrogen bond formation. Moreover, the weak acidity of CH3 COOH (pKa 4.8) will allow a small number of undissociated species to exist at pH 5.69 and thus the formation of mixed intermolecular hydrogen bonds with humic molecules. As noted already, such energy-driven rearrangements will outweigh the weak humic associations in the control solution that were stabilized mainly by hydrophobic forces. However, the shift to higher elution volumes, and the general reduction of molecular absorptivity in the UV chromatogram (hypochromism), as well as the large shift to the total column volume of the humic mass in the RI chromatogram, would suggest that the methyl group (the apolar end) of acetic acid played an additional role in further disrupting the weakly bound humic associations. The apolar methyl group of acetic acid must have altered the residual hydrophobic forces which still stabilized the humic associations even after the hydrogen-bonding rearrangement had taken place. The work of Conte and Piccolo (1999b) on both humic and fulvic acids provided further direct evidence for the conformational model based on the reversible selfassociation of small humic molecules rather than on the macropolymeric randomcoil concept. Furthermore, using HPSEC it was shown that the molecular-size distribution of HS must be interpreted by a combination of two factors: the elution volume and the molar absorptivity of the chromatographic peaks. Earlier research had failed to address the combination of these factors because either closely similar HAs were analyzed only by UV detection in less sensitive low-pressure size-exclusion systems (Swift and Posner, 1971), or poorly UV-absorbing fulvic acids and/or aqueous dissolved organic matter (DOM) samples were used in studies with HPSEC systems without the support of a RI detector (Becher et al., 1985; Berden and Berggren, 1990; Chin et al., 1994; Chin and Gschwend, 1991). More recent HPSEC investigations which coupled UV, RI, and multiangle light scattering (MALS) detectors have also indicated different molecular-size distribution for HS according to the detector employed (von Wandruszka et al., 1999). Humic size reductions that were reported by Conte and Piccolo (1999b) upon modiﬁcations of the mobile phase may be attributed to a disruption of weak humic associations giving rise to separated smaller molecules rather than to the compaction of macromolecular coils. Intermolecular hydrophobic interactions appear to be the predominant binding forces for associations of relatively small humic

86

ALESSANDRO PICCOLO

molecules. These are consequently stabilized by the entropy-driven (the hydrophobic effect) tendency to exclude water molecules from humic aggregates, thus decreasing the total molecular energy in the solution (Israelachvili, 1994; Tanford, 1980). While ionic-exclusion interferences were carefully eliminated in the work of Conte and Piccolo (1999b), permanent adsorption of HS on the column was not found to occur (Conte and Piccolo, 1999a), in accordance with what was reported by Becher et al. (1985), who studied adsorption effects on the same TSK 3000GW column under elution with a 0.02 M phosphate buffer at 6.5. However, a modiﬁcation of the mobile phase with the addition of methanol, HCl, and acetic acid, although in very small amounts, may have produced pore-size changes which determine non-size-exclusion chromatography. In order to verify that the mobile phase modiﬁcation did not alter the exclusion properties of the HPSEC column, Piccolo, Conte et al. (2001) subjected polymeric standards of known molecular weight such as the negatively charged polystyrenesulfonates (PSS) and neutral polysaccharides (PYR) to size-exclusion chromatography in the same mobile phases used for HS by Conte and Piccolo (1999b). Moreover, for the very reason of the undisputed polymeric nature of both nonionic PYR and polyelectrolytic PSS, a comparison between the behavior of real covalently bound polymers and that of HS under the same chromatographic conditions would have provided information on the humic conformational structure. Figure 6 shows curves relating elution volumes and log of MW obtained for both PYR (RI detector) and for PSS standards (both UV and RI detectors) when these were dissolved and eluted in the four different mobile phases as in Conte and Piccolo (1999b). No signiﬁcant differences were shown from the curves obtained for the uncharged PYR standards in the different mobile phases (Fig. 6, I). The linear equations were similar and the slight differences well within the statistical analytical signiﬁcance (2% of relative standard deviation): y = 27.8–2.94x (r2 = 0.996) for the control solution, y = 28.6–3.13x (r2 = 0.997) for solution B, y = 29.1–3.22x (r2 = 0.990) for solution C, and y = 28.2–3.03x (r2 = 0.993) for solution D. The curves obtained for the polyelectrolytic PSS standards were found to be very similar in the different elution phases by either the UV or the RI detector (Fig. 6, II and III). The relative equations for the respective four mobile phases were A, y = 25.6–2.63x (r2 = 0.994); B, y = 25.0–2.53x (r2 = 0.986); C, y = 26.1– 2.76x (r2 = 0.998); D, y = 26.4–2.79x (r2 = 0.990), for the UV detector; and A, y = 25.4–2.68x (r2 = 0.986); B, y = 24.8–2.54x (r2 = 0.979); C, y = 24.8–2.54x (r2 = 0.986); D, y = 25.5–2.73x (r2 = 0.922), for the RI detector. Regardless of the charge density of the employed polymers, the lack of differences in their chromatographic behavior indicated that the slight variation in the composition of mobile phases was not sufﬁcient to alter either the conformational stability conferred to these macromolecular polymers by strong covalent bondings or their interactions with the stationary phase. Conversely, the size-exclusion

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

87

Figure 6 HPSEC calibration curves by RI detector of polysaccharide standards (I) and by both UV (II) and RI (III) detectors of polystyrenesulfonate standards of known MW ﬁrst dissolved and then eluted in different mobile phases. A, (䊊) control mobile phase (0.05 M NaNO3, pH 7, I = 0.05); B, () same as A but 4.6 × 10−7 M in methanol (ﬁnal pH 6.97); C, ( ) same as A but to pH 5.54 with HCl; D, (∇) same as A but 4.6 × 10−7 M in acetic acid (ﬁnal pH 5.69) (Piccolo, Conte et al., 2001).

88

ALESSANDRO PICCOLO

chromatograms of three different humic acids (HA1 from a volcanic soil, HA2 from an oxidized coal, and HA3 from a lignite), revealed by either UV or RI detectors, varied dramatically with the composition of mobile phases in both peak absorbance and elution volumes. The resulting Mw and polydispersity values as well as the percentage of reduction with respect to control mobile phase are shown in Table III. Piccolo, Conte et al. (2001) concluded that while slight modiﬁcations in the mobile phase did not affect the column capacity for size exclusion, the substantial difference between the response to HPSEC of the polymeric standards and that of HS under the very same chromatographic conditions proved that humic materials have a different conformational structure. As proposed earlier, this may well be a self-assembling association of relatively small and heterogeneous molecules instead of a coil of polymeric macromolecules. Piccolo, Conte et al. (2001) further studied the decrease of molar absorptivity observed in HS chromatograms with modiﬁcations in the mobile phases. To verify that such variations were not limited to the wavelength used to record chromatograms (280 nm), thereby being simply accountable to shifts of peak maxima, UV spectra of HS solutions were recorded over a range of wavelengths. Figure 7 shows that the three humic materials produced different absorbance values upon modiﬁcation of their solutions. While the addition of methanol caused a reduction Table III Weight-Average Molecular Weight (Mw) and Polydispersity (P) of Humic Samples in Different Mobile Phases as Determined by UV and RI Detectorsa Ab

B

C

D

Sample

Mw

P

Mw

P

%

Mw

P

%

Mw

P

%

HAI UV RI

23000 22700

2.8 2.9

29000 9200

3.5 2.1

+26.1 −68.8

8200 3400

1.9 1.3

−64.3 −85.2

4000 2250

1.2 1.1

−82.6 −90.2

HA2 UV RI

9700 9500

1.7 1.6

4200 2930

1.1 1.4

−56.7 −69.1

3600 2100

1.0 1.0

−62.9 −77.9

2900 2200

1.2 1.0

−70.1 −76.8

HA3 UV RI

17000 16650

2.0 3.5

7900 7790

1.5 1.7

−53.5 −53.2

9000 7990

1.8 1.5

−47.0 −52.0

3500 2200

1.1 1.0

−79.4 −86.7

a

Mw changes (%) as compared to control mobile phase (A) are reported (Piccolo, Conte et al., 2001). b A, 0.05 M NaNO3 (pH 7); B, As solution A but added with CH3OH (pH 6.97); C, As solution A but added with HCl (pH 5.54); D, As solution A but added with AcOH (pH 5.69); P, Polydispersity (Mw/Mn).

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

89

Figure 7 UV spectra (250–450 nm) of solutions (0.5 g L−1) of HA1, HA2, HA3 in A, control solution (0.05 M NaCl, pH 7, I = 0.05); B, same as A but 4.6 × 10−7 M in methanol (ﬁnal pH 6.97); C, same as A but to pH 5.54 with HCl; D, same as A but 4.6 × 10−7 M in acetic acid (ﬁnal pH 5.69) (Piccolo, Conte et al., 2001).

in molecular absorptivity at all wavelengths in HA2 and only for some wavelengths in HA1, it did not have any effect in HA3. The addition of HCl produced, instead, lower absorbance values than control solution for all HAs, except for the 290- to 330-nm range in HA1. After adding acetic acid, a signiﬁcant decrease in absorbance was observed for all humic solutions and over all wavelengths. These

90

ALESSANDRO PICCOLO

results showed that molar absorptivity of the bulk HS varied with the composition of the solution and was reduced by adding chemicals which disrupt their weakly stabilized molecular associations. Furthermore, the authors noted that the different light absorption of HS in modiﬁed solutions could have been related to their speciﬁc molecular composition as evaluated by 13C-CPMAS–NMR solid-state spectroscopy. The major discrepancy among UV spectra of humic matter was the different molecular absorptivity, as compared to HA2 and HA3, shown by HA1 in the methanol-added solution (Fig. 7). This was explained by the relatively larger aromaticity and carboxylic acidity measured in the CPMAS–NMR spectrum of HA2. The slight amount of methanol in solution B was believed to be capable of a larger conformational disruption of HA2 due to the formation of hydrogen bonds between methanol and humic carboxylate groups. These should have been sufﬁcient to disaggregate a humic association weakly stabilized mainly by π –π interactions among aromatic structures and thus decrease molar absorptivity in comparison to the control solution (Fig. 7). Conversely, the lower acidic-C and larger aliphatic-C content calculated from the CPMAS–NMR spectrum of HA1 suggested that a smaller number of intermolecular hydrogen bonds could be formed with methanol. In the case of HA1, the same bonds should have associated, instead of disrupting, the humic aggregates which had been already stabilized by hydrophobic forces among aliphatic moieties. The overall large hydrophobicity of HA3 possibly balanced the two opposite methanol effects towards a new humic conformation showing an absorptivity similar to that of the control solution. By the same reasoning, the authors were able to attribute to molecular properties of HS the differences in molar absorptivity observed in HPSEC chromatograms. HPSEC was applied by Piccolo, Conte et al. (1999) to conﬁrm previous results obtained by low-pressure gel-permeation chromatography (Piccolo et al., 1996a,b). Humic solutions were titrated to pH 3.5 using either a mineral acid (HCl) or different monocarboxylic (formic, acetic, propionic, and butyrric) acids, and then eluted through a HPSEC Biosep S2000 column (Phenomenex). In this case, a mobile phase of constant composition and pH 7 (0.05 M NaNO3) was used. The UV chromatograms for a soil humic acid from that study are shown in Fig. 8, and the changes with the different acids of the weight-average molecular weights are reported in Table IV. All HS used by the authors showed a decrease in the UV absorbance of chromatographic peaks when treated with either HCl or monocarboxylic acids. Again, such decreases in peak intensities were attributed to the hypochromic effect arising when the closely associated molecules in the conformational arrangements of the humic materials at pH 7 were separated by the additions of both mineral and monocarboxylic acids. The combined effects of reduction of peak intensity and the shift of peaks to higher elution volumes were taken as evidence of the disrupting effects of the added acids on the original conformations of the HS. As previously explained, molecular separation upon acid treatment was attributed to a formation

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

91

Figure 8 HPSEC chromatograms (UV detected at 280 nm) of a humic acid from an Italian volcanic soil eluted in a mobile phase made of 0.05 M NaNO3, pH 7, I = 0.05. A, control humic sample dissolved in the mobile phase prior to injection; B, as for A but with HCl added to lower pH of solution to 3.5; C, as for A but with formic acid added to lower pH of solution to 3.5; D, as for A but with acetic acid added to lower pH of solution to 3.5; E, as for A but with propionic acid added to lower pH of solution to 3.5; F, as for A but with butyric acid added to lower pH of solution to 3.5 (Piccolo, Conte et al., 1999).

of stronger intermolecular hydrogen bonds which altered the original conformations that were stabilized mainly by weaker hydrophobic interactions. The net effect allowed the separation of smaller molecules during HPSEC elution. Addition of organic acids not only further decreased the peak absorbances of HS but also enhanced their shifts to higher elution volumes, indicating a more extensive disruption of the original association compared with that obtained for HCl addition.

92

ALESSANDRO PICCOLO Table IV Weight-Average Molecular Weight (Mw) Values and Percentage of Variation with Respect to Control Solutions () of a Humic Acid from a Volcanic Soil Treated with Different Monocarboxylic Acidsa

Mw

Control

HCl

HCOOH

CH3COOH

CH3CH2COOH

CH3(CH2)2COOH

32961

37462 +13.7

23483 −20.8

23140 −29.8

31065 −5.75

33716 +2.29

a

Piccolo, Conte et al. (1999).

An interesting result of the study of Piccolo, Conte et al. (1999) was that the molecular-size changes could be related to the number of carbons present in the added monocarboxylic acids and to the hydrophilic/hydrophobic (HI/HB) carbon ratio of HS as measured by 13C-CPMAS–NMR spectroscopy. It was found that the higher the carbon content of organic acids and the lower the HI/HB ratio of humic materials, the larger the decrease of the average molecular size of HS. In the case of the humic acid whose chromatograms and Mw results are reported here, the HI/HB ratio was the highest among the investigated HS. The supposed high hydrophilicity of this humic material and the consequent high hydration of its molecular association allowed an easier penetration (and hence disrupting effect) of the most hydrated and hydrophilic monocarboxylic acids. In fact, the molecular-size reduction was largest for the addition of formic and acetic acids (Fig. 8 and Table IV). Conversely, HS containing more “hydrophobic” carbons were reduced to lower molecular sizes by the action of monocarboxylic acids with higher numbers of carbon atoms, such as propionic and butyrric acids. The work of Piccolo, Conte et al. (1999) represents additional evidence that HS do not behave as polymeric random coils. It also indicates that the HPSEC technique can be used to reproducibly decrease the apparently large dimensions of humic associations into fractions of smaller molecular sizes as the result of simple interactions with monocarboxilic acids. The authors conﬁrmed the results obtained by Piccolo et al. (1996a,b) by low-pressure size-exclusion chromatography and indicated that the extent of size reductions of humic associations depends on the aliphatic chain lengths of the acids and the hydrophobicities of HS. The difference between the effect of HCl observed by HPSEC and that by the low-pressure mode may be attributed to the larger resolution and sensitivity of the HPSEC technique. By using a similar column (Polysep GFC-P3000 by Phenomenex), Piccolo, Conte et al. (2000) compared the chromatographic behavior of HS with that of undisputed polymers such as neutral PYR and polyelectrolytic PSS, when solutions (in 0.05 M NaNO3) to be injected into the HPSEC system were titrated from pH 7 to 3.5 with either HCl or acetic acid. Contrary to the work of Piccolo, Conte et al. (2001), the mobile phase was kept constant while the solutions under analysis were modiﬁed. It was found that curves of retention volumes versus

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

93

Ln (MW) at either the RI detector for PYR and PSS polymers of known MW or the UV detector for only PSS standards did not show signiﬁcant differences between control and samples added with either HCl or acetic acid. This result indicated that the capacity of the Phenomenex column in aqueous media to exclude by size was not altered by injecting the 100 μl of standard solutions brought to pH 3.5 with a very small amount of acids. However, when different modiﬁed humic solutions were chromatographed with the same mobile phase, the elution of the acid-added humic samples differed signiﬁcantly from the control humic solution. Figure 9 shows the elution proﬁles

Figure 9 HPSEC chromatograms by RI detector (negative mode) of four humic acids dissolved (0.2 g L−1) in mobile phase (0.05 M NaNO3) at pH 7 and in the same solution, but with pH lowered to 3.5 by addition of either HCl or acetic acid (Piccolo, Conte et al., 2000).

94

ALESSANDRO PICCOLO

of control and modiﬁed solutions as detected by the RI detector for the four HS studied by Piccolo, Conte et al. (2000). This ﬁgure indicates that addition of acetic acid determined the largest shift of humic mass to low-molecular-size range. As in previous experiments, these results were considered as evidence that a humic association, stabilized only by weak dispersive forces at pH 7 (mainly hydrophobic forces), varies considerably with respect to its conformational arrangement and its size distribution upon treatment with an acid. In contrast, covalently linked polymers have more stable conformations, and their HPSEC behavior is not affected by an interaction with either a mineral or an amphiphilic acid such as acetic acid. Cozzolino et al. (2001) studied the effect of organic acids of plant, microbial, or anthropic origin on the molecular-size distribution of dissolved HS. They used a Phenomenex Biosep S2000 under elution with a 0.05 M NaNO3 solution to evaluate size changes in four different HS upon addition of hydroxy-(glycolic and malic), keto-(glyoxylic), and sulfonic (benzenesulfonic and methanesulfonic) acids. All HS showed a decrease in peak absorbance when humic matter was dissolved in the HPSEC mobile phase at pH 7, and the pH of the solution was lowered to 3.5 by acid addition before analysis. This effect was generally accompanied by an increase in peak elution volumes. The combination of the two effects led the authors to compare the changes observed for treated samples in relation to the control by measuring the total area of chromatograms (Table V). The results were also explained with a disruption of supramolecular humic associations into smaller-sized but richer energy conformations brought about by the formation of mixed intermolecular hydrogen bonds upon acid treatment. Table V shows that the hydroxy-bicarboxylic malic acid was the most effective in disrupting the original humic associations among the carboxylic acids. This was

Table V Variation (%) with Respect to Control of Total Area in HPSEC Chromatograms of Four Different Humic Acids upon Addition of Acidsa Humic substances Acid

HA1

HA2

HA3

HA4

HCl Glyoxylic Glycolic Malic Methanesulfonic Benzenesulfonic

−51.5 −23.2 −54.6 −65.3 −73.0 −36.7

−14.4 −24.0 −46.6 −69.6 −50.5 −52.8

−63.0 −20.8 −47.8 −72.6 −74.6 −43.2

−56.6 −21.1 −50.6 −73.1 −54.7 −25.8

a

Cozzolino et al. (2000).

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

95

attributed to its greater capacity to form new hydrogen bonds with complementary functions of HS. In fact, malic acid is a bicarboxylic acid (HOOC–CH2–CHOH– COOH) with two carboxyl groups (pKa1 = 3.4; pKa2 = 5.11) which can be either protonated or partially dissociated at pH 3.5, thereby allowing for a larger number of mixed hydrogen bonds on its oxygen-containing functions than for either the strong hydrochloric acid that is only a proton donor or the more weakly acidic monocarboxylic glycolic and glyoxylic acids. The extent of conformational variation was related not only to the pKa’s of acids but also to the chemical and stereochemical afﬁnity of humic components which may allow penetration of acids into the inner humic domains depending on acid structures. For instance, the strongly acidic methanesulfonic and benzenesulfonic acids showed effects which varied with the humic properties (Table V). While methanesulfonic acid was more effective in conformational disruption of HS predominantly containing aliphatic and alkyl moieties, the size distribution of an aromatic-rich humic material was equally varied by benzenesulfonic acid because of the probable larger π –π interactions with aromatic humic components (Fig. 10). Naturally occurring bicarboxylic acids with a progressively higher number of carbon atoms (oxalic, malonic, succinic, and glutaric acid) were evaluated by Cozzolino and Piccolo (2002a) as to their capacity to affect the conformational structure of humic associations. Chromatograms of dissolved HS showed a decrease of peak absorbance as well as an increase in peak elution volume when the solution pH was lowered from 7 to 3.5 by the addition of bicarboxylic acids before HPSEC analysis. The extent of size variation brought about by bicarboxylic acids was in the order: oxalic < malonic < succinic < glutaric (Table VI), and depended on the capacity of acids to come in contact with the inner hydrophobic domains of humic associations and form mixed H-bonds between their carboxyl groups and the complementary humic functions. The contact of bicarboxylic acids with humic hydrophobic associations was found to be a function of the pKa’s of acids which are, in turn, a result of the methylene content in each acid. An increase in number of carbon atoms in the acid molecule enhanced the pKa’s of carboxyl groups and hence their degree of protonation at pH 3.5 at which the acid interacted with dissolved HS. Protonation of acidic groups progressively reduced electrostatic repulsion from humic molecules and made the acid less hydrophilic. More hydrophobic bicarboxylic acids were capable of being in closer contact with the hydrophobic domains of the weakly bound humic associations and capable to promoting more extensive conformational rearrangements by hydrogen bond formation. In line with the importance of hydrophobic forces in the stabilization of humic associations, the authors reported that bicarboxylic acids, with increasing hydrophobicity, reduced humic molecular sizes generally more effectively than HCl which, by providing a small but hydrated proton, could only affect surface hydrophilic components of HS.

96

ALESSANDRO PICCOLO

Figure 10 HPSEC chromatograms of HA2. A, control solution at pH 7; B, as for A but pH lowered to 3.5 with HCl; C, as for A but pH lowered to 3.5 with methanesulfonic acid; D, as for A but pH lowered to 3.5 with benzenesulfonic acid (Cozzolino et al., 2001).

C. PREPARATIVE HPSEC AND CHARACTERIZATION OF FRACTIONS The indication of a supramolecular nature of HS obtained by the analytical HPSEC studies described previously has caused Piccolo, Conte, Trivellone et al. (2001) to apply a preparative HPSEC column to separate more chemically homogeneous fractions to be subjected to chemical and spectroscopic characterization. A preparative Biosep SEC-S-2000 (600 mm × 21.2 mm i.d.) column by Phenomenex eluted with NaCl (2.89 g L−1)/NaN3 (0.3 g L−1) solution at pH 7.0 was used to fractionate solutions (0.6 g L−1) of a humic acid from lignite. Another fractionation was conducted on the same HS after disrupting the humic superstructure by bringing the pH to 3.5 with acetic acid.

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

97

Table VI Variation (%) with Respect to Control of Total Area in HPSEC Chromatograms of Four Different Humic Acids upon Addition of Acids and Standard Deviations (SD)a Humic substances Acid HCl SD Oxalic SD Malonic SD Succinic SD Glutaric SD a

HA1

HA2

HA3

HA4

−51.5 ± 0.7 −35.8 ± 0.2 −59.3 ± 0.2 −65.5 ± 0.3 −64.2 ± 0.5

−14.4 ± 0.1 −41.7 ± 0.5 −48.2 ± 0.8 −50.0 ± 0.9 −72.2 ± 1.2

−63.0 ± 1.1 −29.4 ± 0.1 −48.3 ± 0.7 −46.2 ± 0.3 −62.7 ± 1.4

−56.6 ± 0.3 −56.9 ± 0.7 −70.6 ± 0.1 −71.4 ± 1.5 −83.8 ± 1.7

Cozzolino and Piccolo (2001).

Six fractions were collected from the HPSEC separation of the HA solution at pH 7.0 (Fig. 11), whereas eight fractions were obtained from the separation of the humic solution treated with acetic acid to pH 3.5 before HPSEC injections (Fig. 12). Fractionation was run continuously with an autosampler, a fraction collector, and the chromatographic pattern monitored by a UV detector.

Figure 11 Preparative HPSEC chromatogram of an untreated humic acid from lignite at pH 7 (Piccolo, Conte, Trivellone et al., 2001).

98

ALESSANDRO PICCOLO

Figure 12 Preparative HPSEC chromatogram of a humic acid from lignite titrated with acetic acid to pH 3.5 before injection (Piccolo, Conte, Trivellone et al., 2001).

Reproducibility was excellent (CV<5%) over more than 100 runs for each fractionation series. Fractions were ﬁrst freeze-dried, dialyzed, and freeze-dried again. Figures 11 and 12 indicate that molecular-size distribution of the untreated HS was signiﬁcantly different from that of the HS treated with acetic acid, with the latter showing much less peak absorbance (280 nm) and a number of additional peaks appearing at higher elution times (120–180 min). Acetic acid was conﬁrmed to disrupt the humic superstructure, and the higher resolution of the preparative column separated more fractions than ever observed by analytical HPSEC columns. Fractions were analyzed by a Curie point (610◦ C) pyrolysis-gas chromatography–mass spectroscopy (Pyr–GC–MS) technique and by 1H-NMR spectroscopy. Total ion chromatograms by Pyr–GC–MS showed that fractions had a signiﬁcantly different chemical composition after treatment with acetic acid, thereby conﬁrming that acetic acid had caused rearrangement of the humic conformation in solution and that preparative HPSEC is adequate for HS fractionation. In particular, Piccolo, Conte, Trivellone et al. (2001) observed that the unsaturated aliphatic chains, which were mostly found in fraction 1 in the untreated sample (Fig. 11), were distributed in fractions of lower molecular size in the acidtreated sample (Fig. 12). The majority of aromatic moieties were found in fraction 3 for the untreated sample, whereas the aromatic systems shifted to lighter fractions after treatment with acetic acid. Interestingly, the lightest fractions collected by HPSEC separation after acetic acid addition to the humic sample showed signals of carbohydrates which were not visible in any fraction of the untreated sample, as if the acidic treatment had liberated material which was incorporated in the original conformation. These results thus suggested that the new conformation(s) produced by acetic acid treatment allowed interactions of humic constituents with

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

99

the column stationary phase. This produced a progressive exclusion delay of the different components which were, consequently, collected in more homogeneous fractions. 1 H-NMR spectra of HPSEC fractions not only conﬁrmed the ﬁndings by Pyr– GC–MS analysis, but also showed that the chemical composition of fractions was much simpler after acetic acid treatment. From the 1H-NMR spectra of separated humic fractions, it was even possible to assign resonances to chemical structures of small molecules. The authors proposed that the following structures should be present in the original humic superstructure:

Piccolo, Conte, Trivellone et al. (2001) showed once more that HS are supramolecular association of small heterogeneous molecules, and their loose conformational structure can be disaggregated by the action of organic acids as previously indicated (Conte and Piccolo, 1999a,b; Cozzolino et al., 2001; Cozzolino and Piccolo, 2001a; Piccolo and Conte, 2000; Piccolo, Conte et al., 1999). They proved by chemical and spectroscopic means that treatment with acetic acid separated fractions with a reduced chemical complexity, and that the relatively small molecules which make up the humic supramolecular associations could be identiﬁed by more resolved pyrograms and 1H-NMR spectra. The combination of fraction separation by HPSEC with chemical and spectroscopic characterization showed that the fraction with the largest apparent molecular size is actually composed of a number of either saturated or unsaturated alkyl chains of relatively low molecular size held together in random association because of the hydrophobic effect of the aqueous medium. Fractions of lower apparent molecular size are composed of small aromatic systems and hydrophilic compounds. It must be noted that these conclusions match with previous research which attempted characterization of HS fractions separated by either low-pressure gel permeation or ultraﬁltration (Handerson and Hepburn, 1977; Piccolo, Rausa et al., 1990; Yonebayashi and Hattori, 1987).

100

ALESSANDRO PICCOLO

D. SOLUTE–GEL–ELUENT AS AN INTERACTIVE SYSTEM IN SIZE-EXCLUSION CHROMATOGRAPHY Though the previous results by HPSEC should be regarded as free of interferences from ionic exclusion and adsorption (coefﬁcient of variation for more than ﬁve runs constantly < 5%), a further check for possible hydrophobic adsorption phenomena was conducted on the Polysep column. Seven different monomers which may be constituents of HS (Stevenson, 1994) were subjected to HPSEC separately and in full mixture (1 g L−1) under elution with a phosphate buffer at pH 3. The size-excluded monomers alone gave the elution volumes reported in Table VII while the mixture of the seven monomers produced an elution proﬁle with only two peaks (Fig. 13). It is interesting to note that the retention volumes of the two peaks for the mixture did not correspond to any of the elution volumes found for the single monomers (Table VII). The ﬁrst peak of the mixture eluted (23.07 mL) before any monomer, with the exception of the highly hydrophilic glucosamine, and the more intense second peak eluted before all other monomers except for dihydroxyphenilacetic and gallic acid. These results suggest that when the monomers are together in solution, they form weak mutual associations with a resulting hydrodynamic radius larger than when they are eluted alone. Based on the larger single elution volumes of the more hydrophobic monomers ( hydrocaffeic acid, resorcinol, and catechol), it can be inferred that their elution is retarded by adsorption on the stationary phase of the column. Conversely, when the same compounds are in mixture with other monomers the thermodynamic drive to reduce exposure of hydrophobic components to the aqueous medium increases the reciprocal attraction of monomers. This interpretation is substantiated by the observation that the retention volumes of mixture peaks decreased in value with time of standing before injection (Table VII), thereby indicating a further hydrophobic strengthening of the

Table VII HPSEC Retention Volumes (mL) of Different Monomers and Mixture of the Same Monomers on a Column Polysep GFC-P3000 Eluted with a Phosphate Buffer at pH 7 Dihydroxyphenylacetic acid Gallic acid Protocatechuic acid Hydrocaffeic acid Glucosamine (RI) Resorcinol Catechol Mixture of the above (0 h) Mixture of the above (after 144 h standing)

23.39 25.49 25.94 26.60 20.23 26.90 26.95 First peak: 23.07; second peak: 25.91 First peak: 23.01; second peak: 25.70

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

101

Figure 13 HPSEC chromatogram of a mixture (1 g L−1) of seven monomeric precursors of HS eluted by a phosphate buffer at pH 7 through a Phenomenex column (Polysep GFC-P3000). (1) Dihydroxyphenylacetic acid; (2) gallic acid; (3) protocatechuic acid; (4) hydrocaffeic acid; (5) glucosamine (RI detected); (6) resorcinol; (7) catechol.

monomer association. A consequence is that the hydrophobic interactions of single monomers with the column matrix are reduced while exclusion of the monomer association is expedited. Evidence that hydrophobic adsorption on the column is not irreversible but only affects elution volumes was obtained by eluting progressively less concentrated solutions of the monomer mixture. The absorbance decrease of both peaks shown by the mixture gave a highly signiﬁcant linear correlation with the decrease in mixture concentration, as expected by the Lambert and Beer law (Fig. 14), thereby proving that no humic-like monomers were lost due to permanent adsorption on the column. The behavior of the monomer mixture in HPSEC may well be assimilated to that of associations of humic molecules. When the apparently large-sized weakly

102

ALESSANDRO PICCOLO

Figure 14 Relationship of peak absorbances (ﬁrst and second peak of Fig. 13) versus concentration of a mixture of seven monomeric precursors of HS eluted by a phosphate buffer at pH 7 through a Phenomenex column (Polysep GFC-P3000). (1) Dihydroxyphenylacetic acid; (2) gallic acid; (3) protocatechuic acid; (4) gydrocaffeic acid; (5) glucosamine (RI detected); (6) resorcinol; (7) catechol.

bound humic associations are altered by the action of organic acids or by other species (uncharged compounds, electrolytes, cations) smaller humic clusters are separated, and their elution is retarded by size exclusion but also because the hydrophobic interaction with the column matrix has changed. Other evidence that humic molecules are superstructural associations in aqueous solutions is given by HPSEC elution of HS in a solution of 8 M urea. Concentrated urea is employed to disrupt protein–protein interactions and solubilize aggregating hydrophobic proteins for their further separation (Hjelmeland, 1990). Concentrated urea has been proposed also for low-pressure size exclusion and PAGE separation of HS since it was believed to produce elution patterns similar to 0.1 M Tris-HCl buffer at pH 9 (Trubetskoj et al., 1997). Figure 15 shows the elution proﬁles of three different puriﬁed HS (0.2 g L−1) dissolved in 8 M urea and eluted in the same solution. The chromatograms were detected by both UV and RI detectors in series. This experiment is illustrative

Figure 15 UV- and RI-detected HPSEC chromatograms of humic acids dissolved and eluted in 8 M urea solution. I, humic acid from an oxidized coal; II, humic acids from a lignite, III, humic acid from a Danish agricultural soil.

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

103

104

ALESSANDRO PICCOLO

of the capacity of urea to separate hydrophobic from the hydrophilic components of HS. The chromatograms show a net separation of humic matter: a highmolecular-size fraction at the column V0 visible only by the UV detector and a lowmolecular-size fraction eluting at the column Vt detected only by the RI detector. The humic hydrophobic molecules are strongly associated through a hydrophobic effect induced by the highly concentrated urea that interacts more favorably than hydrophobic molecules with the network of the water structure. Moreover, since urea is known to form complexes with nonionic detergents (Hjelmeland, 1990), it may be also possible that complexes between urea and nonionic humic hydrophobic compounds enhanced the hydrodynamic radius of the hydrophobic associations which thus elute at the earliest exclusion volume of the column. Differences among HS could also be noted. The humic acid from an oxidized coal was the only one to show two well-separated peaks at the RI detector (Fig. 15, I-RI), while only one very intense peak (>1000 mV) at column V0 was shown by the UV detector (Fig. 15, I-UV). This indicates that while lightabsorbing chromophores (280 nm) with a high molar absorptivity were excluded rapidly from the column, their concentration in the humic sample was relevant and of a similar order of magnitude to the nonabsorbing hydrophilic, probably ionized, compounds detected by the RI detector at the total exclusion volume of the column. The humic acid from lignite gave a similar complete separation between the large-sized hydrophobic chromophore fraction (Fig. 15, II-UV) and the small-sized ionized and hydrophilic components shown by the RI detector (Fig. 15, II-RI). However, while the molar absorptivity of the excluding chromophores was still high (>800 mV), the lack of corresponding peaks at the RI detector indicated that their concentration in the sample was much less than that in the hydrophilic components excluded at large elution volumes. Identical behavior was shown by the humic acid extracted from an agricultural soil. Also for this sample, the chromophore association excluded at the column void volume (Fig. 15, III-UV), was hardly representative of the mass of humic sample since the corresponding peak at the RI detector was almost irrelevant in comparison to the signal of the small-sized nonabsorbing hydrophilic constituents (Fig. 15, III-RI). The two experiments reported here (Piccolo, unpublished data) suggest that in the HPSEC analysis of associations of small molecules such as HS, the stationary and mobile phases loosen their properties associated to speciﬁc single molecules. They rather become part of an interactive system with the mixture of solutes and are the whole solute–gel–eluent system that produces size exclusion of the association under analysis. Moreover, more than ionic exclusion, so often accounted for in the past to explain the SEC behavior of HS, is the humic conformational structure and its degree of potential hydrophobic interaction between the associated molecules and either the stationary or the mobile phase that determines the elution volume of a humic association in aqueous solution.

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

105

E. CONCEPTS OF SUPRAMOLECULAR ASSOCIATION OF HUMIC SUBSTANCES The described experiments by either analytical or preparative size-exclusion chromatography cannot be explained by analytical interferences or the traditional polymeric model of HS. They can rather be interpreted with the concept of loosely bound humic supramolecular associations. By this concept one can imagine HS as relatively small and heterogeneous molecules of various origin which selforganize in supramolecular conformations. Humic superstructures of relatively small molecules are not associated by covalent bonds but stabilized only by weak forces such as dispersive hydrophobic interactions (van der Waals, π –π , and CH–π bondings) and hydrogen bonds, the latter being progressively more important at low pH’s. Hydrophilic and hydrophobic domains of humic molecules can be contiguous to or contained in each other and, with hydration water, form apparently large-molecular-size associations. In humic supramolecular organizations, the intermolecular forces determine the conformational structure of HS, and the complexity of the multiple noncovalent interactions control their environmental reactivity. The deﬁnition given by Lehn (1995) may well be applied to HS: “supramolecular assemblies (are) molecular entities that result from the spontaneous association of a large undeﬁned number of components into a speciﬁc phase having more or less well-deﬁned microscopic organization and macroscopic characteristics depending on its nature (such as ﬁlms, layers, membranes, vesicles, micelles, mesomorphic phases, solid-state structures, etc.).” By the concept of supramolecular association, the classical deﬁnitions of humic and fulvic acids should be reconsidered. Fulvic acids may be regarded as associations of small hydrophilic molecules in which there are enough acidic functional groups to keep the fulvic clusters dispersed in solution at any pH. Humic acids are made by associations of predominantly hydrophobic compounds (polymethylenic chains, fatty acids, steroid compounds) which are stabilized at neutral pH by hydrophobic dispersive forces (van der Waals, π –π , and CH–π bondings). Their conformations grow progressively in size when intermolecular hydrogen bondings are increasingly formed at lower pH’s, until they ﬂocculate.

VI. CHEMICAL AND SPECTROSCOPIC EVIDENCE OF SUPRAMOLECULAR ASSOCIATIONS The model of self-assembling supramolecular association of HS is related to the reciprocal afﬁnity of molecules in aqueous solution. Molecules tend to associate by intermolecular forces (Israelachvili, 1994) and the strength of the

106

ALESSANDRO PICCOLO

association depends on their molecular structure. Particularly strong associations are formed by apolar compounds via the hydrophobic effect (Tanford, 1980). A humic supramolecular association in solution is thus formed by the self-organization of hydrophobic and amphiphilic compounds which are progressively isolated from the network of water structure because such separation results in an increase in the system entropy and in the overall energy stabilization of different humic molecules into a superstructure. The importance of hydrophobic humic components in phenomena of aggregation in solution and on surfaces as well as in controlling the reactivity of HS is well documented (Wershaw, 1999). Investigations by ﬂuorescence quenching techniques provided evidence of hydrophobic microdomains in loose humic associations (Engebretson and von Wandruszka, 1994; Morra et al., 1990). By following diffusion of 1,2-dichloroethane into humic matter, Aochi and Farmer (1997) showed that there are discrete microregions of different polarity in humic structures. Chien et al. (1997) studied 19F-NMR spectroscopy and the interactions of a triﬂuoromethylated atrazine with a soil humic acid by measuring the NMR relaxation of atrazine in the presence of both hydrophilic and hydrophobic paramagnetic probes. They conﬁrmed the existence of hydrophobic domains by showing that atrazine occupied a domain of HS accessible only to neutral hydrophobic molecules. Similar results were reached by Piccolo et al. (1998), who studied adsorption isotherms of atrazine on HS extracted from two soils with four different extractants. The most hydrophobic humic extract, isolated from soil by an acetone– HCl solution, adsorbed more and desorbed less atrazine than the more hydrophilic humic materials extracted by either an alkaline or an alkaline–pyrophosphate solution. Since the acetone–HCl extract represents only a small fraction of soil HS (Piccolo, Campanella et al., 1990), the study of Piccolo et al. (1998) inferred that humic microdomains of different polarity are also present in soil, and their mutual distribution determines their reactivity. Direct evidence of the presence of hydrophobic domains in HS was further given by Kohl et al. (2000), who studied the sorptive uptake of hexaﬂuorobenzene by two peat samples with solid-sate 19F-NMR spectroscopy. They found that the sorption process was rapid and related to the soil lipid content, whereas removal of the lipids decreased signiﬁcantly the sorptive capacity of SOM. Hu et al. (2000) detected crystalline domains composed of poly(methylene) chains by solid-state NMR and wide-angle X-ray scattering (WAXS) in several samples of soil organic matter, humins, and humic acids from soil and coal. Their results indicated that no more than 25 CH2 units were present in the hydrophobic crystalline domains. A comparable amount of noncrystalline and more isotropically mobile poly(methylene) chains were also found, and together with the crystalline materials larger aggregates were formed. In accordance with the supramolecular association model described here, Hu et al. (2000) concluded that the crystallites are expected to be resistant to environmental attack and thus inert in the soil and

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

107

have long residence times, while amorphous regions may play a role in the sorption of nonpolar molecules in soil. Disruption of humic supramolecular associations due to formation of hydrogen bondings stronger than the hydrophobic forces stabilizing the original conformation was shown by Miano et al. (1992). They used ﬂuorescence and infrared spectroscopy to investigate the effects of the addition of a glyphosate (N-(phosphonomethyl)glycine) herbicide to a puriﬁed humic solution to bring the pH from 9 to progressively lower values. The excitation spectra revealed an increasing quenching effect on the ﬂuorescence bands at high wavelengths with increasing glyphosate content, while synchronous spectra showed a decrease in the main peak intensity with herbicide addition. These results were interpreted with a disaggregation of the humic supramolecular association by formation of multiple hydrogen bonds between glyphosate and the small humic molecules, implying that a decrease in electron delocalization was responsible for the bands at high wavelengths given by the apparent high-molecular-size HS. The molecular and conformational structure of HS were conﬁrmed to determine adsorption of glyphosate on different humic materials (Piccolo, Celano et al., 1996). The high content of aliphatic components and large and ﬂexible molecular size increased adsorption of the herbicide on HS, possibly because of the involvement of dispersive hydrophobic interactions together with hydrogen bonds. Conte et al. (1997) showed that formation of tightly bound hydrophobic domains in solutions of HS could alter the quantitative evaluation of humic carbon distribution in 13C-NMR spectra recorded in the liquid state. A comparison between liquid-state NMR spectra and solid-state CPMAS–NMR spectra of several HS showed that the amount of alkyl groups measured by the former technique was invariably lower than for the latter method. The authors attributed this result to the separated phase that the humic hydrophobic components create to reduce the total solvation energy in solution and to the consequent limited spin-lattice relaxation time of hydrophobic carbons. While liquid-state spectra may therefore show lower signals in the alkyl-C region, this does not occur in the solidstate mode where signal recording occurs by polarization transfer from protons near the carbons to be measured. In another approach, Kenworthy and Hayes (1997) used the ﬂuorescence quenching of pyrene by bromide to investigate the nature of humic associations. They found that pyrene in HS solution was protected from bromide quenching. This protection was lost, however, when acetic acid, followed by base, was added to the medium. The authors considered that hydrophobic associations of the humic molecules protected the pyrene from the bromide. Treatment of the humic solution with acetic acid, as proposed by Piccolo et al. (1996a,b), removed that protection because of disaggregation of the loose humic superstructure. This suggested that the HS in solution were associations of low-molecular-weight masses held together by hydrophobic bonding and in which pyrene was enveloped.

108

ALESSANDRO PICCOLO

Supramolecular associations of HS were found by Haider et al. (2000) to be disrupted in smaller components by the action of derivatizing reagents (trimethylsilyl compounds) to silylate acidic oxygenated functions. These reagents cannot break the ether and ester linkages that are believed by the supporters of the polymeric model to hold humus constituents together. The authors showed that substitution of trimethylsilyl groups for active hydrogens in the functional groups of soil organic matter caused a disaggregation of the humic pseudomacromolecular structures into smaller entities that readily dissolve in organic solvents and elute in lowmolecular-size ranges by HPSEC. Wanner et al. (2000) explained with a disaggregation of humic supramolecular associations the shift in gel permeation to larger elution volumes of the radioactivity of a 14C-labeled dithianon fungicide bound to HS extracted from soil after 64 days of incubation together with straw. The same material extracted from soil without straw incubation indicated that the radioactivity was mainly in the large-molecularsize fraction of HS. The authors reasoned that organic acids or other amphiphilic compounds produced during microbial degradation of maize would have disrupted the association of humic molecules as proposed by Piccolo et al. (1996a,b). The use of a NMR technique such as diffusion ordered spectroscopy (DOSY) with enhanced sensitivity provided by keeping the probe circuitry at 22◦ K allowed Simpson et al. (2000) to support the supramolecular model proposed by Piccolo and co-workers. In their preliminary study, the authors noticed that the Nuclear Overhauser Effect (NOE) of a humic material was explained more by an association of small molecules rather than by a macromolecular polymer. Buurman et al. (2002) studied the thermal stability of soil humic extracts saturated with H, Na, Ca, or Al before and after wetting with organic solvents such as methanol, formic acid, and acetic acid. Table VIII shows that while thermal characteristics of H–humates did not change upon addition of the organic solvents, thermal decomposition (oxidation) of Na–humates shifted to much higher temperatures (750–850◦ C) than that for the control. The effect on Ca–humates was substantially less dramatic, whereas some molecular rearrangement but hardly any alteration in thermal stability was observed when the organic solvents were added to Al–humates (Table VIII). Buurman et al. (2002) explained these results on the basis of the different forces that hold small humic molecules together in the various samples. In fact, humic components are strongly bound to each other by hydrogen bonding in H–humates, whereas divalent and trivalent cations bind humic molecules together by electrostatic bridges in Ca– and Al–humates. These bonds were not overcome by the simple addition of organic solvents, and their thermal stability remained generally unchanged (Table VIII). Conversely, associations of humic molecules in Na–humates are stabilized only by nonspeciﬁc hydrophobic interactions. Hence, the signiﬁcant increase in thermal stability occurred by wetting the relatively ﬂexible Na-humates with organic solvents, which are slightly less polar than water, was explained by a rearrangement of largely hydrophobic

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

109

Table VIII Temperatures (T◦ C) of Exotherms and Relative Weight Losses (%) in Themogravimetric Analyses (TGA) of a Soil Humic Extract and Their Saltsa Temperature ranges 250–350◦ C Sampleb

MLc

T

WLe

9.8

319

42.7

350–450◦ C T

WL

450–550◦ C T

WL

550–700◦ C T

WL

>700◦ C T

WL

Rd

H H-sat

510−

Na-sat Control MeOH HFor HAc

530 8.3 13.5 7.5 25.4

285 282 288 282

43.8 35.3 43.1 30.2

428

Ca-sat Control MeOH HAc

16.6 40.0 25.1

309

25.4

322

37.5

365 362 433

Al-sat Control MeOH HFor HAc

10.1 64.5 84.6 19.6

295 288

21.9 24.4

35.4−

3.0

50.8

51.2

481

19.5

37.4 57.3 30.8

484 477 510

13.8 23.7 8.1

381 344

39.9 48.2

338

15.3

467 450 492 461

47.3 29.3 46.2 49.7

807 822 756 663

9.5

623

10.2

53.9 40.4 31.8

3.0 11.7 7.7 11.5 13.9 12.5 11.2 7.5 14.0 14.7 7.3

a

Buurman et al. (2002). Control; humic sample without any addition. MeOH, methanol; HFor, formic acid; HAc, acetic acid. c ML, moisture loss (%) (20–105◦ C) refers to the total weight of sample. d R; Residual weight at 1000◦ C, percentage of dry sample (105◦ C). Temperature readings are from TGA and may differ slightly from DTA curves. e WL, weight loss belonging to the respective exotherm, percentage of dry sample (105◦ C). b

humic components leading to an increase of the association energy. The authors also observed that the intensity and the reversibility of the thermal stabilization suggested that the rearrangement of the humic association must have taken place among small molecules rather than in macromolecules.

VII. TURNING LOOSE HUMIC SUPERSTRUCTURES INTO STABLE POLYMERS Understanding humus as a supramolecular association of small molecules means overcoming the limitations imposed by the paradigmatic polymeric model. If HS

110

ALESSANDRO PICCOLO

are seen as weakly bound supramolecular associations, their unstable conformation could be then stabilized in real polymeric structures. This could be achieved by increasing the number of intermolecular covalent bonds via an oxidative-coupling reaction catalyzed by oxidative enzymes as the phenoloxidases. This class of enzymes has been shown to promote, through a free-radical mechanism, oligo- and polymerization of phenols and anilines, and hence believed to contribute to soil detoxiﬁcation from related organic contaminants (Kim et al., 1997). However, no evidence of the catalytic action of these enzymes directly on humic molecules has ever been produced. This is because, while covalent binding of contaminants to HS was observed, there was no reason to evaluate a size increase in a humic matter that was “assumed” to be already polymeric. Piccolo, Cozzolino et al. (2000) attempted to turn a loosely bound humic superstructure into a covalently linked polymer by treating with horse-radish peroxidase (HRP) and hydrogen peroxide (oxidant), a humic material dissolved in 0.1 M phosphate buffer at pH 7. They used a HPSEC Biosep S2000 column (Phenomenex) to evaluate the changes in molecular-size distribution brought about by the oxidative reaction with HRP catalysis. Moreover, addition of acetic acid to the reacted humic mixture to pH 4 before HPSEC injection was used to assess the stability of humic conformation following the polymerization reaction. Figure 16 shows the HPSEC chromatograms obtained with control humic solution before (Fig. 16A) and after (Fig. 16B) acetic acid addition. The control solution showed only a slight absorption at the void volume (V0) characteristic of high-molecular-size fractions. Treatment of this solution with acetic acid to pH 4 decreased the molecular-size distribution of the humic material as also observed in other research described previously. Similar behavior was shown by the humic solution when treated with either H2O2 (Fig. 16, C) or HRP (Fig. 16, E) alone and after acetic acid addition (Fig. 16, D and F, respectively). The lower intensity of peak absorption (hypochromism) in the latter solutions suggests an inﬂuence of both oxidant and enzyme on the relative distance (and dipole orientation) among chromophores. A degree of disaggregation of the humic supramolecular structure into smaller associations by the presence of peroxidase alone was indicated in the acetic acid-treated sample (Fig.16, F) by the concomitant reduction of intensity in the diffused peak and its enhancement in the peak eluted after the solvent hump (around 25 mL). The chromatogram of the humic solution subjected to the oxidative-coupling reaction with both H2O2 and HRP (Fig. 16, G) was distinctly different from the control chromatograms (Fig. 16, A, C, E). The peak at V0 was increased; a new peak appeared at around 14.7 mL; and the large diffused peak was not only more intense than that in control solutions but also shifted to lower elution volumes (21.5 versus about 22.5 mL). These changes indicated a signiﬁcant increase in the molecular size of humic material with oxidation catalyzed by HRP.

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

111

Figure 16 HPSEC chromatograms of a control humic solution in phosphate buffer at pH 7 (A), of the same solution as in A but added with H2O2 (C), of the same solution as in A but added with peroxidase (E), of the same solution as in A but added with both H2O2 and peroxidase (G) and of the same solutions but added with acetic acid to pH 4 before injection (B, D, F, H, respectively) (Piccolo et al., 2000d).

Treatment of the humic solution with acetic acid conﬁrmed that the size increase was due to a true polymerization of humic molecules via formation of carbon–oxygen or carbon–carbon bonds rather than to a different supramolecular association stabilized by weak forces. In fact, unlike control solutions, the peak at V0 not only maintained its intensity but also was even increased in the chromatogram of sample treated with acetic acid (Fig. 16, H). The same behavior was partially shown by the new peak that appeared at 14.7 mL after the polymerization

112

ALESSANDRO PICCOLO

reaction. This suggested that the high-molecular-size material excluded at these elution volumes was stabilized by stronger forces than in control samples, and their macromolecular arrangement could not be disrupted by the addition of acetic acid. However, the reduced intensity of the large diffused peak (about 21.5 mL) after acetic acid treatment suggests a hypochromic effect due to chromophores which, being not yet covalently bound in polymeric structures, were separated from their weakly bound associations and eluted at larger elution volumes. Piccolo, Cozzolino et al. (2000) studied, by infrared spectroscopy, humic samples which underwent an oxidative catalyzed reaction to collect future evidence of the formation of covalent bonds. Diffuse reﬂectance infrared Fourier transform) (DRIFT) spectra of HRP alone, a control HS, and HS oxidized by HRP catalysis are shown in Fig. 17, A, B, and C, respectively. In comparison to the control, the DRIFT spectrum of the humic material subjected to oxidative coupling showed a substantial change in the 1500–900 cm−1 frequency interval with the appearance of three main bands at 1247, 1097, and 947 cm−1 and a decrease in the 1400and 1227-cm−1 bands. The absorptions shown at 1247 and 1097 cm−1 was reasonably assigned to bond deformation of aryl and alkyl ethers (Bellamy, 1975), respectively, which were formed during free-radical coupling reactions catalyzed by HRP and, hence, conﬁrm the interpretation of HPSEC measurements. The HPSEC and DRIFT results of Piccolo, Cozzolino et al. (2000) suggest that the small heterogeneous molecules present in HS, as in weakly associated superstructures, can be covalently bound into true oligo- or polymers by an oxidative coupling reaction catalyzed by a peroxidase enzyme. The extent of covalent polymerization should be a function of the amount of humic molecules, mainly phenolic or benzencarboxylic acids derived from lignin and microbial biosynthesis which may undergo oxidative-coupling reactions. However, it should be imagined that other classes of compounds may become conﬁned into the macromolecular conformations of polymerized humus. Cozzolino and Piccolo (2002b) extended the polymerization catalyzed by HRP to other HS and studied the effect of solution pH (4.7 and 7) and composition of humic associations. By HPSEC experiments they conﬁrmed that an increase in weight-average molecular weight (Mw) occurred invariably for all humic substances with oxidative polymerization. Table IX shows that Mw values of the polymerized samples increased for three different HS with respect to the control. Moreover, comparison of chromatograms (not shown) and Mw values (Table IX) obtained by treating humic solutions with acetic acid to pH 3.5 before HPSEC injection, conﬁrmed that the increase in molecular size by HRP catalysis was stable and due to formation of covalent bonds among reacting humic molecules. However, covalent polymerization of humic molecules was found to proceed to a further extent at pH 7 than at pH 4.7 (Table IX), despite the fact that HRP is most active at the latter pH. The difference in reactivity was attributed to the large mobility of reacting molecules in the hydrated and relatively smaller humic

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

113

Figure 17 DRIFT spectra of horseradish peroxidase (A), humic acid (B), and humic acid subjected to oxidation catalyzed by horseradish peroxidase (C) (Piccolo, Cozzolino et al., 2001).

114

ALESSANDRO PICCOLO Table IX Weight-Average Molecular Weight (Mw) Values and Percentage of Variation with Respect to Control Solutions () of Humic Sample Treated with Acetic Acid (from pH 7 to 4.5 and from pH 4.7 to 3.5) before and after (Mixture) Oxidative Polymerization by Enzymatic Catalysisa Sample

Control

Control + AcOH

HA-A Mw

39705 ± 1538

45047 ± 1634 +13.4

pH 4.7 42582 ± 1620 +7.2

45679 ± 1672 +15.0

25803 ± 296

42772 ± 2027 +65.7

pH 7.0 33154 ± 492 +28.4

47771 ± 1763 +85.1

9535 ± 574

11160 ± 576 +17.0

pH 4.7 13233 ± 698 +38.7

13853 ± 678 +45.28

8461 ± 570

11256 ± 556 +33.0

pH 7.0 12827 ± 676 +51.6

11554 ± 486 +36.5

13498 ± 685

13782 ± 586 +2.1

pH 4.7 21159 ± 1044 +56.7

23574 ± 918 +74.6

14051 ± 291

13284 ± 327 −5.4

pH 7.0 15788 ± 168 +12.4

19009 ± 178 +35.3

Mw HA-B Mw Mw HA-C Mw Mw a

Mixture

Mixture + AcOH

Cozzolino and Piccolo (2002b).

associations stabilized only by weak dispersive (hydrophobic) forces at pH 7. Reactive humic molecules are more mobile at neutral pH and the polymerization via a free-radical mechanism is more efﬁcient. Conversely, intermolecular hydrogen bonds formed at pH 4.7 confer a larger conformational size and rigidity to humic associations, while the mobility as well as the reactivity of small molecules are thereby reduced. Cozzolino and Piccolo (2002b) also showed that the extent of polymerization depends on the molecular composition of the humic association undergoing oxidative reactions. They noticed that polymerization was somewhat inhibited when HS were richer in alkyl carbons and poor in carboxyl carbons as assessed by 13 C-CPMAS–NMR spectroscopy. The HA-A sample of Table IX had the largest alkyl-C and lowest carboxyl-C content among the tested HS. Its degree of polymerization was low at pH 4.7 because reactive humic molecules remained conﬁned in strong hydrophobic associations of apparently large molecular size and could hardly become available to the oxidative enzyme. This limitation was partly solved at pH 7 when negative charges arising form dissociation of existing carboxyl groups

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

115

disrupted the hydrophobic associations from which humic constituents were liberated to move and come concomitantly in contact with catalyst and oxidant. In accordance with this mechanism, HS were also found to possess a greater reactivity towards the transformation of the herbicide 2,4-D into 2,4-dichlorophenol when dissolved at pH 7 rather than at pH 4.7 (Piccolo et al., 2001a). The larger reactivity towards polymerization shown by the other two HS even at pH 4.7 (Table IX) was attributed to their generally high content of carboxyl groups and low amount of alkyl-C that prevented the formation of tight hydrophobic domains and kept the humic molecules in relative high mobility in solution at pH 4.7. However, their larger reactivity may be also ascribed to their high content of aromatic groups which are potentially highly reactive in free-radical reactions such as those catalyzed by HRP. Involvement of these groups in covalent bond formation had been previously shown using DRIFT spectroscopy (the aryl ether band in Fig. 17). Synthetic complexes of OH–Al–humate–montmorillonite were made using humic acids from an oxidized coal and a lignite (Violante et al., 1999) to model organomineral complexes of soils. In a separate experiment, these synthetic clay– humic complexes were subjected to the oxidative reaction with HRP as a catalyst and H2O2 as an oxidant. No changes in the organic carbon content were observed with the applied oxidative conditions. Extraction of the complexes with alkaline– pyrophosphate solutions allowed for the determination of the amount of HS available to solubilization before and after the treatment with HRP. Figure 18 shows that the yields of extraction for both humic–clay complexes decreased signiﬁcantly after the oxidative-coupling reaction, ranging from about 42 to 32% and from 40 to 29% for the complexes made of humic acid from oxidized coal and lignite, respectively. These results indicate that polymerization of humic molecules occurred also in the solid phase of the clay–humic complexes, and the increase in molecular size of the humic materials was the most probable cause for the reduction in extraction yields. It also seems possible to induce the polymerization of HS in natural soil samples in order to control or change the properties of native soil organic matter. The evidence shown here that humic supramolecular associations can be turned into more stable covalently linked conformation of truly larger molecular size can be interpreted as additional evidence that HS should not be considered macromolecular polymers as they have been viewed for so long.

VIII. ROLE OF HYDROPHOBIC HUMIC SUPERSTRUCTURES IN SOIL A model of humus as a supramolecular association of small molecules, originated from extended microbial degradation of different plant and animal

116

ALESSANDRO PICCOLO

Figure 18 Yields (%) of HS extracted with an alkaline–pyrophosphate solution from synthetic OH–AL–humate–montmorillonite complexes formed by using humic acids from oxidized coal and lignite before (COX-M; LIG-M) and after polymerization reaction catalyzed by (HRP) horseradish peroxidase (COX-M+HRP; LIG-M+HRP).

biomolecules and assembled together by mainly hydrophobic forces strengthened by the hydrophobic effect, may well have implications on how we regard the phenomena of accumulation and decomposition of soil organic matter. It has been increasingly proved in recent years that simple, mainly alkyl, organic compounds deriving from both plant residue decomposition and microbial resynthesis are progressively incorporated into the most stable SOM fractions (Almendros et al., 1996, 1998; Jambu et al., 1991; Lichtfouse, 1998). It has also been ascertained that the most recalcitrant humic fractions mainly contain aliphatic compounds (Augris et al., 1998; Kohl et al., 2000; Nierop et al., 1998; Nierop et al., 1999). Piccolo (1996) reviewed the literature on the stability of soil humus and presented several examples of increased humus adsorption and accumulation in soils through nonionic entropic mechanisms. He proposed that hydrophobic humic components in soil protect easily degradable compounds. He postulated that incorporation of polar molecules in associations of hydrophobic components may contribute to prevent an otherwise rapid microbial degradation of hydrophilic molecules and

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

117

enhance their persistence in soil. This hypothesis is in accordance with the model of humic superstructures described here, by which humic molecules self-assemble into hydrophobic or hydrophilic domains according to their reciprocal afﬁnity. It could be thus assumed that tightly bound humic associations containing mainly resistant alkyl remains of vegetative tissues may incorporate, by a random selforganizing process, also a few hydrophilic molecules or associated clusters of them. Evidence that hydrophilic domains are part of humic supramolecular associations in solutions were reported previously. However, a mechanism for incorporation of labile hydrophilic compounds such as carbohydrates into hydrophobic humic domains was also suggested to occur in soil by Spaccini, Zena et al. (2000). They found that, for three European soils in a north–south gradient climate, carbohydrate content increased with decreasing size of soil aggregates and was related to the humic content in aggregates and the climate regime prevailing for each soil. A similar ﬁnding was reported by Spaccini, Zena et al. (2001) in several soils of two contrasting tropical ecosystems (Ethiopian highlands and Nigerian lowlands). In both forested and cultivated soils, carbohydrates content was found to be larger in smaller-size fractions of both soil aggregates and particles. The content of HS was highly correlated to that of carbohydrates, thereby implying that in both types of land use the microbially labile carbohydrates were protected from mineralization in humic associations and in the relative organomineral complexes. The persistence of HS in soil should be no longer attributed to a presumed large macromolecular size that is formed by a process of catalyzed polymerization, highly unprobable under natural conditions such as the multiphase soil matrix (see previous discussion), but rather to a mechanism of hydrophobic protection. Humic molecules associate in soil solution or on mineral surfaces by weak forces and progressively build up in apparently large-size molecular clusters which are extruded from water and its biological activity, and thus reside for longer time in soil. Until a new event (plowing, deforestation, etc.) alters the established equilibrium, supramolecular associations of humic molecules are stable in soil and may persist indeﬁnitely. The hypothesis that incorporation of degradation products of fresh organic matter in native humic materials represents a basic mechanism of soil organic matter accumulation as well as the reason for its long-term stabilization have been proposed also by other authors (Baldock et al., 1989; Lichtfouse et al., 1998; Nierop and Buurman, 1999). The concept of HS as spontaneous associations of different molecules bound together by multiple weak dispersive bonds offers several different important implications: (1) organic matter in soils accumulates by hydrophobic protection; (2) given enough interaction with the soil solution any humic molecular superstructure (regardless of the classiﬁcation into humic and fulvic acids and humin) can exchange with hydrophilic molecules freshly added via degradation of biological

118

ALESSANDRO PICCOLO

tissues; (3) stabilization of humic superstructures by hydrophobic forces should result in a consequent stabilization of organo-mineral complexes and overall soil structure.

A. SEQUESTRATION OF ORGANIC CARBON IN SOIL BY HYDROPHOBIC PROTECTION Accumulation of SOM by hydrophobic protection was shown by Spaccini, Piccolo et al. (2000). They characterized the labile and stable pools of organic matter in three European soils after a 1-year incubation with fresh organic matter such as maize straw either deposited on soil (mulched) or incorporated (mixed) in soil. Quantitative analyses of carbohydrates released by acidic hydrolysis from particle-size fractions of incubated soils and relative δ 13C measurements in hydrolysates indicated that, rather than ﬁxing carbohydrates derived from added maize, incubation either with or without maize enhanced decomposition of this native labile component which had been previously stabilized in the SOM. Conversely, the authors found that the stable organic matter pool represented by different humic fractions could incorporate the organic C derived from maize straw decomposition especially when this was thoroughly mixed with samples. Incorporation of organic matter from maize into humic substances was generally shown by 13 C-CPMAS–NMR spectra of humic fractions extracted by an alkaline solution, suggesting that incorporated material in HS was mainly aliphatic compounds, such as polysaccharides and peptides of plant and microbial origin. Values of C isotopic abundance (Table X) together with structural information of humic extracts obtained by NMR spectra showed that both hydrophilic and hydrophobic components from maize straw were incorporated into the humic pool in the soil. Table X shows that, in comparison to HS extracted with alkaline solutions (HA1 and FA1), increased carbon from maize was found in humic and fulvic fractions isolated again with alkaline extractants (HA2 and FA2) after removal from soils of a highly hydrophobic organic fraction (HE) solubilized in an acetone–HCl solution (Piccolo, Campanella et al., 1990). These results indicated that the mutual interactions of different classes of compounds inﬂuence solubility and chemical reactivity of humic matter in soil. In particular, the hydrophobic material solubilized in the HE fraction appeared to have covered and/or incorporated the maize-derived hydrophilic components and thus favored their persistence by precluding contact with water and microorganisms. The δ 13C values of Table X indicated that while the hydrophobic fraction (HE) behaved as a protective sink for carbon released by maize straw during incubation, a larger incorporation of carbon from maize was generally shown by the FA fractions especially after removal of the protecting hydrophobic layer of humus by acetone. This conﬁrmed previous reports (Wander and Traina, 1996; Zalba and

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

119

Table X Isotopic Abundance (δ 13 C, %o) and Maize-Derived OC (%) in Bulk Soil and Humic Fractions of Control and Maize-Treated Samples before (t0 ) and after (t1 ) Incubation (Standard Deviation was Less than ± 0.1%o for all Values)a Control (t0) 13

Control (t1) 13

Mulched (t1) 13

Mixed (t1) 13

Sample

δ C ( ‰)

δ C (‰)

δ C (‰)

OCmaize(%)

δ C (‰)

Danish soil Bulk soil HA1 FA1 HE HA2 FA2

−26.5 −27.6 −27.2 −29.4 −27.7 −27.0

−26.6 −27.8 −26.6 −29.4 −27.5 −26.2

−24.6 −27.3 −26.5 −28.1 −26.0 −25.2

NSb NS 7.5 10.0 7.2

−24.5 −26.2 −25.2 −27.0 −25.7 −23.5

9.7 9.5 13.9 12.0 19.3

German soil Bulk soil HA1 FA1 HE HA2 FA2

−26.0 −26.9 −26.6 −28.4 −27.4 −26.8

−26.1 −27.5 −26.1 −28.4 −27.3 −25.5

−23.8 −27.3 −26.1 −28.2 −27.2 −24.5

NS NS NS NS 7.4

−23.1 −25.6 −24.5 −26.3 −25.6 −24.0

12.3 11.3 13.2 10.7 11.4

Italian soil Bulk soil HA1 FA1 HE HA2 FA2

−24.2 −25.3 −24.4 −26.8 −25.2 −24.4

−24.2 −25.3 −23.6 −26.1 −25.6 −24.0

−22.0 −25.2 −23.8 −25.9 −25.1 −22.5

NS NS NS NS 12.7

−21.7 −23.5 −22.2 −24.3 −25.4 −21.9

13.5 11.8 12.7 NS 17.5

a b

OCmaize(%)

Spaccini, Piccolo et al. (2000). NS, not signiﬁcant.

Quiroga, 1999) which indicated fulvic acids to be the humus fraction most sensitive to soil management. However, in addition to fulvic acids, also humic acids and the humin-like HE fraction incorporated carbon from maize (Table X), thereby suggesting that all humic fractions had short-term interactions with degradation products of vegetative tissues. The ﬁndings of Spaccini, Piccolo et al. (2000) are in line with the dynamic model of small humic molecules of different origin driven by weak forces to either form supramolecular associations in the soil solution or accumulate in layers on soil surfaces. The same results would be hardly explainable by the polymeric model, unless one assumes that an extensive polymerization of maize-derived molecules had spontaneously occurred in the multiphase soil matrix during the relatively short 1-year period.

120

ALESSANDRO PICCOLO

The sequestration of organic carbon (OC) in soil by hydrophobic protection in humic material was proved by Piccolo, Spaccini et al. (1999). They synthesized a 13 C-labeled 2-decanol as a model of an easily degradable molecule in soil. They partitioned the labeled molecule into solutions of two humic acids, one from compost (HA–C) and one from lignite (HA–L), of different degrees of hydrophobicity. The two labeled humic solutions and one solution containing only the labeled 2-decanol (soil+13C) were added to a soil and incubated at ﬁeld capacity for 3 months. The treated samples and a control soil were sampled periodically and the 13 C content was measured by high-resolution mass-spectrometry. Figure 19 shows the variation in δ 13C values (‰) for control and treated samples with incubation time and the signiﬁcant differences in OC sequestration among treatments. The biolabile 13C-labeled 2-decanol was protected from mineralization when incorporated into the hydrophobic domains of the HS. The highly hydrophobic and more aromatic humic acid from lignite was more effective than the one from compost in sequestering the carbon from 2-decanol. After incubation, the residual 13C-labeled OC recovered in bulk soil was equal to 28, 45, and 58% of the original content for samples containing the labeled alcohol alone or with HA from compost and lignite, respectively. These percentage values

Figure 19 Variation in δ 13C values with incubation time (weeks) in relation to soil treatments with C-labeled material and to control soil sample. Vertical lines indicate SD (n = 3) (Piccolo, Spaccini et al., 1999). 13

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

121

calculated as kilograms per hectare of soil OC indicated that the protection exerted by the most hydrophobic HA from lignite signiﬁcantly increased the 13C-OC content by 25 and 40 kg ha−1 compared to HA from compost and the alcohol alone. The same experiment allowed Spaccini et al. (2001) to also follow the 13C-OC distribution in the particle-size fractions of the treated samples. Figure 20 reports the residual 13C-OC among soil particle sizes and indicates that the hydrophobic protection was most effective in the silt- and clay-sized fractions. This result conﬁrms the importance of associations between ﬁne textural fractions and microbially recalcitrant OM and suggests that SOM accumulation due to hydrophobic protection preferentially occurs within organomineral association of ﬁner soil particles. Nevertheless, hydrophobic sequestration of carbon in soil may also take place within larger size fractions, provided that humiﬁed matter of large hydrophobic character is applied. In fact, the highly hydrophobic HA from lignite was able to reduce OC decomposition, with respect to treatments with HA from compost and 13 C-2-decanol alone, even in the coarser fractions which are commonly associated with rapid cycling of SOM pools (Angers et al., 1997; Gregorich et al., 1989). Further evidence of the hydrophobic mechanism for 13C-OC incorporation into soil humic material was found by Spaccini et al. (2001) when they measured the 13 C-OC content in HS extracted by alkaline solutions from different treatments at the end of incubation. Table XI shows that in the course of incubation the labeled carbon strictly interacted with HA and FA, which, together, retained from 27.7 to 39.7% of total 13C-OC present in the whole soil at the end of incubation. Interestingly, despite the ﬁnding that NMR spectroscopy showed that the original 13C-methyl group of 2-decanol had been oxidized, presumably by microbial activity during incubation, to a highly hydrophilic carboxyl group, the resulting labeled structure was still tightly held by humic matter. This was taken as an indication that hydrophobic protection must have played a role in sequestering the hydrophilic residual 13C-OC in the soil humic matter. In fact, 13C-OC was found to be incorporated to a larger extent in the most hydrophobic humic fraction, the HA, which retained from three to ﬁve times more 13C-OC than the more hydrophilic FA (Table XI). The importance of these ﬁndings may be signiﬁcant for controlling the carbon cycling between the soil and the atmosphere at a global scale. Innovative soil management practices aimed at increasing the hydrophobicity of SOM, by amendments with organic matter of advanced humiﬁcation such as mature compost or humic acids from geological sources, may have an impact in substantially reducing CO2 emissions from soils. Combined with the present and future efforts in limiting CO2 emissions from fossil fuels, the hydrophobic sequestration of carbon in soils may contribute to reduce the rate of CO2 increase in the atmosphere. If the values reported by Piccolo, Spaccini et al. (1999) and Spaccini et al. (2001) for the decrease of SOC losses obtained under the conditions of their experiment are multiplied for the estimated world cultivated areas which are around

122

ALESSANDRO PICCOLO

Figure 20 Variation in comparison to time 0 of 13C-SOM content in soil particle-size fractions according to treatments (13C-2-dec. = treatment with only 13C-labeled 2-decanol; 13C-HAC = treatment of HA from compost previously added with 13C-labeled 2-decanol; 13C-HAL = treatment of HA from lignite previously added with 13C-labeled 2-decanol). Bars in graph indicate standard deviation (n = 3) (Spaccini et al., 2001).

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

123

Table XI Content of 13C-Labeled Organic Carbon (13C-OC2-dec) in Soil Humic (HA) and Fulvic (FA) Acids Extracted from Soils Treated with 13 C-Labeled 2-Decanol Alone (13C-2-dec) or with Humic Acid from Compost (13C-HAC) or from Lignite (13C-HAL), at the Start (t0) and the End (t2) of Incubationa HA

Treatments Control 13

C-2-dec

13

C-HAC

13

C-HAL

t

b

t0 t2 t0 t2 t0 t2 t0 t2

δ 13C (‰)c −26.21 −25.95 −25.72 −23.54 −22.02 −18.70 −25.64 −19.56

OC2-dec (mg)

0.12 0.58 0.92 1.60 0.14 1.60

FA OC2-dec (%tot)

1.2 20.0 8.8 34.0 1.3 25.8

δ 13C (‰)c −25.80 −25.65 −25.04 −23.42 −23.17 −22.75 −23.74 −22.45

OC2-dec (mg)

OC2-dec (%tot)

0.08 0.22 0.25 0.27 0.22 0.34

0.8 7.6 2.4 5.7 2.0 5.5

a

Spaccini et al. (2000e). Incubation time (weeks): t0 = 0, t2 = 12. c LSD among soil treatments and incubation time = 0. 1‰ (P = 0.05; n = 3). b

16–17 × 108 ha (Paustian et al., 1997; Schlesinger, 1997), the calculation relative to 13C-OC decomposition gives a hypothetical decrease in carbon losses of about 0.7 × 1014 g of soil OC. This value represents a considerable reduction of C losses from cultivated soils as compared to the global estimate of 8 × 1014 g C year−1 (Paustian et al., 1997; Schlesinger, 1997).

B. HYDROPHOBIC HUMIC ASSOCIATIONS IN THE STABILIZATION OF SOIL STRUCTURE A good structure is important for sustaining long-term crop production on agricultural soils because it inﬂuences water status, workability, resistance to erosion, nutrient availability, and crop growth and development. One of the measures of good structure is the stability of soil aggregates in water, and this is inﬂuenced mostly by both the quality and the quantity of OM in the soil. An increasing number of studies have pointed out the importance of nonionic bonds or entropic factors in the interactions between clay and humic components of soils. On the other hand, the inﬂuence of the molecular size of HS and their stereochemical ﬂexibility was shown in several instances. These ﬁndings, reviewed by Piccolo (1996), substantiated the understanding that soil structure needs the

124

ALESSANDRO PICCOLO

essential long-term stabilization in both macro- and micro-aggregates by humic molecules rather than by transient polysaccharides. Moreover, they suggested that hydrophobic forces holding together the apparently large-size humic superstructures may well be also responsible for the ultimate stabilization of soil structure. Piccolo and Mbaguw (1999) investigated the simultaneous aggregate-stabilizing effects of different SOM components. They studied the effect of model compounds such as polysaccharide gum (G) and stearic (S) acid (alone or in combination with a HA from lignite) on aggregation stability (AS) of an arable soil before and after removal of its native OM and during incubation time. They found that removal of OM reduced AS of unmodiﬁed soil by about 40 and 20% after soil incubation for 7 and 40 days, respectively. With reference to the amended soils (Table XII), G increased AS by 382, 264, and 22% at 7, 15, and 40 days of incubation in soil(−OM) (where OM was removed), whereas pretreatment with HA reduced the effectiveness of G by 20, 2, and 21%, respectively, for the same incubation period. The HA pretreatment before S addition enhanced AS at 7 days by 18% and at 15 days by 69%. When compared with S alone the S + HA treatment gave a relative increase in AS of 15% at 7 days and 36% at 15 days. It was only at 15 days that a signiﬁcant increase in AS of 21% was obtained with HA alone. Similar trends (Table XII) were obtained on soil(+OM) (where OM was retained) although the relative improvements in AS obtained from G and G + HA were lower than those in soil(−OM). Conversely highly signiﬁcant improvements

Table XII a

Percent Change in Aggregate Stability (WSA > 0.50 mm) Relative to the Unamended Soils as Influenced by Type of Amendment and Incubation Periodb Treatmentsd Soils

DAIc

G

S

HA

G+HA

S+HA

A(−OM)

7 15 40

382 264 22

3 33 16

2 21 1

362 262 1

18 69 10

B(+OM)

7 15 40

76 6 −14

10 60 24

3 1 −4

43 −5 −4

17 86 34

a Relative change (RC) = {[(WSAt/WSAc)−1] × 100}, where t = treated and c = control. b Piccolo and Mbagwu, (1999). c DAI = days after incubation. d Symbols are deﬁned in the text; negative values indicate decrease in stability relative to control.

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

125

in AS compared with controls were obtained from S and S + HA additions in this soil than in soil(−OM). Relative to S alone, S + HA improved AS by 7% at 7 days, 26% at 15 days, and 10% at 40 days. Piccolo and Mbagwu (1999) showed that soil–aggregate stability is improved and maintained with time more by hydrophobic than by hydrophilic components of organic matter. Their results are in line with the model of supramolecular associations of humic molecules by hydrophobic forces and suggest that soil aggregation is favored and maintained in large degree by interparticle hydrophobic associations. The implication is that long-lasting aggregate stability of soils can be achieved by promoting the hydrophobicity of native organic matter (for instance, through notillage practices) or by additions of hydrophobic humic material from hydrophobic materials such as organic wastes like compost or hydrophobic exogenous HS like those from lignite.

IX. FUTURE PERSPECTIVES IN RESEARCH AND TECHNOLOGY The clariﬁcation of the conformational structure of humic substances represents a major innovation in humus chemistry. The notion that humic substances are not macromolecular polymers as they have been described for so long but rather superstructures of only apparent large size and self-assembled by relatively small heterogeneous molecules held together by mainly hydrophobic dispersive (van der Waals, π –π , CH–π ) forces opens up new opportunities to enlarge the knowledge of both their detailed chemistry and their management in the soil and the environment. Chromatographic methods of separations such as HPSEC were found to produce reproducible and more homogenous fractions of the molecules constituting the humic superstructures. Awareness of the weak forces which cause the self-assembling of humic molecules has allowed one to devise methods based on interactions with chemical species such as amphiphilic organic acids, urea, mono- and polyvalent cations, which can disrupt the apparently large humic associations and obtain fractions which are chemically even simpler and more homogeneous. The combination of chromatographic methods with spectroscopic techniques such as NMR, IR, and ESR spectroscopy together with the different modern variations of mass spectrometry has increased enormously the potential to derive a complete picture of the secondary molecular structure of humic substances. The moment in which a sound molecular and conformational structure, based on chemical methods rather than on computer models, will be produced, is very near. From then on, our capacity to obtain a full molecular structure of any humic material or natural organic matter (NOM) from any environment and ecosystem will be limited by only the advances in analytical automation. Considering the rapidity with

126

ALESSANDRO PICCOLO

which certain biological research ﬁelds can grow, this limitation could be easily overcome, provided there is the necessary public and private interest to pursue knowledge in a ﬁeld that is vital for the well-being of our planet and the biological life living on it. The novel understanding of humic substances as supramolecular associations described here has great implications on soil and environmental management, which may only be partially imagined at this stage. One example is the mentioned possibility to turn the loose humic superstructures into real covalently linked polymers by a catalytic technology that can ensure polymerization of humic molecules in both water and soil environments. The fallout of such a technology is viable in order to improve our capacity to control soil organic matter management, reduce the risk of erosion, and limit the extent of soil desertiﬁcation, with the obvious improvement of soil productivity. Another example of the potential of the polymerization of soil humic molecules in situ is the possibility to control CO2 emission from agricultural soils, by sequestering the organic carbon in more stable polymeric humus. This technology should be easily coupled with the potential shown here of exploiting the hydrophobic capacity of humic substances alone to sequester organic carbon in soils and limit its microbial mineralization. Finally, there are a large number of observations indicating that humic molecules released from the large supramolecular associations can inﬂuence nutrient uptake by plants and increase crop yields signiﬁcantly. Combinations of reﬁned chemical analysis of humic molecules with physiological studies on their effect on plants may clarify the mechanism(s) by which soil humic substances increase crop yield and, possibly, revive the past interest in humus fertility. Use of humic molecules either native or exogenous together with inorganic fertilizers to maximize plant nutrient uptake and ﬁnal yields may have tremendous impact in both increasing the economic efﬁciency of fertilizers and protecting the environment from the pollution of excess fertilization.

ACKNOWLEDGMENTS Most of the research in this area was funded by the Ministry of University and Scientiﬁc Research and Technology (MURST) of Italy and the Commission (DGXII) of the European Union. Their support is gratefully acknowledged. I am also grateful to my former doctorate students, Drs. P. Conte and R. Spaccini, and to Ms. A. Cozzolino who is now completing her doctorate.

REFERENCES Aiken, G. R., and Gillam, A. H. (1989). Determination of molecular weights of humic substances by colligative property measurements. In “Humic Substances II. In Search of Structure” (M. H. B. Hayes, P. McCarthy, R. L. Malcolm, and Swift R. S., Eds.), pp. 516–543. Wiley, New York.

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

127

Aiken, G. R., McKnight, D. M., Wershaw, R. L., and MacCarthy, P. (1985). Introductory remarks. In “Humic Substances in Soil, Sediment, and Water” (G. R. Aiken, D. M. McKnight, R. L. Wershaw, and P. MacCarthy, Eds.), pp. 1–12. Wiley, New York. Almendros, G., Guadalix, M. E., Gonz´alez-Vila, F. J., and Martin, F. (1996). Preservation of aliphatic macromolecules in soil humins. Org. Geochem. 24, 65–659. Almendros, G., Guadalix, E., Gonz´alez-Vila, F. J., and Martin, F. (1998). Distribution of structural units in humic substances as revealed by multi-step selective degradation and 13C-NMR of successive residues. Soil Biol. Biochem. 30, 755–765. Amalﬁtano, C., Pignalosa, V., Auriemma, L., and Ramunni, A. (1992). The contribution of lignin to the composition of humic acids from a wheat-straw amended soil during 3 years of incubation in pots. J. Soil Sci. 43, 495–504. Angers, D. A., Recous, S., and Aita, C. (1997). Fate of carbon and nitrogen in water-stable aggregates during decomposition of 13C- and 15N-labelled wheat straw in situ. Eur. J. Soil Sci. 48, 295– 300. Aochi, Y. O., and Farmer, W. J. (1997). Role of microstructural properties in time-dependent sorption/ desoprtion behaviour of 1,2-dichloroethane on humic substances. Environ. Sci. Technol. 31, 2520– 2526. Augris, N., Balesdent, J., Mariotti, A., Derenne, S., and Largeau, C. (1998). Structure and origin of insoluble and non-hydrolizable, aliphatic organic matter in a forest soil. Org. Geochem. 28, 119–124. Baldock, J. A., Oades, J. M., Vassallo, A. M., and Wilson, M. A. (1989). Incorporation of uniformly labeled 13C-glucose carbon into the organic fraction of a soil. Carbon balance and CP/MAS 13 C NMR measurements. Austr. J. Soil Res., 27, 725–746. Ballarin, B., Seeber, R., Tonelli, D., and Zappoli, S. (1999). High performance size exclusion chromatography in the evaluation of molecular size distributions of humic acids fractionated by ultraﬁltration. Ann. Chim. (Rome) 89, 211–220. Barth, H. G. (1980). A practical approach to steric-exclusion chromatography of water soluble polymers, J. Chromatogr. Sci. 18, 409–429. Becher, G., Carlberg, G. E., Giessing, E. T., Hongslo, J. K., and Monarca, S. (1985). High-performance size exclusion chromatography of chlorinated natural humic water and mutagenicity studies using the microscale ﬂuctuation essay. Environ. Sci. Technol. 19, 422–426. Beckett, R., Jue, Z., and Giddins, J. C. (1987). Determination of molecular weight distributions of fulvic and humic acids using ﬂow ﬁeld fractionation. Environ. Sci. Technol. 21, 289–295. Bellamy, L. J. (1975). “The Infrared Spectra of Complex Molecules,” pp. 129–139. Chapman and Hall, London. Berden, M., and Berggren, D. (1990). Gel ﬁltration chromatography of humic substances in soil solutions using HPLC-determination of the molecular weight distribution. J. Soil Sci. 41, 161–172. Buurman, P., van Lagen, B., and Piccolo, A. (2002). Increase in thermal stability of soil humic substances as a result of self-association. Org. Geochem. (in press). Cameron, R. S., Swift, R. S., Thornton, B. K., and Posner, A. M. (1972). Calibration of gel permeation chromatography materials for use with humic acid. J. Soil Sci. 23, 343–349. Cameron, R. S., Thornton, B. K., Swift, R. S., and Posner, A. M. (1972). Molecular weight and shape of humic acid from sedimentation and diffusion measurements on fraction extracts. J. Soil Sci. 23, 394–408. Campbell, C. A., Paul, E. A., Rennie, D. A., and McCallum, K. J. (1967). Applicability of the carbon dating method of analysis to soil humus studies. Soil Sci. 104, 217–224. Cantor, C. R., and Schimmel, P. R. (1980). “Biophysical Chemistry. Part II: Techniques for the Study of Biological Structure and Function,” pp. 399–404. Freeman, NewYork. Carter, C. W., and Suffet, I. H. (1982). Binding of DDT to dissolved humic material. Environ. Sci. Technol. 16, 735–740.

128

ALESSANDRO PICCOLO

Ceccanti, B., Calcinai, M., Bonmati-Pont, M., Ciardi, C., and Tarsitano, R. (1989). Molecular size distribution of soil humic substances with ionic strength. Sci. Total Environ. 81/82, 471–479. Chicz, R. M., and Regnier, F. E. (1990). High-performance liquid chromatography: Effective protein puriﬁcation by various chromatographic modes. In “Guide to Protein Puriﬁcation” (M. P. Deutscher, Ed.), pp. 392–421. Methods of Enzymology, Vol. 182. Academic Press, San Diego, CA. Chien, Y.-Y., Kim, E.-G., and Bleam, W. F. (1997). Paramagnetic relaxation of atrazine solubilized by humic micellar solutions. Environ. Sci. Technol. 31, 3204–3208. Chin, Yu-P., Aiken, G. R., and O’Loughlin, E. (1994). Molecular weight, polydispersity, and spectroscopic properties of aquatic humic substances. Environ. Sci. Technol. 28, 1853–1858. Chin, Yu-P., and Gschwend, P. M. (1991). The abundance, distribution, and conﬁguration of porewater organic colloids in recent sediments. Geochim. Cosmochim. Acta 55, 1309–1317. Clapp, C. E., Emerson, W. W., and Olness, A. E. (1989). Sizes and shapes of humic substances by viscosity measurements. In “Humic Substances II. In Search of Structure” (M. H. B. Hayes, P. McCarthy, R. L. Malcolm, and Swift R. S., Eds.), pp. 497–514. Wiley, New York. Clapp, C. E., and Hayes, M. H. B. (1999). Sizes and shapes of humic substances. Soil Sci. 164, 777–788. Conte, P., Piccolo, A., van Lagen, B., Buurman, P., and de Jager, P. A. (1997). Quantitative differences in evaluating soil humic substances by liquid- and solid-state 13C-NMR spectroscopy. Geoderma 80, 339–352. Conte, P., and Piccolo, A. (1999a). High pressure size exclusion chromatography (HPSEC) of humic substances. Molecular sizes, analytical parameters, and columns performance. Chemosphere 38, 517–528. Conte, P., and Piccolo, A. (1999b). Conformational arrangement of dissolved humic substances. Inﬂuence of solution composition on association of humic molecules, Environ. Sci. Technol. 33, 1682–1690. Cornel, P. K., Summers, R. S., and Roberts, P. V. (1986). Diffusion of humic acid in dilute aqueous solution. J. Colloid Interface Sci. 110, 149–164. Cozzolino, A., and Piccolo, A. (2002a). Effects of bicarboxylic acids on the supramolecular associations of dissolved humic substances. Soil Biol. Biochem. (in press). Cozzolino, A., and Piccolo, A. (2002b). Polymerization of dissolved humic substances catalyzed by peroxidase. Effects of pH and humic composition. Org. Geochem. (in press). Cozzolino, A., Conte, P., and Piccolo, A. (2001). Conformational changes of soil humic substances induced by some hydroxy-, cheto-, and sulphonic acids. Soil Biol. Biochem. 33, 563–571. De Haan, H., Jones, R. I., and Salonen, K. (1987). Does ionic strength affect the conﬁguration of aquatic humic substances, as indicated by gel ﬁltration? Freshwater Biol. 17, 453–459. De Nobili, M., Gjessing, E., and Sequi, P. (1989). Sizes and shapes of humic substances by gel chromatography. In “Humic Substances II. In Search of Structure” (M. H. B. Hayes, P. McCarthy, R. L. Malcolm, and Swift R. S., Eds.), pp. 562–587. Wiley, New York. Determann, H., and Walter, I. (1968). Source of aromatic afﬁnity of Sephadex dextran gels. Nature (London) 219, 604–605. Dubach, P., and Metha, N. C. (1963). The chemistry of soil humic substances. Soils Fertil. 26, 293–300. Engebretson, R., and von Wandruszka, R. (1994). Micro-organization of dissolved humic acids. Environ. Sci. Technol. 28, 1934–1941. Engebretson, R., and von Wandruszka, R. (1997). The effect of molecular size on humic acid associations. Org. Geochem. 26, 759–767. Eswaran, H., Van den Berg, E., and Reich, P. (1993). Organic carbon in soils of the world. Soil Sci. Soc. Am. J. 57, 192–194. Flaig, W. (1958). Die Chemie der organischer Stoffe in Boden und deren physiologische Wirkung. Verhandl., II u. IV Komm. int. Bodenk. Ges. 2.

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

129

Flaig, W. (1988). Generation of model chemical precursors. In “Humic Substances and their Role in the Environment” (F. H. Frimmel and R. F. Christman, Eds.), pp. 75–92. Wiley, Chichester. Flaig, W., and Beutelspacher, H. (1968). Investigations of humic acids with the analytical ultracentrifuge. In “Isotopes and Radiation in Soil Organic Matter Studies,” pp. 23–30. IAEA, Vienna. Flaig, W., Beutelspacher, H., and Riets, E. (1975). Chemical composition and physical properties of humic substances. In “Soil Components” (J. E. Gieseking, Ed.), Vol. 1, pp. 1–219. Springer-Verlag, Berlin. Freifelder, D. (1982). “Physical Biochemistry,” pp. 500–503, 2nd ed. Freeman, New York. Ghosh, K., and Mukherjee, S. K. (1971). Hymatomelanic acids as polyelectrolytes: 1. Viscosimetric and osmometric studies. J. Appl. Polym. Sci. 15, 2073–2077. Ghosh, K., and Schnitzer, M. (1980). Macromolecular structures of humic substances. Soil Sci. 129, 266–276. Goring, D. A. I. (1971). Polymer properties of lignin and lignin derivatives. In “Lignins, Occurrence, Formation, Structure and Reactions” (K. V. Sarkanen and C. H. Ludwig, Eds.), pp. 695–768. Wiley, New York. Gregorich, E. G., Kachanosky, R. G., and Voroney, R. P. (1989). Carbon mineralization in soil size fractions after various amount of aggregate disruption. J. Soil Sci. 40, 649–659. Haider, K. (1987). The synthesis and degradation of humic substances in soil. In “Trans. 13th Congr. Intern. Soc. Soil Sci.,” pp. 644–656. Hamburg, Germany. Haider, K., and Martin, J. P. (1967). Synthesis and transformation of phenolic compounds by Epicoccus nigrum in relation to humic acid formation. Soil Sci. Soc. Am. Proc. 31, 766–772. Haider, K., Spiteller, M., Dec, J., and Sch¨affer, A. (2000). Silylation of soil organic matter. Extraction of humic compounds and soil-bound residues. In “Soil Biochemistry” (J.-M. Bollag and G. Stotzky, Eds.), pp. 139–170. vol. X. Dekker, New York. Handerson, H. A., and Hepburn, A. (1977). Fractionation of humic acid by gel permeation chromatography. J. Soil Sci. 28, 634–644. Hayase, K., and Tsubota, H. (1983). Sedimentary humic and fulvic acid as surface active substances. Geochim. Cosmochim. Acta 47, 947–952. Hayes, M. H. B. (1985). Extraction of humic substances from soil. In “Humic substances in Soil, Sediment, and Water” (G. R. Aiken, D. M. McKnight, R. L. Wershaw, and P. MacCarthy, Eds.), pp. 329–362. Wiley, New York. Hayes, M. H. B. (1997). Emerging concepts of the compositions and structures of humic substances. In “Humic Substances, Peats and Sludges” (M. H. B. Hayes and W. S. Wilson, Eds.), pp. 3–30. The Royal Society of Chemistry, Cambridge. Hedges, J. I. (1988). Polymerization of humic substances in natural environments. In “Humic Substances and their Role in the Environment” (F. H. Frimmel and R. F. Christman, Eds.), pp. 45– 58. Wiley, Chichester. Hjelmeland, L. M. (1990). Solubilization of native membrane proteins. In “Guide to Protein Puriﬁcation” (M. P. Deutscher, Ed.), Methods of Enzymology, vol. 182, pp. 253–264. Academic Press, San Diego, CA. Hu, W.-G., Mao, J., Xing, B., and Rohr, K. S. (2000). Poly(methylene) crystallites in humic substances detected by Nuclear Magnetic Resonance. Environ. Sci. Technol. 34, 530–534. Insam, H. (1996). Microorganisms and humus in soils. In “Humic Substances in Terrestrial Ecosystems” (A. Piccolo, Ed.), pp. 265–292. Elsevier, Amsterdam. Israelachvili, J. N. (1994). “Intermolecular and Surface Forces,” p. 282, 2nd ed.. Academic Press, London. Jambu, P., Ambl`es, A., Dinel, H., and Secouet, B. (1991). Incorporation of natural hydrocarbons from plant residues into an hydromorphic humic podzol following afforestation and fertilization. J. Soil Sci. 42, 629–636.

130

ALESSANDRO PICCOLO

Jenkinson, D. S. (1981). The fate of plant and animal residues in soil. In “The Chemistry of Soil Processes” (D. J. Greenland and M. H. B Hayes, Eds.), pp. 505–561. Wiley, Chichester. Kenworthy, I. P., and Hayes, M. H. B. (1997). Investigations of some structural properties of humic substances by ﬂuorescence quenching, In “Humic Substances, Peats and Sludges” (M. H. B. Hayes and W. S. Wilson, Eds.), pp. 39–45. The Royal Society of Chemistry, Cambridge. Kim, J-E., Fernandes, E., and Bollag, J-M. (1997). Enzymatic coupling of the herbicide betazon with humus monomers and characterization of reaction products. Environ. Sci. Technol. 31, 2392–2398. Kirkland, J. J. (1976). Porous silica microspheres for HPSEC. J. Chromatogr. 125, 231–250. Kohl, S. D., Toscano, P. J., Hou, W., and Rice, J. A. (2000). Solid-state 19F NMR investigation of hexaﬂuorobenzene sorption to soil organic matter. Environ. Sci, technol. 34, 204–210. Kononova, M. M. (1961). “Soil Organic Matter. Its Nature, Its Role in Soil Formation and in Soil Fertility.” Pergamon Press, New York. Kudryavtsev, A. V., Perminova, I. V., and Petrosyan, V. S. (2000). Size-exclusion chromatographic descriptors of humic substances. Anal. Chim. Acta 407, 193–202. Laue, T. M., and Rodhes, D. G. (1990). Determination of size, molecular weight and presence of subunits. In “Guide to Protein Puriﬁcation” (M. P. Deutscher, Ed.), pp. 566–587. Methods of Enzymology, vol. 182. Academic Press, San Diego, CA. Lehn, J.-M. (1995). “Supramolecular Chemistry” p. 7. VCH, Weinheim. ´ Chenu, C., Baudin, F., Leblond, C., Da Silva, M., Behar, F., Derenne, S., Largeau, Lichtfouse, E., C., Wehrung, P., and Albrecht, P. (1998). A novel pathway of soil organic matter formation by selective preservation of resistant straight-chain biopolymers: Chemical and isotope evidence. Org. Geochem. 28, 411–415. Linqvist, I. (1967). Adsorption effects in gel ﬁltration of humic acid. Acta Chem. Scand. 21, 2564–2566. Maillard, L. C. (1912). Formation d’humus et de combustibles mineraux sans intervention de l’oxygene atmospherique, des microorganismes des hautes temperatures ou des fortes pressions. C. R. Acad. Sci. Paris, 154, 66. Malcolm, R. L. (1990). The uniqueness of humic substances in each soil, stream, and marine environments. Anal. Chim. Acta 232, 19–30. Martin, J. P., and Haider, K. (1969). Phenolic polymers of Stachibotrys atra, Stachibotrys chartarum and Epicoccus nigrum in relation to humic acid formation. Soil Sci. 107, 260–270. Miano, T. M., Piccolo, A., Celano, G., and Senesi, N. (1992). Infrared and ﬂuorescence spectroscopy of glyphosate-humic acid complexes. Sci. Total Environ. 123/124, 83–92. Miles, C. J., and Brezonik, P. L. (1983). High performance size exclusion chromatography of aquatic humic substances. J. Chrom. 259, 499–503. Morra, M. J., Corapcioglu, M. O., von Wandruszka, R., Marshall, D. B., and Topper, K. (1990). Fluorescence quenching studies of Naphtalene and 1-Naphthol interaction with humic acid. Soil Sci. Soc. Am. J. 54, 1283–1289. Mukherjee, P. N., and Lahiri, A. (1958). Polyelectrolytic behaviors of humic acids. Fuel 37, 220–226. Nardi, S., Arnoldi, G., and Dell’Agnola, G. (1988). Release of the hormone-like activities from Allophobora rosea (sav.) and Allophobora caliginosa (sav.) feces. Can. J. Soil Sci. 68, 563–567. Nierop, K. G. J. (1998). Origin of aliphatic compounds in a forest soil. Org. Geochem. 29, 1009–1016. Nierop, K. G. J., and Buurman, P. (1999). Water-soluble organic matter in incipient podzols: Accumulations in B-horizons or in ﬁbres? Eur. J. Soil Sci. 50, 701–711. Nierop, K. G. J., Buurman, P., and de Leeuw, J. W. (1999). Effect of vegetation on chemical composition of H horizons in incipient podzols as characterized by 13C NMR and pyrolysis-GC/MS. Geoderma 90, 111–129. Nishio, M., Hirota, M., and Umezawa, Y. (1998). “The CH/π Interaction. Evidence, Nature and Consequences.” Wiley, New York. Orlov, D. S., Ammosova, I. A., and Glebova, G. I. (1975). Molecular parameters of humic acids. Geoderma 13, 211–229.

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

131

Paustian, K., Andr´en, O., Janzen, H. H., Lal, R., Smith, P., Tian, G., Tiessen, H., van Noordwijk, M., and Woomer, P. L. (1997). Agricultural soils as a sink to mitigate CO2 emissions. Soil Use Manag. 13, 230–244. Perminova, I. V., Frimmel, F. H., Kovalevskii, D. V., Abbt-Braun, G., Kudryatsev, A. V., and Hesse, S. (1998). Development of predictive model for calculation of molecular weight of humic substances. Water Res. 32, 872–881. Peuravuori, J., and Pihlaja, K. (1997). Molecular size distribution and spectroscopic properties of aquatic humic substances. Anal. Chim. Acta 337, 133–149. Piccolo, A. (1988). Characteristics of soil humic substances extracted with some organic and inorganic solvents and puriﬁed by the HCl–HF treatment. Soil Sci. 146, 418–426. Piccolo, A. (1996). Humus and soil conservation, In “Humic Substances in Terrestrial Ecosystems” (A. Piccolo, Ed.), pp. 225–264. Elsevier, Amsterdam. Piccolo, A., Campanella, L., and Petronio, B. M. (1990). 13C-NMR spectra of humic substances extracted with different mechanisms. Soil Sci. Soc. Am. J. 54, 750–755. Piccolo, A., Celano, G., and Conte, P. (1996). Interactions of Glyphosate with humic substances. J. Ag. Food Chem. 44, 2442–2446. Piccolo, A., Conte, P., and Cozzolino, A. (1999). Effects of mineral and monocarboxylic acids on the molecular association of dissolved humic substances, Eur. J. Soil Sci. 50, 687–694. Piccolo, A., Conte, P., and Cozzolino, A. (2000). Differences in high performance size exclusion chromatography between humic substances and macromolecular polymers. In “Humic Substances. Versatile Components of Plant, Soils and Water” (E. A. Gabbhour and G. Davies, Eds.), pp. 111–124. Royal Society of Chemistry, Special Publication no. 259, Cambridge, UK. Piccolo, A., Conte, P., Cozzolino, A., and Paci, M. (2001). Combined effects of an oxidative enzyme and dissolved humic substances on 13C-labelled 2,4-D herbicide as revealed by high-resolution 13 C-NMR spectroscopy. J. Ind. Microbiol. Biotechnol. 26, 70–76. Piccolo, A., Conte, P., and Cozzolino, A. (2001). Chromatographic and spectrophotometric properties of dissolved humic substances as compared to macromolecular polymers. Soil Sci. 166, 174–185. Piccolo, A., Conte, P., Cozzolino, A., and Spaccini, R. (2001). Molecular Sizes and Association Forces of Humic Substances in Solution. In “Humic Substances and Chemical Contaminants” (C. E. Clapp, M. H. B. Hayes, N. Senesi, P. Bloom, and P. M. Jardine, Eds.), pp. 89–118. Soil Science Society of America, Inc., Madison, WI. Piccolo, A., Conte, P., Scheunert, I., and Paci, M. (1998). Atrazine interactions with soil humic substances of different molecular structure. J. Environ. Q. 27, 1324–1333. Piccolo, A., Conte, P., Trivellone, E., van Lagen, B., and Buurman, P. (2001). Reduced heterogeneity of a lignite humic acid by preparative HPSEC following interaction with an organic acid. Characterization of size separates by PYR-GC-MS and 1H-NMR spectroscopy. Environ. Sci. Technol. (in press). Piccolo, A., Cozzolino, A., Conte, P., and Spaccini, R. (2000). Polymerization of humic substances by an enzyme catalyzed oxidative coupling. Naturwissenschaften 87, 391–394. Piccolo, A., and Mbagwu, J. S. C. (1990). Effects of different organic waste amendments on soil microaggregates stability and molecular sizes of humic substances. Plant Soil 123, 27–37. Piccolo, A., and Mbagwu, J. S. C. (1999). Role of hydrophobic components of soil organic matter in soil aggregate stability. Soil Sci. Soc. Am. J. 63, 1801–1810. Piccolo, A., and Mirabella, A. (1987). Molecular weight distribution of peat humic substances extracted with different inorganic and organic solutions. Sci. Total Environ. 62, 39–46. Piccolo, A., Nardi, S., and Concheri, G. (1992). Structural characteristics of humic substances as related to nitrate uptake and growth regulation in plant systems, Soil Biol. Biochem. 24, 373– 380. Piccolo, A., Nardi, S., and Concheri, G. (1996a). Macromolecular changes of soil humic substances induced by interactions with organic acids. Eur. J. Soil Sci. 47, 319–328.

132

ALESSANDRO PICCOLO

Piccolo, A., Nardi, S., and Concheri, G. (1996b). Micelle-like conformation of humic substances as revealed by size-exclusion chromatography. Chemosphere 33, 595–60. Piccolo, A., Rausa, R., and Celano, G. (1990). Characteristics of molecular size fractions of humic substances derived from oxidized coal. Chemosphere 24, 1381–1387. Piccolo, A., Spaccini, R., Haberhauer, G., and Gerzabek, M. H. (1999). Increased sequestration of organic carbon in soil by hydrophobic protection. Naturwissenschaften 86, 496–499. Piccolo, A., Zaccheo, P., and Genevini, P. L. (1992). Chemical characterization of humic substances extracted from organic-wastes-amended soils. Bioresource Technol. 40, 275–282. Piret, E. L., White, R. G., Walther, H. C., and Madded, A. J. (1960). Some physicochemical properties of peat humic acids. Sci. Proc. R. Dublin Soc., Ser. 1, 69–79. Posner, A. M., and Creeth, J. M. (1972). A study of humic acids by equilibrium centrifugation. J. Soil Sci. 23, 333–341. Rausa, R., Mazzolari, E., and Calemma, V. (1991). Determination of molecular size distributions of humic acids by high-performance size-exclusion chromatography, J. Chrom. 541, 419–429. Reid, P. M., Wilkinson, A. E., Tipping, E., and Malcolm, N. J. (1990). Determination of molecular weights of humic substances by analytical (UV scanning) ultracentrifugation. Geochim. Cosmochim. Acta 54, 131–138. Reuter, J. H., and Perdue, E. M. (1981). Calculation of molecular weights of humic substances from colligative data: application to aquatic humus and its molecular size fractions. Geochim. Cosmochim. Acta 45, 2017–2022. Ritchie, G. S. P., and Posner, A. M. (1982). The effect of pH and metal binding on the transport properties of humic acids. J. Soil Sci. 33, 233–247. Ryan, D. K., and Weber, J. H. (1982). Fluorescence quenching titration for determination of complexing capacities and stability constants of fulvic acid. Anal. Chem. 54, 986–990. Saito, Y., and Hayano, S. (1979). Application of high performance aqueous gel permeation chromatography. J. Chrom. 177, 390–392. Saiz-Jimenez, C. (1995). Analytical pyrolysis of humic substances: Pitfalls, limitations and possible solutions. Environ. Sci. Technol. 28, 1773–1780. Saiz-Jimenez, C. (1996). The chemical structure of humic substances: recent advances. In “Humic Substances in Terrestrial Ecosystems” (A. Piccolo, Ed.), pp. 1–44. Elsevier, Amsterdam. Scheffer, F., and Ulrich, B. (1960). “Lehrbuch der Agrikulturchemie und Bodenkund. Teil III. Humus und Humusd¨ungung,” 2nd ed. Ferdinand Enke, Stuttgart. Schlesinger, W. H. (1997). “Biogeochemistry. An Analysis of Global Change,” 2nd ed. Academic Press, San Diego, CA. Schnitzer, M., and Khan, S. U. (1972). “Humic Substances in the Environment.” Dekker, New York. Schulten, H.-R., and Leinweber, P. (2000). New insights into organic-mineral particles: Composition, properties, and models of molecular structure. Biol Fertil Soils 30, 399–432. Schulten, H.-R., Leinweber, P., and Schnitzer, M. (1998). Analytical pyrolysis and computer modeling of humic and soil particles. In “Structure and Surface Reactions of Soil Particles” (P. M. Huang, N. Senesi, and J. Bufﬂe, Eds.), pp. 281–324. Wiley, New York. Schwarzenbach, R. P., Gschwend, P. M., and Imboolen, D. M. (1993). Environmental Organic Chemistry, Wiley, New York. Simpson, A. J., Kingery, W. L., Hayes, M. H. B., Spraul, M., Humpfer, E., Dvorstsak, P., Kesserbaum, R. M., Godejohann, M., and Hofmann, M. (2000). The structure and association of organic molecules in the terrestrial environment. In “Proceedings of the 10th International Meeting of the Humic Substances Society,” 24–28 July 2000, vol. 2, pp. 1109–1112. Toulouse, France. Spaccini, R., Piccolo, A., Haberhauer, G., and Gerzabek, M. H. (2000). Transformation of organic matter from maize residues into labile and humic fractions of three European soils as revealed by 13 C distribution and CPMAS-NMR spectra. Eur. J. Soil Sci. 51, 583–594.

SUPRAMOLECULAR STRUCTURE OF HUMIC SUBSTANCES

133

Spaccini, R., Zena, A., and Piccolo, A. (2000). Carbohydrates distribution in size-aggregates of three European soils in a climate gradient. Fres. Environ. Bull. 9, 468–476. Spaccini, R., Piccolo, A., Conte, P., Haberhauer, G., and Gerzabek, M. H. (2001). Increased soil organic carbon sequestration through hydrophobic protection by humic substances. Soil Biol. Biochem. (in press). Spaccini, R., Piccolo, A., Haberhauer, G., Stemmer, M., and Gerzabek, M. H. (2001). Decomposition of maize straw in different European soils as revealed by DRIFT spectra of soil particle fractions. Geoderma, 99, 245–260. Spaccini, R., Zena, A., Igwe, C. A., Mbagwu, J. S. C., and Piccolo, A. (2001). Carbohydrates in water-stable aggregates and particle size fractions of forest and cultivated soils in two contrasting tropical ecosystems. Biogeochem. 53, 1–22. Staudinger, H. (1935). “Die hochmolekularen organischen Verbindungen, Kautschuk und Cellulose,” Springer, Berlin. Stevenson, F. J. (1994). “Humus Chemistry. Genesis, Composition, Reactions,” 2nd ed. Wiley, New York. Stevenson, F. J., van Winkle, Q., and Martin, W. P. (1953). Physico-chemical investigations of clayadsorbed organic colloids: II. Soil Sci. Soc. Am. Proc. 17, 31–34. Summers, R. S., Cornel, P. K., and Roberts, P. V. (1987). Molecular size distribution and spectroscopic characterization of humic substances. Sci. Total Environ. 62, 27–37. Swift, R. S. (1989). Molecular weight, shape, and size of humic substances by ultracentrifugation. In “Humic Substances II. In Search of Structure” (M. H. B. Hayes, P. McCarthy, R. L. Malcolm, and Swift R. S., Eds.), pp. 449–466. Wiley, New York. Swift, R. S. (1999). Macromolecular properties of soil humic substances: Fact, ﬁction, and opinion. Soil Sc. 164, 790–802. Swift, R. S., and Posner, A. M. (1971). Gel chromatography of humic acid. J. Soil Sci. 22, 237–249. Tanford, C. (1980). “The Hydrophobic Effect: Formation of Micelles and Biological Membranes.” Wiley, New York. Tate, R. L. III (1999). Micelles, subunits, and the Mediterranean sun. Soil Sci. 164, 775–776. Thurman, E. M., Wershaw, R. L., Malcolm, R. L., and Pinkney, D. J. (1982). Molecular size and weight measurements of humic substances. Org. Geochem. 4, 27–35. Trubetskoj, O. A., Trubetskaja, O. E., Afanas’eva, G. V., Reznikova, O. I., and Saiz-Jimenez, C. (1997). Polyacrilamide gel electrophoresis of soil humic acid fractionated by size-exclusion chromatography and ultraﬁltration. J. Chrom. 767, 285–292. van Dijk, H. (1972). Colloidal chemical properties of humic matter. In “Soil Biochemistry” (A. D. McLaren and G. H. Peterson, Eds.), pp. 16–34, Vol. 2. Dekker, New York. Vartianen, T., L¨umataien, A., and Keuiranen, S. (1987). The use of TSK size exclusion columns in determination of the quality and quantity of humus in raw and drinking waters. Sci. Tot. Environ. 62, 75–84. Violante, A., Arienzo, M., Sannino, F., Colombo, C., Piccolo, A., and Gianfreda, L. (1999). Formation and characterization of OH-Al-humate-montmorillonite complexes. Org. Geochem. 30, 461–468. Visser, S. A. (1964). A physico-chemical study of the properties of humic acids and their changes during humiﬁcation. J. Soil Sci. 15, 202–219. von Wandruszka, R., Schimpf, M., Hill, M., and Engebretson, R. (1999). Characterization of humic acid size fractions by SEC and MALS. Org. Geochem. 30, 229–235. Waksman, S. A. (1936). “Humus.” Williams and Wilkins, Baltimore. Wander, M. M., and Traina, S. J. (1996). Organic matter fractions from organically and conventionally managed soils: II. Characterization of composition. Soil Sci. Soc. Am. J. 60, 1087–1094. Wanner, U., Burauel, P., and F¨uhr, F. (2000). Characterization of soil-bound residue fractions of the fungicide dithianon by gel permeation chromatography and polyacrilamide gel elctrophoresis. Environ. Poll. 108, 53–59.

134

ALESSANDRO PICCOLO

Welte, E. (1955). New results of humus research. Angew. Chem. 67, 153–155. Wershaw, R. L. (1986). A new model for humic materials and their interactions with hydrophobic chemicals in soil-water and sediment-water systems. J. Contam. Hydrol. 1, 29–45. Wershaw, R. L. (1989). Size and shape of humic substances by scattering techniques. In “Humic Substances II. In Search of Structure” (M. H. B. Hayes, P. McCarthy, R. L. Malcolm, and Swift R. S., Eds.), pp. 545–559. Wiley, New York. Wershaw, R. L. (1993). Model for humus in soils and sediments. Environ. Sci. Technol. 27, 814–816. Wershaw, R. L. (1999). Molecular aggregation of humic substances. Soil Sci. 164, 803–813. Wershaw, R. L., and Aiken, G. R. (1985). Molecular size and weight measurements of humic substances. In “Humic Substances in Soil, Sediment, and Water” (G. R. Aiken, D. M. McKnight, R. L. Wershaw, and P. MacCarthy, Eds.), pp. 477–492. Wiley, New York. Yonebayashi, K., and Hattori, T. (1987). Surface active properties of soil humic acids. Sci. Total Environ. 62, 55–64. Zalba, P., and Quiroga, A. R. (1999). Fulvic acid carbon as a diagnostic feature for agricultural soil evaluation. Soil Sci. 164, 57–61.

WATER-SAVING AGRICULTURE IN CHINA: AN OVERVIEW Huixiao Wang,1 Changming Liu,2 and Lu Zhang3 1 State Key Laboratory of Water Environment Simulation Key Laboratory for Water and Sediment Sciences, Ministry of Education Beijing Normal University, Beijing 100875, China 2 Institute of Geographic Sciences and Natural Resources Research The Chinese Academy of Sciences, Beijing 100101, China 3 CSIRO Land and Water, Canberra Laboratory PO Box 1666 Canberra, ACT 2601, Australia

I. Introduction II. Water-Saving Agriculture as a System III. Water-Use Efficiency A. Molecular Level B. Single-Leaf Level C. Canopy (Community) Level D. Field Level E. Regional Level IV. A Rationale for the Use of Water Resources A. Rational Utilization of Rainwater B. Five-Water Interaction Mechanism and Its Application C. Utilization of Low-Quality Water V. Water-Saving Engineering Measures A. Water Conveyance Structures B. Water-Saving Irrigation Techniques VI. Water-Saving Agronomic Practices A. Water-Matched Production B. Biological Water-Saving Technology C. Water-Saving Irrigation Schemes D. Soil Moisture Conservation E. Soil Fertility VII. Water-Saving Management A. Administrative Systems B. Water Prices and Fees C. Laws and Regulations D. Technical Training and Services E. Expert System for Water-Saving Agriculture F. Water-Saving Products and Marketing VIII. Concluding Remarks References 135 Advances in Agronomy., Volume 75 C 2002 by Academic Press. All rights of reproduction in any form reserved. Copyright 0065-2113/02 $35.00

136

H. WANG ET AL.

Agricultural development in China is limited by available water. Currently, agricultural water use accounts for over 70% of the total water usage. To ensure China’s food security, it is necessary to promote water-saving agriculture, which is considered as an integrated system including four components: rational utilization of agricultural water resources, water-saving irrigation, agronomic water-saving techniques, and agricultural management. This paper reviews the current status of the water-saving agricultural research in China. It outlines some of the problems in the current agricultural systems and discussed the potential of the various techniques and measures for improving overall water use efficiency. It is suggested that biological and engineering measures need to be integrated. There is also a need to establish a commercial environment for promoting water-saving agricultural technology. This review is largely based on the research conducted in China during the C 2002 Academic Press. last two decades.

I. INTRODUCTION Over half of the area of China is arid or semiarid with an annual precipitation of less than 400 mm. In the 1990s, the drought-affected area was around 26.7 million hm2, and this reduced grain yield by 100 million tons per year (Luo, 1999). When drought strikes, it becomes a serious natural disaster for agricultural production. In recent times, water shortages have become major factors restricting agricultural development. In China, agricultural production depends highly on irrigation with two-thirds of the agricultural output derived from irrigated land, representing 46% of the total cultivated land (Zhang Q, 1999). Due to industrialization and urbanization, there is an increasing competition for limited water resources between rapidly growing urban and industrial sectors and the agricultural sector. During the last 2 decades, the percentage of agricultural water use in China decreased from 88 to 72% of the total. This percentage will be even smaller by the middle of the 21st century (Zhang Q, 1999). United Nation projections show the population of China reaching 1.5 billion in 2030, and as a result agricultural water demand will need to increase to meet the additional food requirements. According to Han (1999), China will need 640 million tonnes of grain each year, and the deﬁcit in water resources will be 260 billion m3 annually. This presents a serious challenge to the agricultural sector in terms of managing water resources and maintaining sustainable development. Precipitation in China is highly variable, both in time and in space. About 44% of the population and 58% of the cultivated land are in the northern and northeastern provinces, but these regions possess only 14% of the total water

137

WATER-SAVING AGRICULTURE IN CHINA

Major drainage net work in Nen

R iver

CHINA H ulun L ake

Erq

isi

B urqin

R iv e

r

Talimu

o L ia

K orla

R i ver

ve

r

Changcun r R iv e

Wulumuqi

Yili R iver

Kashi

Ri ng aji a gh u S on H aerbin

Lu H uhehaote

R

B eij ing

Ri

R iver

Yaiyuang

ve

r

Ji nan Z h engzhou

L anz hou

Y el

lo w

ll o

Xi ning

Ye

Golmud

T i anjing S h ij iazhuang

w

Y i nchuan

S h enyang

an

Y E L L OW S E A Xi an

H uai R

Yalu z ha ng

bu

Chengdu

R iver

L as ha

Nanjing

S h anghai

W uhan

Chan gjiang

R

Chongqing

Nans hang Changs ha Fu zhou

K unming

Taibei

T A IW AN

Xi

R iver

Nanning

H aik ou

Guangzhou H ongkong

S OU T H CH I N A S E A H AIN AN IS L AND

Figure 1 A map of China showing the different catchments

(Brown and Halweil, 1998). The percentage of water resources used —i.e., the ratio between water consumption and the available water—has reached 46% for the northern rivers, over 50% for the Huaihe and Yellow rivers, and over 80% for the Haihe River. Different catchments of China are shown in Fig. 1. According to the publication of the UN and other international organizations, “Overall Estimation on World Fresh Water Resources,” serious water shortage occurs when the percentage of water resources used is above 40% (Feng, 1999). There is no reliable surface water resource for irrigation in these regions, so groundwater has been overdrawn, causing the regional groundwater table to drop dramatically (as shown by later examples) and resulting in a series of environmental problems such as land subsidence, drying up of rivers, seawater intrusion, secondary salinization, and soil desertiﬁcation. Changes in the buried groundwater tables in the middle part of the North China Plain (NCP) from the piedmont to

138

H. WANG ET AL.

Groundwater table (m)

YS

SJZ

LC

JX

SL

SX

WQ

XX

BT

NP

CZ

YS

HX

0 10 20 30 40

1964

1984

1993

Figure 2 Changes in the groundwater tables in the middle part of the North China Plain from the piedmont to the coast. (YS: Yuanshi, SJZ: Shijiazhuang, LC: Luancheng, JX: Jinxian, SL: Shulu, SX: Shenxian, WQ: Wuqiang, XX: Xianxian, BT: Botou, NP: Nanpi, CZ: Cangzhou, YS: Yanshan, HX: Haixing).

the seashore are shown in Fig. 2. Form the data in Fig. 2, the speed of the drop in the groundwater table from 1984 to 1993 was doubled compared to that from 1964 to 1984, especially at the piedmont of the Taihang Mountains in the NCP including Shijiazhuang, Luancheng, and Jinxian. In these areas, the groundwater table dropped to 0.64 m per year from 1964 to 1984 and to about 1.22 m per year from 1984 to 1993. It is estimated that the shallow aquifer in the groundwater irrigation district of the Haihe Plain will be used up within 20–30 years if there are no appropriate measures taken. The development of water-saving agriculture is essential in northern China. In southern China where the annual rainfall is high, seasonal droughts still occur and threaten agricultural production. Water saving is also necessary in this region both for overcoming drought and for controlling water pollution. However, the problem is different from that in northern China and will not be discussed in this review. Water resources management is still underdeveloped in most parts of northern China, which adds another dimension to the water-shortage problem. The seriousness of the water shortages is not yet fully appreciated. Water-use efﬁciency (WUE) is very low in all sectors, particularly in agriculture due to irrigation methods and inappropriate management practices. It is estimated that with traditional ﬂood irrigation in northern China, up to 60% of the water evaporates from open canals and ﬁelds. The actual irrigation water supply is 0.5–1.5 times more than the crop water requirement, and the effective-use coefﬁcient of irrigation water is rather low, only 0.4–0.5, while in the developed countries it has reached 0.8–0.9 (Liu and Li, 1999). Water pollution is also threatening agriculture in China, and there are increased risks of soil degradation.

WATER-SAVING AGRICULTURE IN CHINA

139

Since an event of severe drought in 1972 in northern China, research on watersaving agriculture has been conducted for over 20 years. Especially after the 1990s, the central government has increasingly paid much more attention to water-saving research than ever before, and it emphasizes that water-saving irrigation is a revolutionary measure. As a result, a number of projects have been supported. Two examples are “The Applied Scientiﬁc Study on Water-Saving Agriculture in the North China Plain” and “Research and Demonstration on the Technology System of Water-Saving Agriculture in the North China Plain.” Another project is, “Scientiﬁc and Technological Demonstration Project on High Efﬁcient Water Use in Agriculture,” which was supported by the Ministry of Science and Technology and the Ministry of Water Resources in eight provinces or autonomous regions (Hebei, Shandong, Shanxi, Shaannxi, Gansu, Inner Mongolia, Ningxia, and Xinjiang). The general objective is to develop models and key techniques for highly efﬁcient water use in agriculture in northern China. However, the dissemination and application of the water-saving techniques are rather slow, and fail to help solve the growing water shortage problem in northern China. Most of the water-saving studies have been characterized by single measures, a move which would be unfavorable to an overall solution of water saving. Some important issues on water-saving agriculture still need to be explored. Here, the authors will cite research achievements on the water-saving agriculture in recent years in northern China and discuss the essence of water-saving agriculture. In this review, water-saving agriculture is considered to be an integrated system including four components: rational utilization of agricultural water resources, water-saving irrigation, agronomic water-saving techniques, and agricultural management in water saving. The aim is to provide an overall picture of the water-saving agriculture developments in northern China.

II. WATER-SAVING AGRICULTURE AS A SYSTEM “Water-saving agriculture” is a complex system involving agronomic and hydraulic engineering techniques in the integrated exploitation of water, soil, and crop resources. Only when water-saving agriculture is considered as an integrated system can comprehensive water-saving measures be properly evaluated and applied. Water for agriculture is comprised of waters from both precipitation and irrigation. There are three main stages from the supply of irrigation water to its ﬁnal use on crops. The ﬁrst stage involves taking water from reservoirs or groundwater pumping stations and applying it to ﬁelds; the second stage can be considered the transfer of the irrigation water to the soil; and the ﬁnal stage is its uptake by plants. Water losses will occur at each stage. To achieve an overall water-saving target, it is essential to examine appropriate water-saving measures for each stage.

140

H. WANG ET AL.

Water-saving agriculture system

Rational exploitation and utilization of water resources

Irrigation engineering and techniques

Agronomic water-saving practices

Water-saving management measures

Figure 3 A water-saving agricultural system.

First, wise utilization of water resources and modern engineering measures can reduce transmission water losses. Second, updated irrigation techniques and ﬁeld works can increase the ratio of irrigation water and soil water in the root zone. Finally, measures to maximize crop water-use efﬁciency, such as agronomic measures and biological techniques, can favorably adjust the ratio of crop transpiration and soil evaporation. Water-saving agriculture aims at more efﬁcient water use, and the key issue is to increase the efﬁciency and economics of the water supply. So water-saving agriculture does not simply mean water-saving irrigation; it also includes effective use of precipitation on dry land. Apart from making full use of precipitation, agricultural water resources should be rationally exploited, utilized, and managed through both agronomic and hydraulic engineering techniques. Water-saving agriculture means obtaining high yields, good crop quality, and high-water-use efﬁciency. In this way, sustainable agriculture can be achieved. It will require better water management, soil amelioration, the scheduling of agricultural production, and the improvement of tillage and planting systems. The details of the agricultural water-saving measures are rather involved and include agronomic water saving (crop physiology, ﬁeld regulation), irrigation water saving (irrigation engineering, irrigation techniques), and management water saving (policy, law, and system). The combination of all these components forms a framework of a water-saving agriculture system, and this is generalized in Fig. 3. Each component has different aspects and will be analyzed and discussed in detail in the following.

III. WATER-USE EFFICIENCY A central issue in water-saving agriculture is how to effectively use irrigation water or precipitation. It is therefore clear that water-use efﬁciency is the basic indicator for gauging the effectiveness of water-saving agriculture. In the past, irrigation quotas were normally set by crop water requirements in order to achieve

WATER-SAVING AGRICULTURE IN CHINA

141

high crop yields, but this method ignored ways of using limited water resources to obtain the highest proﬁt under these conditions. As water shortages have developed, a number of governments have turned their attention to research more efﬁcient water use so that water resources can be used in a sustainable way. In a previous review on water-use efﬁciency, Stanhill (1986) used the term WUE in at least two different ways, hydrologically and physiologically. In a purely hydrological sense, WUE has been deﬁned as the ratio of the volume of water used productively by crops to that potentially available for this purpose, that is, WUE =

ET , P + I + SWa

(1)

where ET is the sum of water transpired and evaporated from the study area; P, I, SWa are the rainfall, applied irrigation, and available soil water, respectively. For irrigation areas, WUE was expressed by Bos and Nugteren (1974) as WUE = SW/W,

(2)

where SW is the increase in the soil water content of the root zone following irrigation; and W is the total quantity of water supplied to the irrigation area. Kang et al. (1994) referred to this as the ﬁeld water-storing efﬁciency. The total irrigation coefﬁcient (η) is divided into three components: η = η1 × η2 × η3 ,

(3)

where η1 , η2 , and η3 are the water conveyance, farm ditch, and ﬁeld application coefﬁcients, respectively. The term WUE in its physiological sense is more commonly used in several different scales, such as molecular, single-leaf, canopy, ﬁeld, and regional levels. It can be called crop water-use efﬁciency.

A. MOLECULAR LEVEL The WUE at this level has been extensively studied. It can be used to analyze the biological differences in WUE among species and varieties, providing the theoretical basis for using the molecular biotechnique to increase the WUE of crop varieties. However, this issue will not be discussed in detail here.

B. SINGLE-LEAF LEVEL At the single-leaf level, WUE is deﬁned as the net CO2 uptake by leaf per unit of transpiration, and is expressed as the ratio of leaf photosynthesis rate to leaf

142

H. WANG ET AL.

transpiration rate. It can be used as the upper limit for crop water-use efﬁciency. The water vapor and CO2 ﬂuxes can be expressed as the ratio of concentration gradient and diffusion resistance and can also be measured with gas-exchange equipment. Assuming that CO2 and water vapor take identical paths between the leaf cell walls and bulk air, the WUE at this level can be calculated as (Fischer and Turner, 1978) WUE = ⌈c · Dc (ra + rs )⌉ / [e · De (ra + rs + ri )] ,

(4)

where c and e are the leaf-to-air concentration gradients for CO2 and water vapor, respectively; Dc and De are the diffusivities of CO2 and water vapor, respectively; and ra, rs, and ri are the boundary layer, stomatal, and internal resistances to diffusion, respectively. WUE is affected by such environmental factors as air saturation deﬁcit, air temperature, incident irradiance, and leaf orientation/leaf movement. WUE also varies with genotype, the leaf traits ra, rs, and ri, , and leaf water potential ψleaf . According to the studies by Condon et al. (1987) and Wright et al. (1994), leaf WUE was negatively related to 13C discrimination. This parameter could be a good index to use when screening for high WUE in variety breeding trials.

C. CANOPY (COMMUNITY) LEVEL The WUE at the canopy level is deﬁned as the ratio of the net CO2 assimilation of a crop community to that of transpiration, that is, the ratio of the canopy CO2 ﬂux to the water vapor ﬂux for the canopy transpiration. It can be expressed as WUE = Fc /T,

(5)

where Fc is the canopy CO2 ﬂux; T is the water vapor ﬂux for the canopy transpiration.

D. FIELD LEVEL The WUE at the ﬁeld level is deﬁned as the grain yield per unit of water used. The yield can be expressed as the net biomass Yb (including roots) or the grain yield Ye . The ﬁeld level WUE is calculated as WUE = Y /W U,

(6)

where Y is the dry matter yield (Yb ) or the grain yield (Ye ) in kilograms per hectare; and water use (WU in millimeters) can be total evapotranspiration, irrigation water applied, or precipitation for different purposes.

WATER-SAVING AGRICULTURE IN CHINA

143

E. REGIONAL LEVEL The WUE at this level is deﬁned as the ratio of the yield per year in a region to the water use (tonne per cubic meter) in the same period. Its calculation is complex because there are usually several crops growing in the same period, and also there are different kinds of landscapes within a region. In China, statistics on grain yield, planting areas, water use including irrigation, and precipitation are made, and WUE is calculated at the county (subregion) level. The WUEs for every county are integrated to obtain the WUE of the whole region, and are then mapped.

IV. A RATIONALE FOR THE USE OF WATER RESOURCES Precipitation, surface water, soil water, groundwater, and evapotranspiration are closely linked to each other to form the hydrological cycle, and also transform into various hydrologic states to form a continuum governed by the water balance. Rational exploitation and utilization of water resources include the following aspects (see Fig. 4) and will be described separately later.

Rainwater harvesting system Effective use of precipitation

Rain-fed agriculture

Soil water reservoir establishment

Five-water interaction systems Rational exploitation and utilization of water resources

Five-water interaction mechanisms

Inter-face processes

Conjunctive well & canal irrigation Utilization of brackish water Utilization of low quality water Sewage utilization after treating

Figure 4 A rationale for the use of water resources.

144

H. WANG ET AL.

A. RATIONAL UTILIZATION OF RAINWATER Dryland agriculture is highly dependent on precipitation and its variability, while irrigated agriculture relies on additional water from irrigation. The effective utilization of rainfall is of great signiﬁcance for both dry land and irrigated agriculture. The effective precipitation during the crop growing season depends on rainfall characteristics, surface cover, soil texture, cropping system, and irrigation. Effective precipitation is controlled by inﬁltration into the soil and the water-storage capacity of the soil. Also, the vertical inﬁltration process is closely related to the groundwater table. Experimental results from the Linxi Irrigation Station, Hebei Province, showed that the optimum groundwater table for rainfall inﬁltrating into soil and recharging to groundwater is about 4–6 m (Fan, 1986). The water-conserving ability relates to soil structure and organic matter content. According to the calculation of Liu and Wei (1989), the annual effective rainfall was 423 mm in the plain north of the Yellow River; the effective rainfall at Haihe and Luanhe basins was 50–70% of the total, and the rainwater inﬁltration depth was normally less than 0.6 m. In the Shandong irrigation districts, the ratios between the effective rainfall and the total at different crop growing seasons ranged from 40 to 90%, and the effective rainfall could meet 30 to 40% of the crop water requirement. This means that the potential for increasing rainfall-use efﬁciency is quite high in northern China. Dryland contour cultivation, deep tillage, and other agronomic biochemical water-conservation measures could greatly increase the effective rainfall ratio. Field small-scale water-storing cisterns could retain rainfall-derived surface runoff during heavy rainfall, which then could be used for irrigation (Lei, 1999). There are large areas with annual precipitation of over 200 mm (mostly concentrated from June to September) in northern China, which could make use of rainwater by collecting runoff from bare slopes, road surfaces, courtyards, house roofs, and other compressed surfaces (Zhang et al., 1999). In the Loess Plateau, 76% of cultivated land has an annual precipitation of over 400 mm, so the potential for developing rainwater-harvesting agriculture is high. However, the current rainwater-use coefﬁcient is only about 30% (Li Y, 1999). Rainfall in the NCP is highly variable and cannot meet water requirements at every stage of crop growth. Regulating and storage measures, such as the establishment of different kinds of reservoirs (surface, soil, and ground reservoir), have been used to make full use of precipitation in this region. When large areas are considered, the soil water reservoir becomes increasingly important. Experimental data from the report by the Ministry of Water Resources (1990a) for Hanwang, Ranzhuang, Shangqiu, and Yucheng showed that the seasonal regulating capacity of the soil water reservoir could have a range of 200–300 mm. Storing rainfall in the soil during a rainy season following a continuous period of drought could regulate water usage between seasons. The storage capacity of the soil water reservoir can be calculated from three basic soil water parameters: saturated soil water

WATER-SAVING AGRICULTURE IN CHINA

145

content, ﬁeld capacity, and wilting point. At the Xueye irrigation district station, Laiwu in Shandong Province, the total capacity, ﬁeld capacity, wilting point, and the effective storage capacity of clay soil (0–200 cm) were 783, 688, 382, and 306 mm, respectively. The regulating capacity of soil water could reach 306 mm, which accounted for 62% of the winter wheat water requirement (Li X, 1999). In recent years, a catchment system for rainwater utilization has been successfully developed both for domestic water use and for agricultural production in arid and semiarid regions, especially in poor mountain areas in northwest China, where annual precipitation is 200–400 mm and no irrigation schemes have been developed. Rainfall in these areas is concentrated from July to September as heavy storms, and the resulting surface runoff is on sloping lands with poor vegetation covers. The summer dominance of precipitation is more obvious in areas with lower rainfall. In these regions, precipitation could not satisfy the water requirements of summer-harvested crops such as wheat and pea, the main crops in the region, resulting in low and variable yields. Although rainfall is the only source of water for agriculture in the region, rainwater-use efﬁciency was still low at only 14–32% (Shang and Chang, 1999). Developing rainwater catchment utilization techniques is therefore of great signiﬁcance in arid areas. There has been a trend in mountain areas from traditional rain-fed agriculture to the rainwater-harvesting agriculture. Water supply in the rainwater-harvesting agriculture could match crop water requirements; hence soil evaporation loss may be reduced by up to 30% (Zhang C, 1999). Rainwater utilization is an old technique. Some 2500 years ago, a large-scale reservoir was built in Shou County, Anhui Province, to collect rainwater for agricultural use (Zhang et al., 1999). In the northwest arid and semiarid mountain areas of the Loess Plateau, many rainwater utilization techniques have been developed in the struggle against drought. Here are two prime examples. One in the Gansu Province is called “engineering 1-2-1,” which means one (1) family building a 100-m2 rainwater-collecting compressed area, with two (2) 30- to 50-m2 cisterns used to collect rainwater for domestic use, and using the water collected to irrigate an economic crop of one (1) Chinese mu (0.067 hm2) in the family courtyard. The other is “engineering 1-1-2” in Inner Mongolia which calls for one (1) family having one (1) cistern for storing rainwater as the supplementary irrigation water for the water-saving farmland of two (2) Chinese mu (Li and Shi, 1999). The entire system of rainwater catchment utilization has three main components: a rainwater-collecting compressed surface, a rainwater-storage cistern, and an irrigation scheme. Several experiments have been conducted in this area on the position, area, and materials for forming a leakage-proof collecting site, and the volume, materials, and method of construction of the cistern (Li and Shi, 1999; Xiao et al., 1999). The study on the complementary irrigation techniques has been given more attention. The techniques mainly include the optimum irrigation period and the amount of water applied for wheat, corn, potato, pepper, and the

146

H. WANG ET AL.

microirrigation methods such as drip irrigation under mulching ﬁlm and ditch irrigation under mulching ﬁlm (Deng et al., 2000; Han et al., 1999; Yin et al., 2000; Xiao et al., 1999). From these experiments, it has been demonstrated that the combination of rainwater catchment and complementary water-saving irrigation at the critical stage of crop growth could increase both crop yield and water-use efﬁciency. This method is an effective way to solve water-shortage problems in semiarid and arid regions and to promote a sustainable development of dryland agriculture. The theoretical basis of rainwater catchment agriculture and its application need to be improved. Small-scale and scattered rainwater catchment projects have had no harmful environmental impacts, and have potential beneﬁts in protecting the environment. Good prospects exist for widespread rainwater utilization in regions with an effective annual precipitation of over 250 mm. In general, building reservoirs, digging wells, and interbasin water transfer have been considered as measures against drought. Understanding the importance of rainwater utilization and rainwater catchment projects is not enough to solve water shortages. Practical steps such as local rainwater resource utilization are important ways forward. At present, rainwater catchment projects have been developed only in regions of the Loess Plateau and in the Hebei and Henan provinces in the NCP. Close attention should be paid to rainwater utilization not only in the regions with low precipitation but also in those with plentiful rain. Speciﬁc studies on rainwater utilization have been conducted in the past, including rainwater catchment projects on rainwater collection, storage, puriﬁcation, and efﬁcient use. Integrated research including all these aspects needs to be undertaken in the future.

B. FIVE-WATER INTERACTION MECHANISM AND ITS APPLICATION 1. Five-Water Interaction Systems and SPAC Interface Processes Water movement in the Soil–Plant–Atmosphere Continuum (SPAC) regarding hydrological cycles includes interactions among ﬁve waters, which are atmospheric water, surface water, groundwater, soil water, and plant water. The issue of a “ﬁve states of water interaction” system was once proposed in 1988 (Liu and Ren, 1988), and the study on their interactions is called “ﬁve states of water transfer.” It is clear that the study of “ﬁve states of water transfer” has expanded the connotation of the SPAC study proposed by Philip (1966). The study on the interface process control of water ﬂuxes was considered as the key objective for water-saving agriculture through careful experimentation. Meticulous experimentation was conducted in recent years on this issue including

WATER-SAVING AGRICULTURE IN CHINA

147

the (1) the simulation of soil water movement and calculation of soil water by using a large-sized lysimeter, (2) studies on ﬁeld evapotranspiration regulations for water saving, (3) the interface processes of water ﬂuxes and water-saving regulation in SPAC, (4) development of an integrated model of water movement in SPAC. Three representative experimental stations of the Chinese Academy of Sciences (CAS) at Luancheng (Hebei Province), Nanpi (Hebei Province), and Yucheng (Shandong Province) with different conditions of buried groundwater tables, 26, 6, and 2 m, respectively, were selected to conduct the experiments and observations on the water and energy processes of groundwater, soil, crop, and atmosphere system at the same time. Several interfaces with different characteristics existed between soil, plant, and atmosphere. In the studies on water-saving agriculture in the North China Plain, taking soil water as the key factor, the analysis on the main interface processes, such as soil–root, plant–atmosphere, soil–atmosphere, soil–groundwater interfaces, was combined with soil water movement. The interface research will become the hot issue for water-saving agriculture research. The mass, energy, and information exchange actively on the interfaces, in which the most important parameter is water. All the water processes involved in water saving such as evaporation, inﬁltration, water uptake by the root system, and transpiration occurred on the interfaces of soil–groundwater, soil–atmosphere, plant–soil, and plant–atmosphere. The strengthening of the interface study could regulate the water ﬂuxes on the interfaces, which can give the theoretical basis for efﬁcient water use. 2. Conjunctive Use of Surface Water and Groundwater The conjunctive use of surface water and groundwater is a basic way of rationally exploiting water resources, and is favorable for controlling drought, waterlogging, and salinization, as well as increasing the replenishment of groundwater. During dry seasons, groundwater is pumped to supply water for crops, which could effectively regulate the groundwater table. The groundwater reservoir can store precipitation during rainy seasons, reduce the surface runoff, and increase the transfer ratio of rainfall to soil water and groundwater. For example, the Report of the Institute of Hydrology, Hebei Province (1985), reported that the groundwater table in the Heilonggang region was between 2.5 and 5.0 m below the surface before the rainy season. A 1-m reduction in the groundwater table could result in a decrease of surface runoff by 11–26 mm. The critical groundwater table, before the rainy season, should be controlled at 4.5–8.5 m. This varies with different hydrological years in the region. This was estimated based on the principle of the average discharge of groundwater for years not exceeding the average replenishment of groundwater. It was observed that in the lower reaches of the Yellow River, where the buried groundwater table was controlled to about 5 m before the rainy season, there would be no surface runoff to occur at daily rainfall rates of 150–200 mm

148

H. WANG ET AL.

so that precipitation could be fully captured (Jia, 1999). At the end of the ﬂood season, the shallow groundwater should be used for irrigation before wheat to suppress groundwater upward evaporation and increase the available groundwater. The data from the Nanpi Experimental Station, Hebei Province, showed that controlling the buried groundwater table at 5 and 2.5 m before and after the rainy season could effectively reduce upward groundwater evaporation. Wang et al. (1993) calculated that recoverable reserves of groundwater in the investigated area could be increased from 4.34 to 12.59 million m3 per year using such a scheme. Irrigation combining groundwater and surface water is a very effective way of conjunctively utilizing precipitation, surface water, and groundwater. In the NCP, groundwater was overdrawn in successive years, resulting in a series of environmental problems. The extraction of deep groundwater should be strictly controlled, especially in the groundwater depression cone areas. The extraction of shallow groundwater should be used instead so that the balance between groundwater replenishment and discharge can be maintained. For instance, in the areas of Shangqiu, Henan Province, the groundwater table dropped dramatically due to continuous pumping of groundwater to counteract droughts for several years. The practice of diverting Yellow River water for supplemental irrigation could effectively limit the groundwater depletion (Liu and Wang, 1995). A rise in the buried groundwater table occurred and resulted in the increased upward groundwater evaporation and consequently in secondary soil salinization. The irrigation method combining wells and canals could help to save water resources and to achieve the integrated mitigation of drought, ﬂooding, and control of salinity. The combined well and canal irrigation project could control groundwater and surface water through the regulating mechanism of the soil water reservoir, ﬁnally increasing the effective-use coefﬁcient of irrigation. Groundwater irrigation during the Spring in the surface water irrigation district of Linxi, Hebei Province, could increase the effective-use coefﬁcient of irrigation from 48 up to 69% as reported by the Ministry of Water Resources (1990a). The Jinghui irrigation district, Shaanxi Province, is another good example of the combined well and canal irrigation scheme, which uses water diverted from the Jinghe River. Before the surface irrigation scheme was introduced in 1932, the groundwater table in this region was at a buried depth of 15–30 m. The buried groundwater table was increased by 10–24 m until 1954, resulting in a secondary soil salinization threat. Groundwater irrigation was introduced at the beginning of the 1960s, and the area under combined well and canal irrigation reached 66,700 hm2 in the 1970s. The annual average groundwater extraction was 110–130 million m3, which effectively contained the increase in the groundwater table and secondary soil salinization. Associated with other agronomic measurements such as preventing leakage and leveling land, the irrigation water-use efﬁciency rose to 0.505; while comparable ﬁgures for the Ningxia, Shandong, and Henan surface water irrigation districts were only 0.21, 0.25, and

WATER-SAVING AGRICULTURE IN CHINA

149

0.40. The gross irrigation water under combined well and canal conditions was only 135–220 mm, much lower than the designed irrigation norm of 460–630 mm in the Henan and Shandong provinces (Jia, 1999).

C. UTILIZATION OF LOW-QUALITY WATER Low-quality water, such as city, industrial, and domestic sewage and brackish water, cannot meet the water-quality standard for agriculture. But such water would be a useful agricultural water resource after treatment, and its utilization would be a rational way to broaden sources of irrigation water. Sewage can be used to irrigate crops after puriﬁcation to reach a standard of water quality suitable for agriculture. The economic and environmental beneﬁts of treated sewage irrigation are evident. Efﬂuent irrigation can fully use the water and fertilizer resources in sewage. It is also beneﬁcial to agriculture that the organic content of the sewage can improve soil structure and fertility, and can reduce pollution loads. However, improper sewage use can result in crop yield reductions, soil degradation, disease, and ecological damage. Efforts have therefore increasingly been made to explore the most economically effective, technically practical, and energy-saving aspects of sewage disposal and recycling systems. Efﬂuent should be mainly applied to the economic crops rather than to vegetables, and to cereal crops only at the early stage of growth. The amount and timing of efﬂuent irrigation are determined by soil fertility and the stage and conditions of the crop. The amount of efﬂuent irrigation should be limited to crop water requirements in order to avoid deep drainage and surface runoff. According to statistics from the Ministry of Agriculture, the area of efﬂuent irrigation in China has reached 3.33 million hm2, and is mainly conﬁned to the northern arid regions (Liu and Wang, 1995). Attention should be paid to soil, groundwater pollution, and environmental sanitation issues during the development of efﬂuent irrigation. In China, the available brackish groundwater with a total soluble salt (TSS) content of 2–3 g/L is about 13 billion m3. There is 5.8 billion m3 of brackish groundwater with a TSS of 2–3 g/L in the NCP, but only 0.66 billion m3 was used in 1996. In addition, the area of saline groundwater with a TSS of 3–5 g/L in the NCP is up to 20,000 km2 (Zhang W, 1999). From the experimental results, the limiting value of soil solute concentration for crop use is 15–20 g/L, and the general irrigation water concentration should be less than 5–6 g/L (Zhang W, 1999). The rational utilization of brackish or saline water is of great signiﬁcance in alleviating water-shortage problems. Salt accumulation in the soil and the salt tolerance of crops at different growth stages must be considered when using brackish water. Mixing brackish and fresh water in irrigation, or alternating the use of the two, should be encouraged, and there have already been some successful examples of

150

H. WANG ET AL.

this. The mixing technique has been implemented in 2.53 million hm2 in the central and eastern parts of the Hebei plain. Given an applied ratio, 0.4, between brackish water and fresh water, some 12.8 × 108 m3 brackish water has been used, which is equivalent to increasing the average water amount per unit of cultivated land by 510 m3/hm2 (Wang et al., 1993). It must be stressed that irrigation with brackish water should use appropriate drainage facilities to ensure that the concentration of soil solutes does not exceed the crop’s physiological salt tolerance, especially during early crop growth stages. Factors such as salt content of irrigation water, crop species, salt tolerance, climate conditions, and irrigation methods should be carefully considered.

V. WATER-SAVING ENGINEERING MEASURES Water-saving engineering and technical measures mainly include water conveyance structures and water-saving irrigation techniques. The details are shown in Fig. 5.

Associated canal systems

Water conveyance works Water-saving engineering and technical measures

Lined canal (leak proof) Water conveyed through low pressure pipe Water conveyed through surface pipe

Watersaving irrigation techniques

Drip irrigation

Border irrigation with small plots

Sprinkler irrigation Irrigation on the film Seeping irrigation Intermittent irrigation Rational surface irrigation Ditch irrigation (fine flow)

Alternative ditch irrigation

Figure 5 The system of engineering and technical measures for water-saving agriculture.

151

WATER-SAVING AGRICULTURE IN CHINA

A. WATER CONVEYANCE STRUCTURES The traditional Chinese surface irrigation, especially ﬂood irrigation, wasted large volumes of irrigation water. According to statistics, the average irrigation water-use coefﬁcient currently in the Shandong Province is 0.5, and only 0.2–0.3 for ﬂood irrigation by gravity. The present water-use coefﬁcient of canal systems in the NCP is 0.37–0.50. The average irrigation water-use coefﬁcient for all the irrigation districts in China is about 0.45 (Wang, 1999), which demonstrates that over half of the irrigation water supplied will be lost during the conveyance due to leakage, evaporation, and poor operation. The water loss during water conveying is the major component of the total loss. The water conveying loss ratios (rc ) in China compared to other countries, which show a potential capacity in China’s water-saving agriculture, are given in Table I. In such a case, the water-saving capacity can be measured. Water-saving measures include associated canal systems, techniques for preventing canal seepage, and water conveyance through pipes. In many irrigation districts of China, the incompletion of associated canal systems and the ﬁeld water-use coefﬁcient are low due to poor management. For example, the completion ratio of associated canal systems at the second subsystem was ﬁnished by 50–60%, and that at the fourth subsystem only by 25% in the irrigation districts of the Haihe River basin. The canal system was old and not maintained for years; and from 40 to 70% of the canal head works and structures of the canal systems need to be repaired based on the investigation of 195 large-scale (over 20,000 hm2) irrigation districts throughout the country. Reconstruction of the old canal systems could increase the water-use coefﬁcient by 0.1 and could save up to 2 billion m3 of water annually (Xu, 1992). The technique for associating canal systems should be taken as an important step toward agricultural water saving. Techniques for preventing canal seepage are important in reducing water loss during the conveyance and the controlling of the groundwater table. Currently, the length of water-tight irrigation canals in China has reached 500,000 km, and the controlled ﬁeld area is 867 km2. The Ministry of Water Resources (1990b,c) has estimated that 75% of the water loss during conveyance could be saved through Table I The Ratio of Water Loss in the Water Conveying System and the Total Water Loss in China Compared to Other Countries (rc) China Region

Israel

United States

Japan

NCP area

Average

rc (%)

<10

22

39

55

50

152

H. WANG ET AL.

canal lining. In general, water-tight canals can increase the canal water-use coefﬁcient from 0.5 to 0.7. The water volume saved from irrigation projects can be divided into two parts. One is recoverable water including canal waste water, ﬁeld drainage, and groundwater replenishment, which can be reused in agriculture and can be considered as real water saving. Irrecoverable water, including reduced soil evaporation and evaporation from the canal water surface, cannot be reused (Shen et al., 2000). In recent years, the use of pipes instead of open canals for irrigation has become a trend in many countries, especially developed countries. Water delivered by pipes can effectively reduce seepage and evaporation during conveyance. It was widely used in the United States in the 1950s so as to increase the waterconveying efﬁciency up to 97%. The area of pipe irrigation was less than 1% in the former Soviet Union during the 1950s, and quickly rose to 63%, saving up to 30–40% of the water. At present, the area of pipe irrigation and drainage accounts for about 30% of the farmland in Japan. A study of pipe irrigation was begun in China in the mid-1960s. The Haihe Water Conservation Commission (1991) conducted a number of surveys. By the end of 1995, the irrigation area using pipes with low pressure has increased to over 3 million hm2, mainly in northern China in the Hebei, Shandong, and Henan provinces, and the cities of Beijing and Tianjing; in recent years it has also been developed in Shanxi, Inner Mongolia, Gansu, and Xinjiang (Li et al., 1999). The pipe water conveyance coefﬁcient is now over 0.9 and could save water by up to 50, 15, and 7% compared with an Earth canal, a canal lined with stone, and a canal lined with concrete in irrigation districts using groundwater; and it could save some 40% in irrigation districts using surface water. Compared to an Earth canal, pipe water conveyance has several advantages including saving water, reliable water supply, saving land taken up by canals, and saving labor. Over the last decade, this irrigation method has been rapidly taken up in China. The groundwater irrigation area in northern China is about 1.1 × 107 hm2, in which pipe water conveyance irrigation is less than one-third. Technical standards, project quality, and administrative loads directly affect the speed at which this water-saving technique is adopted, and should be given much more attention in future studies. The pipe water conveyance technique is more complex in surface water irrigation districts.

B. WATER-SAVING IRRIGATION TECHNIQUES Traditional surface ditch irrigation is wasteful of water, and the amount of water in northern China, applied is one to two times the actual crop water requirement (Lin and Zhao, 1999). The introduction of modern irrigation techniques is a necessity. Experiments on sprinkle irrigation began in the late 1950s in China, and the drip technique was introduced in 1974 (Fu, 1989; Liu and Mou, 1995). After

WATER-SAVING AGRICULTURE IN CHINA

153

several decades of experiment and popularization, the total area of sprinkle and drip irrigation currently stands at about 66.67 thousand hm2, only 1.7% of the total irrigated land available to economic activity (Feng, 1999). The ratio of consumed water use for surface irrigation, sprinkling, and drip irrigation is 1:0.5:0.3. Besides its water-saving effect, the use of sprinkling and drip irrigation increased grain yield by about 10–30% (Fu, 1989). The effects of such advanced irrigation techniques on water saving and yield are considerable, but the monetary investment is high. Due to economic limitations, such expensive irrigation techniques are only suitable for some horticultural cash crops and for high-lift irrigation districts, sloping land, and serious water-deﬁcit regions; but it is not suitable for ﬁeld crops and ﬂat areas. Local natural and economic conditions should be considered when applying sprinkling and drip irrigation techniques. It should be emphasized that surface irrigation does not mean just ﬂood irrigation; similarly, water-saving irrigation does not include just sprinkle and drip irrigation. Improved surface irrigation also could reduce water loss. Some 60% of irrigation in the United States is surface irrigation, and new technology such as leveling land-using lasers, surge irrigation, and pipe irrigation could successfully save water (Jia, 1999). In China, surface irrigation accounts for 98%, and a number of advanced surface irrigation methods have already developed (see Fig. 3). Improved border check irrigation is using small plots by changing ridges from long to short, wide to narrow, i.e., large to small plots. The suitable widths are 2–3 m and 1–2 m for gravity irrigation and pumping irrigation, and the suitable lengths are 30–50 m and 30 m (Lin and Zhao, 1999). The irrigation norm increases with the increase in the length of the ridges. Border check irrigation of small plots can reduce irrigation water volumes, but raise irrigation uniformity up to over 80%. This technique can also control deep drainage to prevent the rise of groundwater tables and soil salinization. From the measurement of soil water content at 0–200 cm before and after irrigation, no deep drainage occurred when the plot ridge length was 30–50 m, while deep drainage accounted for 30% of irrigation water when the length was 200–300 m (Lin and Zhao, 1999). The experimental data from the Yucheng station, Shandong Province, showed that border check irrigation with small plots, 20 m long and 3 m wide, could have the maximum savings in the cost of water and the minimum irrigation amounts (Sun et al., 1992). Surge irrigation or intermittent irrigation has been found to greatly improve the efﬁciency of surface ditch or border irrigation by raising water conveyance. It supplies water intermittently to ﬁeld irrigation ditches or plots at speciﬁed time intervals. It can save up to 10–40% of water compared to ordinary gravity irrigation due to the reduction of deep drainage and tail water loss. Intermittent ditch irrigation and intermittent border irrigation could save 38 and 26% of water, respectively (Jia et al., 1994). Soil evaporation under a crop canopy can be effectively reduced by using plastic ﬁlms. Based on ﬁlm-covering cultivation, new irrigation methods have been

154

H. WANG ET AL.

developed over the last decade in Xinjiang, including irrigation beside, on, and under the ﬁlms. Irrigation on ﬁlm is also called ﬁlm hole irrigation; i.e., irrigation water ﬂows onto the ﬁlm and through the holes on the ﬁlm seeps into the soil of surrounding the crop roots (its rhizosphere). Currently the area of irrigation on ﬁlm is up to 0.2 million hm2 in Xinjiang, and it has quickly spread to the Gansu, Ningxia, Henan, and Hebei provinces (Lin and Zhao, 1999). Under the wheat–cotton intercropping practice in the Shangqiu, Henan Province, 30 and 50% of water were saved by wheat and cotton, respectively, through irrigation on ﬁlm (Si et al., 1992). The Institute of Field Irrigation of the Chinese Academy of Agricultural Sciences conducted an experiment on alternative ditch irrigation in western Henan in 1985. The results showed that the irrigation quotas for alternative ditch irrigation and irrigation in every ditch (the control) were 147 and 294 mm, respectively, saving 50% of the water, although yields were reduced slightly by some 14%. Alternative ditch irrigation suits growth periods with low water requirements, regions with high groundwater tables and wide–narrow-row or wide-row cropping. Problems of large-sized irrigation plots and uneven land commonly exist in the surface water irrigation areas of the Yellow River basin, which causes large amounts of water to be wasted due to a long duration of irrigation, high irrigation quotas, low uniformity of irrigation, and high evaporation rates. Normally the irrigation plot is 667–2667 m2, and height differences within the plots are over 5 cm. Corresponding ﬁgures for current advanced levels for the plot and height differences are 133–167 m2, and less than 8 mm. Given the same plot width, an irrigation quota for a 300-m-long plot is doubled compared to one that is 100 m long.

VI. WATER-SAVING AGRONOMIC PRACTICES Since about 50% of irrigation water is consumed in the ﬁeld, a study on watersaving issues such as reducing soil evaporation and luxury crop transpiration is essential for increasing crop water-use efﬁciency. Such a study is a very important aspect of water-saving agriculture, and also has the potential for the development of water-saving agriculture in the future. Increasing the crop water-use efﬁciency needs substantial support from a number of agronomic measures (see Fig. 6).

A. WATER-MATCHED PRODUCTION 1. Water-Matched Agricultural Structure Different agricultural structures differ in their water requirements, so agricultural structures should be arranged and adjusted after a careful consideration of local water resources conditions. According to statistics from Liu (1989), the outputs

WATER-SAVING AGRICULTURE IN CHINA

Water-matched production

155

Water-matched agricultural structure Water-matched planting

Water-saving biological basis Biological water saving technology Water-saving breeding

Limited irrigation

Water-saving agronomic measures

Regulated deficit irrigation Water-saving irrigation scheme Controlled alternative irrigation

Mulching technique Soil moisture conservancy Tillage measure

Fertilizer application

Figure 6 System of water-saving agronomic measures.

of planting, sideline, husbandry, forestry, and ﬁshery accounted for 68.9, 19.5, 9.4, 1.6, and 0.6% of the total agricultural output, respectively. From the point of view of water consumption, the proportion of planting in agriculture was high, but those of forestry, husbandry, and ﬁshery were too low. Nevertheless, forestry and husbandry are suitable for developing low-water-consumption agriculture because trees and forage grasses (herbage) normally do not need irrigation except for fruit trees and young nursery plants. Therefore, the agricultural structure should be rationally adjusted in accordance with water resources and natural conditions. The ratio of water-consuming planting should be cut down and properly replaced by forestry and husbandry. A diversiﬁed agricultural structure can help counter natural hazards such as ﬂooding and drought. 2. Water-Matched Planting An optimal cropping system is developed according to the temporal and spatial characteristics of precipitation, groundwater resources, and current status of

156

H. WANG ET AL.

irrigation works. Crop varieties that match crop water requirements to rainfall distribution will have a higher drought resistance and water-use efﬁciency. Selective crop planting can increase the use efﬁciency of local water resources. The coupling of crop water consumption with effective precipitation varies among crop breeds. The scientiﬁc basis for water-matched planting is to match crop water requirements with precipitation. It is estimated that the effective precipitation could meet 55–82% of the wheat water requirement in a normal water year to the south of the Yellow River; however, the ﬁgure is only 29–37% to the north. Effective precipitation during the cotton growing season could provide 65–74% of water consumption, and the effective rainfall for corn could basically meet crop water use. There is no big difference between the south and the north of the Yellow River (Xu, 1992). Crops with the characteristics of high rainwater-use efﬁciency should be recommended in order to replace high waterconsuming crops according to the principle of the water-matched agriculture (Liu, 1989). Cropping systems in most areas of the northern China did not make the best use of the available water and heat energy. The results from the Nanpi Experimental Station in the Heilonggang region of Hebei Province showed that multiple crop indices had negative relationships with the irrigation rate. Water-use efﬁciency of 2-year three crops was maximum, while in the system with 1-year two crops there was a shortage of water and heat, especially during a dry year. To reduce the multiple crop index to suit the local water resources, adversity-resistance crops (such as millet, sorghum, and potato) or soil-improving crops such as soybean should be planted, which could save water at the rate of 1200 m3/hm2 (Liu and Wang, 1992). Adjusting the crop sowing dates can help crop water consumption during the growing season to match the precipitation pattern. This is an effective way to reduce drought impacts and increase the efﬁcient use of rainwater. An experiment at the Nanpi Experimental Station, Hebei Province, showed that the grain yield of spring corn could increase by 30% due to late sowing in the middle and end of May compared to that sowing in the middle and end of April. The water-use efﬁciency was increased by 29% (Wang et al., 1993).

B. BIOLOGICAL WATER-SAVING TECHNOLOGY Biological water-saving technology is of particular importance in water-saving agriculture due to its great potential. The dominant position of water-saving engineering at present will be replaced by biological water-saving technology in the 21st century. Innovation of biological water-saving technology and an everincreasing water production efﬁciency will become the lasting forces for the sustainable development of water-saving agriculture.

WATER-SAVING AGRICULTURE IN CHINA

157

1. Biological Basis for Water Saving More and more attention has been paid in recent years to the study of the physiological mechanism of crop water stress. An important issue is the crop root regulating signal in response to soil stress and its role in the optimization of water use. Crop root is able to monitor available soil moisture and can synthesize some biologically active substances when soil drying. The substances are then transferred to the shoot transmitting the information of adversity to regulate the inner physiological biochemical processes to keep the water balance within the crop body. Root-synthesized abscisic acid (ABA), a water stress hormone, can be used as a root signal in the root-to-shoot communication of soil drying, and assumes a dominant role in the information regulation on the optimization of crop water use (Li and Wang, 1994). Experiments using pea showed that the content of endogenous ABA in leaf, leaf epidermis, and root increased as soil water content decreased. The ABA content at the root tip was negatively correlated with leaf conductance and it could regulate stomata behavior (Wu et al., 1993). This physiological root mechanism provides the theoretical basis for modern watersaving irrigation methods such as regulated deﬁcit irrigation (RDI) and controlled alternative irrigation (CAI). 2. Water-Saving Breeding Developing the drought-resistant, drought-tolerant, water-saving varieties with a higher WUE through genetic engineering is a challenge. Modern water-saving breeding is based on the adaptability of crops to drought environment and aims at increasing crop yield and water-use efﬁciency through biological functions. The responses of different crops or varieties to water stress differ signiﬁcantly and can result in the difference in WUE by 30% among varieties (Shan and Zhang, 1999). Water-use efﬁciency can be considered as a crop characteristic in terms of yield and drought resistance, and it is heritable. Hence breeding a new variety of drought tolerance or resistance could achieve the aim of water saving. Due to the differences in the CO2-assimilation process, the water-use efﬁciency of C4 crops is 2.5–3.0 times that of C3 crops. Currently there has been an attempt to change the carbon metabolic pathway of C3 crops using biological techniques. In modern drought-resistant breeding, it is essential to develop and improve a simple and quick screening method for identifying drought-resistant varieties. A lot of work has been done in this area (Condon et al., 1987; Farquhar and Richards, 1984; Nageswara Rao and Wright, 1994; Wright et al., 1994). The most important result was that 13C-isotope discrimination () was apparently negatively related with water-use efﬁciency, which means has the potential to be a screening factor for drought-resistant breeding. However, application was limited because

158

H. WANG ET AL.

measurements of need an expensive mass spectrometer. It was veriﬁed from the experiment that the speciﬁc leaf area (SLA) was closely related with and that the measurement of SLA is simple and costs very little. The development of the 13 C-isotope discrimination theory provides a new thought process and a method for drought-resistant breeding.

C. WATER-SAVING IRRIGATION SCHEMES 1. Relationships of Crop Yield and Applied Irrigation Water A water-saving irrigation scheme coupled with crop water requirements is an important aspect of water-saving agriculture. Yield, WUE, and economic proﬁt should be considered simultaneously for developing a water-saving irrigation scheme. Crop yield has nonlinear relationships with irrigation water supply. At low levels of irrigation input, the crop yield increased quickly with irrigation water; but once irrigation water exceeded a certain volume, crop yield increased slowly, and even decreased when water was added. The optimum irrigation quota could be achieved based on marginal analysis. Taking Luancheng Station at the piedmont of the Taihang Mountains as an example, the measured data were calculated through marginal analysis to obtain the optimum irrigation water for winter wheat (202 mm). The results showed that winter wheat needs to be irrigated three times during a growing season to achieve both yield and WUE at relatively high levels. Comparing the optimum irrigation quota with the irrigation quota of 343 mm for the maximum yield, water could be saved up to 41%, but with only an 8% reduction in yield (Yuan et al., 1992). The determination of the optimum irrigation quota is of particular importance in water-deﬁcit regions. A ﬁxed amount of irrigation water can either be applied to maximize the yield from a small area or to increase the yield from a larger area. In the latter case, the overall yield can be greater than that for the same area with only a small portion irrigated. The crop water requirement differs during different growing stages, and matching the degree of water requirement and precipitation available also varies with time. Crop yield depends not only on the amount of irrigation water but also on its temporal distribution. Under soil water-deﬁcit conditions, the judicious application of irrigation water can lead to a relatively high yield and WUE. The methods for determining optimum irrigation schedules include a yield-reducing coefﬁcient method, marginal analysis, and dynamic planning. The results from the Nanpi Experimental Station in the Heilonggang regions, Hebei Province, showed that WUE reached a maximum value of 11.1 kg/ha/mm when irrigated three times. The best irrigation practice was irrigating at turn-green when irrigated once, at turn-green and booting when irrigated twice, and stem elongation, booting, and grain ﬁlling when irrigated three times.

WATER-SAVING AGRICULTURE IN CHINA

159

Crop sensitivity to water stress differs during different stages, and the most sensitive stage is called the crop water critical stage. The sensitive indices for different stages (λi ) could be calculated from measured data through multielement regression analysis. The period from jointing to heading is identiﬁed as the critical water requirement stage of winter wheat, and irrigation at this stage can obtain the optimum yield. Water management at this critical stage is an important aspect for water saving and yield increasing (Yuan et al., 1992; Zhang and Liu, 1992). 2. Limited Irrigation Studies on the effects of limited water deﬁcit and limited irrigation on grain yield have been developed recently to enrich the contents of water-saving irrigation. The traditional way was to irrigate the ﬁelds thoroughly and to aim at high yield. Under water-shortage conditions, it is necessary to implement inadequate irrigation. The theoretical basis for inadequate irrigation includes crop–water relation, impacts of the water deﬁcit on crop growth at different stages, and the physiological drought resistance of crops. Xu (1991) reported that the water deﬁcit did not always reduce the crop yield, and a moderate water deﬁcit at an early growing stage would be beneﬁcial for some crops (Shan and Zhang, 1999). In semiarid areas, the sunﬂower yield under a slight water deﬁcit was increased by 50% compared to that under frequent irrigation (Xu, 1991). Under certain conditions, a moderate water deﬁcit did not affect crop yield, but increased WUE. Leaf WUE of spring wheat increased under slight to moderate water-deﬁcit conditions. Millet has a high drought-resistant ability, and apparently its leaf WUE was not reduced with a water supply of only 30% of the water requirement. It was also proved that the impacts of the water deﬁcit differed from various physiological processes with an order of growth—transpiration—photosynthesis—matter conveyance. Water deﬁcits at vigorous growing stages and pollination periods should be avoided, while crops could endure drought to some extent without serious yield reduction during the stage of seedling or the late period of crop growth. Waterdeﬁcit treatment at early growing stages could restrain the growth of seedlings, but improve root development, which is helpful for crops which are enduring a more serious drought during a late growth period. A moderate water deﬁcit at the grain ﬁlling stage could improve the ﬁlling process to increase the crop harvest index. 3. Regulated Deficit Irrigation Regulated deﬁcit irrigation (RDI) is a new irrigation method proposed in the middle of the 1970s. The basic concept of RDI differs from the traditional waterabundant and high-yield irrigation, or limited water irrigation. The theoretical bases of RDI are crop physiology and biochemistry. The active regulated deﬁcit is conducted on crops according to their characteristics and water requirements.

160

H. WANG ET AL.

Certain water stresses are imposed initially to some crop growth stages, which can change intrinsic plant physiological and biochemical processes, regulate the distribution of photosynthesis products to different tissue organs, and control the growth dynamic between the aerial part and the root to improve reproductive growth and to eventually increase crop yield. RDI can reduce nonproductive soil evaporation and can restrain transpiration during the water stress period. There is no clear reduction in the photosynthesis rate during water stress, but there is the super-compensating effect on photosynthesis after rewatering, which is beneﬁcial to photosynthetic products which transform, and distribute to seed. The effect of drought on crops follows a process from “adaptation” to “injury.” The compensating effects on physiology, water use, and growth will be produced after rewatering if the water deﬁcit is limited within the adaptation process. A timely moderate water deﬁcit can restrain vegetation growth and increase the root/shoot ratio to raise root conductivity, water absorption, and drought resistance. RDI is an effective way for regulating water–soil–plant–environment interactions. RDI was ﬁrst conducted on fruit trees and quickly became an important issue in horticulture; but RDI experiments on grain crops are still in an initial stage (Guo et al., 1999; Jia et al., 1999). The indicators for RDI include the optimum degree, stages, and duration of regulating the water deﬁcit. The experiment of RDI on corn (1997–1998) in the Huoquan irrigation district, Shanxi Province, showed that crop water requirements at stages of seedling, jointing, and tasselling were decreased due to the reductions in leaf area, soil evaporation, and leaf transpiration rate when the regulating water deﬁcit was conducted at the corn seedling stage, while the crop water requirement at the grain ﬁlling stage was increased because of the retarding of leaf senescence to keep the relatively high photosynthesis rate and harvest index. In the end, RDI can increase crop water production efﬁciency without a large reduction in crop yield (Guo et al., 1999). 4. Controlled Alternative Irrigation According to the relationship among crop photosynthesis, transpiration, and stomatal aperture, and the physiological function for raising WUE, a new ﬁeld water-saving regulation approach called Controlled Alternative Irrigation (CAI) was proposed (Davies and Zhang, 1991; Kang et al., 1997). The principle of CAI is that crop root can produce signals during water stress, and the signals can be transmitted to leaf stomata to control their aperture at optimum levels. Rootsynthesized ABA, as a root signal of water stress, is continuously provided for crop leaf in order to regulate the stomatal statue to avoid luxury transpiration. Partial irrigation can control soil moisture at the root zone vertically and horizontally so that part of the crop roots grows in the drying soil. Drying and wetting occur alternately, and roots in different soil zones experience drought hardening, which in turn, can increase root conductivity.

WATER-SAVING AGRICULTURE IN CHINA

161

The root conductivity experiment conducted at the greenhouse of Hong Kong Baptist University showed that root conductivities at rewatering after a period of water stress for both corn and sunﬂower were higher than those with a full water supply (Kang et al., 1997). CAI can regulate the crop water relation and increase the WUE with no reduction in photosynthesis product accumulation. If combined, CAI techniques and agronomic tillage measures can reduce ineffective water loss of soil evaporation, and rainfall can be collected for root use, which is of particular importance in water-shortage regions.

D. SOIL MOISTURE CONSERVATION Rainwater or irrigation water in the ﬁeld will mostly turn into soil water, and only a small part will recharge groundwater systems. Water evaporated from soil under a crop canopy can be controlled and transformed into productive crop use. Therefore reducing soil evaporation under the crop canopy is an important component of water-saving agriculture. Measures adopted in ﬁeld water-saving practices generally include mulching, water conservation tillage, and chemical water-saving measures, in which the effects of straw mulching on saving water and increasing yield were most evident. 1. Field Mulching Techniques Field mulching, which is an effective measure for increasing the efﬁciency of water use, can reduce nonproductive water loss in the ﬁelds through the artiﬁcial regulation of water conditions between soil surface and crops. Mulching has already been applied to areas of 3.33 million hm2 in 1991 in China, according to a report from the Haihe Water Conservancy Commission (Newsletters of Haihe Hydraulic Commission, Ministry of Water Resources, 1991). Gravel mulching is an old water conservation technique which can improve the soil water conservation capacity and has already been widely used in the arid regions of northwestern China. Plastic ﬁlm mulching was also successfully applied to reduce soil evaporation and conserve water in soil for crop use. The results of the plastic ﬁlm mulching experiment at the Yucheng Station, Shandong Province, showed that soil evaporation under canopy was reduced by 40–60% (Zhang et al., 1992). However, plastic ﬁlm mulching was not widely applied; it is only used for vegetable and commercial crops due to the high cost and soil pollution from the residual ﬁlm. The light-decomposed plastic ﬁlm technique was then developed, and the results of the experiment in cotton ﬁelds using light-decomposed plastic ﬁlm at the Nanpi experimental station showed that WUE reached 9.39 kg/ha/mm, an increase of 73% compared with the control and soil evaporation which was reduced by 50–72% (Wang et al., 1993). Soil moisture at 0–60 cm in a corn ﬁeld with wheat straw mulching could increase by 14–27% in the Heilonggang regions of Hebei Province

162

H. WANG ET AL.

(Xing and Wang, 1992). Straw mulching has also been widely used, and offers a solution to the straw reutilization problem and prevents environmental pollution due to straw burning, which is a traditional treatment used by farmers in the region. 2. Soil Tillage Measures Tillage is a traditional soil water conservation technique and is an important part of agricultural water saving. Different water and heat regimes between the tilled and untilled soils resulted in the effects of soil moisture conserving. The difference in soil porosity between tilled and untilled soils is about 10%, and the difference in inﬁltration is about 20%. Experimental results from the Nanpi Experimental Station showed that tillage could reduce surface runoff by 12–25% and soil surface evaporation by 20% (Wang et al., 1993). Deep tillage can enlarge soil water-storage capacity. The soil with deep tillage could store over 90% of rainwater in the autumn, and the upper-layer soil water was increased by 10.2–12.0% after summer tillage for 2 years, as a report from the Haihe Water Conservancy Commission indicates (Newsletters of Haihe Hydraulic Commission, Ministry of Water Resources, 1991). Harrowing against frozen soil in early spring and the protective minimum or zero tillage in the northern agricultural areas could prevent soil evaporation to conserve soil moisture to a certain degree.

E. SOIL FERTILITY It has been shown that balanced fertilizer application is an important way to increase ﬁeld WUE. Crop stomatal regulation, water-holding capacity, membrane permeability, and photosynthesis are closely linked to nitrogen (N), phosphorus (P), and potassium (K) application. Stomatal conductance and transpiration rate at high and moderate N were lower than those at low N under water stress. N application could increase the leaf net photosynthesis rate under light or moderate water stress, while no difference in the leaf net photosynthesis rate existed at severe water stress (Shen et al., 2000). Increasing fertilizer application to improve soil fertility, i.e., to increase the content of soil organic matter, can improve soil physical properties and build a high and efﬁcient soil water reservoir. The soil water storage capacity can be increased after organic fertilizer application, and the inﬁltration capacity could increase by 40–60% when the content of soil organic matter is increased from 1 to 1.5%. Increasing fertilizer application could also improve crop root growth and its drought resistance so as to fully use soil water stored in the deep soil layer. However, excessive fertilizer application could decrease the fertilizer-use efﬁciency and result in pollution. Therefore the study of the interactions between water and fertilizer to develop an optimum ratio of water and fertilizer application

2

WUE (kg/hm /mm)

WATER-SAVING AGRICULTURE IN CHINA

163

12.0 8.0 4.0 0.0 0

1

2

3

4

5

6

Number of Irrigations High Figure 7

Moderate

Low

Water-use efﬁciency under different levels of soil fertility (high/moderate/low).

is of practical signiﬁcance, especially for the NCP regions where the development of agricultural production is limited by water resources. The experimental results from the Hengshui Agricultural Research Institute, Hebei Academy of Agricultural Sciences, showed that crop yield had a negative linear relationship with irrigation water under different soil fertility (see Fig. 7). It is clear that WUEs with high soil fertility are much higher than those with low soil fertility. For a certain soil fertility, WUE is nonlinearly related with applied irrigation water. The number of irrigations at the maximum WUE are once, twice, and three times, respectively, for low, moderate, and high soil fertility (Xi et al., 1986). Sufﬁcient water conditions should be associated with higher soil fertility so as to gain the maximum efﬁciency of water resources.

VII. WATER-SAVING MANAGEMENT In general, better management can save 50% of irrigation water because the management could ensure a successful implementation of other water-saving measures. The water-shortage problem could not be solved solely based on scientiﬁc understanding. Economic and administrative measures are necessary to ensure that sound water management is in place.

A. ADMINISTRATIVE SYSTEMS The scattered administrative system of water resources management was carried out for years in China. Catchment management is the prerequisite both for raising the beneﬁt of a water resource integrated development and utilization and for

164

H. WANG ET AL.

realizing water saving. A joint water management system involving the catchment administration and a regional regulation need to be set up, which can coordinate several complex relations between the upper reaches and the lower reaches, watersupply area and water-use area, utilization and protection of water resources, and agricultural sector and urban sector. At present, the managed water conservation works are mainly run by the states, and general maintenance, replacements, and technical innovations of the works are also subsidized by the states in the irrigation districts of China. The current water management institutions are rather inefﬁcient and cannot meet the needs of the market economy and the water-saving development. It is imperative to reform the present irrational water management institution. The enterprise-type water administrative organization should be set up in the surface water irrigation districts to establish nonproﬁt water-supply company which has autonomous management functions and responsibilities. Consequently it can aim at maintaining its own water project. In the groundwater irrigation districts, the water-use association by a village should be formed to tighten up the planned regulation on the exploitation and utilization of groundwater so as to reach its replenishment–discharge balance for sustainable use. In superintendence and coordination with the state management institution, the farmers’ own organization can arouse people’s water-saving initiative, so that farmers will commit the obligation to water resources management willingly.

B. WATER PRICES AND FEES Water resources were used almost freely or charged at a very low price in China. This has caused the national overburden to water conservancy facilities and serious waste of water resources. There was no proper system for collecting water fees in China before 1985. In the lower reaches of the Yellow River, water from the river was diverted for irrigation freely before 1979, and a water price was introduced (e.g., 0.05 Chinese Yuan/m3), which is only 1/4–1/3 of the water-supplying cost. Obviously, water resources could be used more efﬁciently if the commodity sense of water resources was adopted. At present, a low water fee is disadvantageous to the implement of water-saving measures. An appropriate system for collecting water fees should be established. Farmer participatory water management should be realized for water fee collection and water utilization. In principle, the criterion for collecting a water fee should be based on the balance between the water fee and the cost in order to maintain and run water conservancy projects. The basis for setting a water price should be the cost of the water supply. The water price determined by the local governments is only 0.01–0.02 Chinese Yuan/m3 water, too low compared to the cost. Water resources wasting is inevitable under a low water fee policy. Regulation of the water price policy is of great importance for

WATER-SAVING AGRICULTURE IN CHINA

165

pushing forward water saving in surface water irrigation districts, especially for gravity irrigation areas. The water price can vary according to different regions and different seasons. The uniﬁed water price adopted both for surface water and for groundwater is the prerequisite for the conjunctive use and management of surface water and groundwater. During the reforms of water prices, the fee subsidy policy should be made according to the local conditions considering the current income standard of the farmers.

C. LAWS AND REGULATIONS A good management system must have the support of the government through laws and policy. Further improvement needs to be made to the existing water laws, such as the “water law,” the “law on water pollution,” the “law on water and soil conservation,” and the “law and ﬂood prevention and control.” For a catchment, the law on catchment management should be considered to ensure that an integrated catchment management system be put into effect. In groundwater irrigation areas, a single-well and quantity-limited withdrawing system has been set up in terms of the “withdrawal permission license” to monitor the use of groundwater. Laws and regulations for managing and protecting groundwater environment should be set up and improved to effectively supervise and control groundwater tables, protect groundwater resources, and prevent land subsidence. An implementation of engineering and technical measures of agricultural water saving requires investment inputs and policy support to ensure the rational development and utilization of water resources. Water pollution damages the ecoenvironment and intensiﬁes the imbalance between water supply and demand. It is necessary to conscientiously protect water resources, to control water pollution, to increase the ratio of sewage disposal, and gradually to achieve sewage recycling. Laws concerned with water pollution and its prevention and control should be made.

D. TECHNICAL TRAINING AND SERVICES In light of local conditions, to make great efforts in spreading available scientiﬁc achievements and advanced practical techniques is one of the actions of good water management. Technical training on water-saving agricultural management has been enforced in the nation so as to raise the technical level and understanding of administrative personnel and farmers. A socialized water-saving service system is a very important part of the development of water-saving industrialization. There has been a big water-saving market in China in recent years, and this is essential

166

H. WANG ET AL.

for improving the socialized water-saving service systems. The main contents of such service systems include the technical consultation and technique services, the technical training, and the services on materials and equipment.

E. EXPERT SYSTEM FOR WATER-SAVING AGRICULTURE Expert systems of water saving have recently been developed in China. The expert system on regional water resources planning and water-saving and yieldincreasing irrigation in irrigation districts has just been developed (Xu et al., 2000). This expert system includes ﬁve subsystems: water resources calculation and planning, water resources numerical simulation, agricultural water use and design of an irrigation and drainage scheme, choice of irrigation techniques, and irrigation decision-making. Another is an expert system of irrigation forecast and decisionmaking for water saving, which was developed by Northwest Sci-Tech University of Agriculture & Forestry (Wang et al., 2000). The users only need to input local meteorological data, soil type, crop variety, and initial soil water content. Through computer inference, it can work out irrigation time, applying the optimum amount of irrigation water, the cost and yield-increasing proﬁt of this irrigation, and the most economical irrigation scheme, which could guide farmers to carry out scientiﬁc water-saving irrigation decisions.

F. WATER-SAVING PRODUCTS AND MARKETING The poor quality of water-saving products, equipment characteristics, and their post-sell, maintaining services greatly hinder the development of water-saving. Many existing water-saving irrigation projects could not normally be used and will even be abandoned due to the substandard water-saving irrigation equipment adopted. To ensure the development of water saving, high-quality water-saving equipment and technical products are now beginning to be developed, and the market competition mechanism has been introduced into the water-saving technical innovation in China. Water is a special commodity. Water enterprise ﬁtted in with the marketing economy running should be founded for the development, management, circulation, and exchange of water resources. The cruxes of the water market are to rear the water commodity consciousness of farmers, to use water resources nongratuitously in agriculture, to calculate water fees using the amount of water being used, and to form a rational water price system. Another market on water-saving technology is also required. The farmers’ demand for water-saving techniques should be stimulated so that farmers are willing to use and afford them. It is necessary to encourage business to invest in water-saving technology, equipment

WATER-SAVING AGRICULTURE IN CHINA

167

development, and production. The water resources market is of utmost importance, so there will be no water-saving technology market and consequently no sustainability in developing water-saving agriculture without the establishment of a water resources market.

VIII. CONCLUDING REMARKS Looking into the 21st century, China’s population is expected to expand to 1.6 billion, and food supply will be a serious challenge for national agriculture production. In order to ensure food security and to increase agricultural yields, the water shortage problem must be dealt with urgently. There are two solutions to the agricultural water shortage problem in China: diverting water from water-rich areas through water transfer schemes and developing water-saving agriculture. Compared with water diversion, which requires more investment with higher costs and a longer period for implementation, the development of water-saving agriculture is an attractive alternative. Therefore, water-saving agriculture should be the ﬁrst priority with water transfer projects to follow. Water-saving agriculture will provide a good basis on which the magnitude of water transfer can be optimized to minimize cost and to maximize the beneﬁt of the projects. Sociologically, water saving would ameliorate the conﬂicts between the water-rich areas and the regions to where the water is diverted. In other words, water saving is not only essential for solving water shortage problems but also socioeconomically and ecoenvironmentally beneﬁcial. The management of water resources is an extremely important issue. There is a pressing need to move away from the past practices of unplanned irrigation infrastructure and move towards the consideration of water management. Water resources management involves many aspects, such as policy, legislation, institutional, organizational, personnel, ﬁnancial, and operational. Generally speaking, the sources of water applied to crops should come from ﬁrst, the use of rain water; second, from surface water sources; third, from groundwater sources. This consecutive context allows locally available water resources to be used more efﬁciently. So-called water interactions regarding water-saving agriculture in the ﬁeld imply the transformation of hydrological states including ﬁve statuses of water in air, surface, soil, aquifer, and plant. An important concept that this review has presented is the study of the interfacial processes controlling the hydrological states. Physiologically, water-saving agriculture is mainly identiﬁed by high water use efﬁciency. As a result, crop growth models can be used to theoretically evaluate water-saving agriculture. Biological water-saving technology is of particular

168

H. WANG ET AL.

importance in water-saving agriculture. The current dominant position of watersaving engineering will be replaced by a shift to biological water-saving technology. The innovation of biological water-saving technology will become the lasting motivation for the sustainable development of water-saving agriculture. This paper provides a comprehensive overview of the water-saving research and practices conducted in China during the last 2 decades, and highlights a number of achievements. However, there is a need for further research in this area, in order to fully realize the potential of water-saving agriculture in China. The following points need to be addressed in order to improve the effectiveness of water-saving agriculture: 1. To enhance and to work out integration of biological and engineering measures for water-saving agriculture 2. To more widely deepen scientiﬁc knowledge, and transfer the water-saving techniques to farmers 3. To establish marketing mechanisms for water-saving agriculture and to promote commercialization of water-saving technology 4. To raise awareness of the importance of water saving through improved training and education.

ACKNOWLEDGMENTS This review was supported by NSFC (National Natural Science Foundation of China) projects (49801003 and 49871020) and the Major State Basic Research Development Program of China (G1999043605). We would like to thank Andrew Bell and Mat Gilfedder for improving the English in this paper.

REFERENCES Bos, M. G., and Nugteren, J. (1974). “On Irrigation Efﬁciencies.” Publ. No. 19. International Institute for Land Reclamation and Improvement, Wageningen. Brown, L. R., and Halweil, B. (1998). China’s water shortage could shake world food security. World Watch 11, 10–21. Chinese Academy of Agricultural Sciences (Ed.) (1985). “Regional control and agricultural development in the Huang-Huai-Hai Plain.” Agricultural Science and Technology Press of China, Beijing. Condon, A. G., Richards, R. A., and Farquhar, G. D. (1987). Carbon isotope discrimination is positively correlated with grain yield and dry matter production in ﬁeld-grown wheat. Crop Sci. 27, 996–1001. Davies, W. J., and Zhang, J. (1991). Root signals and the regulation of growth and development of plants in drying soil. Annu. Revi. Plant Physiol. Plant Mol. Biol. 42, 55–76. Deng, G., Zhang, Y., Wang, P., Li, S., and Zhou, H. (2000). Effect of irrigation with catchment of rainfall on yield and water use efﬁciency of ﬁlm-mulched spring wheat. Agr. Res. Arid Areas 18(1), 91–94.

WATER-SAVING AGRICULTURE IN CHINA

169

Fan, R. (1986). Some issues of the analysis in the effective precipitation during crop growth season. In “Proceedings of the Symposium on Urban Water Supply, Agricultural and Industrial Water Saving, Water Resources Protection in Hebei Province.” Shijiazhuang. Farquhar, G. D., and Richards, R. A. (1984). Isotope composition of plant carbon correlates with water use efﬁciency of wheat genotypes. Aust. J. Plant Physiol. 11, 539–552. Feng, G. (1999). The general ideas of water-saving irrigation development in China. In “Problems on Water Saving Agriculture in China.” (Dept. of Rural and Social Development, Ministry of Science and Technology, China, Ed.), pp. 11–29. China WaterPower Press, Beijing. Fischer, R. A., and Turner, N. C. (1978). Plant productivity in the arid and semiarid zones. Ann. Rev. Plant Physiol. 29, 277–317. Fu, L. (1989). Developing statue and prospects of micro-irrigation technology in China. In “Proceedings of Symposium of Water-Saving Agriculture and Development on Technology of Irrigation and Drainage in China (Institute of Field Irrigation of Chinese Academy of Agricultural Sciences, Ed.),” pp. 51–55. Guo, X., Liu, C., Shao, X., and Peng, S. (1999). The effect of regulated deﬁcit irrigation on water consumption regulation and water production efﬁciency of maize. Agr. Res. Arid Areas 17(3). Haihe Water Conservation Commission (1991). Newsletter of the Haihe Hydraulic Commission, Ministry of Water Resources. Han, D. (1999). Speech on green forum ‘problems on water saving agriculture in China.’ In “Problems on Water Saving Agriculture in China” (Dept. of Rural and Social Development, Ministry of Science and Technology, China, Ed.), pp. 11–29. China WaterPower Press, Beijing. Han, Q., Ma, L., Miao, X., and Ren, S. (1999). Rainwater-collection in water kiln and water saving irrigation for ﬁlm-mulched pepper. Agr. Res. Arid Areas 17(2). Jia, D. (1999). Issues on agricultural water saving in Yellow River basin. Irrigation Drainage 18(3), 1–3. Jia, D., Meng, Z., and Wang, H. (1999). High efﬁcient agricultural water use and agronomical watersaving technology. Water Saving Irrigation 4(4), 7–10. Jia, D., Si, T., and Wang, H. (1994). Study on sustainable development of water-saving agriculture. Eco-agriculture Study 2(2), 30–36. Kang, S., Liu, X., and Zhang, G. (1994). “The application and water transfer theory in the Soil-PlantAtmosphere Continuum.” Hydrology and Hydroelectric Press, Beijing. Kang, S., Zhang, J., Liang, Z., Hu, X., and Cai, H. (1997). The controlled alternative irrigation—A new approach for water saving regulation in farmland. Agr. Res. Arid Areas 15(1), 1–6. Lei, S. (1999). Lecture II. Reformation and development of irrigation water source project. China Rural Water Hydropower 5. Li, X. (1999). Several problems concerning soil. Water Saving Irrigation 1, 13–17. Li, Y. (1999). Scheme of high efﬁcient water use in agriculture in Loess Plateau. China Rural Water Hydropower 1, 10–12. Li, Z., and Shi, Q. (1999). Lecture VI Rainwater collecting irrigation. Water Saving Irrigation 4. Li, Y., and Wang, X. (1994). Root signals and its roles in the optimization of water use in plant. Chinese Bull. Bot. 11(2). Li, L., Wang, Y., and Li, Y. (Eds.) (1999). Technology of pipe water conveyance project. pp. 1–10. China WaterPower Press, Beijing. Lin, X., and Zhao, L. (1999). “Water-Saving Technology for Surface Irrigation of Dry Land Crops.” China WaterPower Press, Beijing. Liu, C. (1989). Agricultural water saving and water regulation in the North China Plain. Geog. Res. 8(3), 1–9. Liu, C., and Li, L. (1999). Sustainable water supply and water-saving industrial engineering in China. In “Problems on Water Saving Agriculture in China” (Dept. of Rural and Social Development, Ministry of Science and Technology, China, Ed.), pp. 42–57. China WaterPower Press, Beijing.

170

H. WANG ET AL.

Liu, C., and Mou, H. (1995). Connotation of rainwater resourcefulization and utilization. In “Proceedings of the 7th International Conference on Rainwater Catchment and Utilization Systems.” Beijing. Liu, C., and Ren, H. (1988). “Air, Surface, Soil and GroundWater Interactions—Experiment and Calculation Analysis.” Science Press, Beijing. Liu, M., and Wang, X. (1992). Analysis on changes in cropping system for saving water in Heilonggang region of Hebei Province. In “Study on Water-Saving Agriculture. (Y. Xu, Ed.),” pp. 52–57. Science Press, Beijing. Liu, C., and Wang, H. (1995). Discussion on the connotation of water saving agriculture. In “Advances in the Applied Fundamental Researches of Water Saving Agriculture” (Y. Shi, C. Liu, and Y. Gong, Eds.), pp. 7–19. Agriculture Press of China, Beijing. Liu, C., and Wei, Z. (1989). “The Agricultural Hydrology and Water Resources in the North China Plain.” Science Press, Beijing. Luo, Q. (1999). Stratedy on sustainable development of water-saving agriculture. Science and Technology Daily, 4 May. Nageswara Rao, R. C., and Wright, G. C. (1994). Stability of the relationship between speciﬁc leaf area and carbon isotope discrimination across environments in peanut. Crop Sci. 34, 98–103. Newsletters of Haihe Hydraulic Commission, Ministry of Water Resources (1991). Special issue: Agricultural water-saving technology. Philip, J. R. (1966). Plant water relations: some physical aspects. Annu. Rev. Plant Physiol. 17, 245–268. Report of the Institute of Hydrology, Hebei. (1985). Rational regulation of groundwater table to increase the available water resources. Shan, L., and Zhang, S. (1999). Water-saving agriculture and its biological basis. In “Problems on Water Saving Agriculture in China” (Dept. of Rural and Social Development, Ministry of Science and Technology, China, Ed.), pp. 30–41. China WaterPower Press, Beijing. Shang, X., and Chang, J. (1999). Storage and utilization of rainwater and development of rural economy in central Gansu. Agr. Res. Arid Areas 17(2). Shen, Z., Wang, L., Yu, F., and Liu, B. (2000). “New Concept of Water Saving—Study and Application of Real Water Saving.” China WaterPower Press, Beijing. Si, T., Zhang, W., and Jia, D. (1992). A new irrigation technique for saving water and increasing yield— irrigation on ﬁlm under wheat-cotton interplanting. In “Water Use Efﬁciency in Agriculture” (Y. Xu, C. Liu, and J. Shalhevet, Eds.), pp. 183–188. Science Press, Beijing. Stanhill, G. (1986). Water use efﬁciency. Adv. Agron. 39, 53–85. Subproject 75-57-01-01. Report of Ministry of Water Resources (1990a). Study on the interactions between atmosphere water-surface water-soil water-groundwater in the North China regions. Subproject 75-57-03-03. Report of Ministry of Water Resources (1990b). Study on agricultural water saving techniques. Sub-project 75-57-03-05. Report of Ministry of Water Resources (1990c). Present situation, developing trend, and strategy study on demand and supply of water resources in North China regions. Sun, R., Wu, K., and Liu, S. (1992). Analysis on water-saving beneﬁt of border irrigation with different sized plots. In “Study on Water-Saving Agriculture” (Y. Xu, Ed.), pp. 194–199. Science Press, Beijing. Wang, H. (1997). Water Transfer and Efﬁcient Water Use in Agro-ecosystem: Experiments and Modelling. Ph.D. thesis of Institute of Geography, Chinese Academy of Sciences. Wang, C. (1999). Brief discussion on drying up of Yellow River and development of water-saving agriculture. Water Saving Irrigation 6, 11–13. Wang, Z., Xiong, Y., and Wang, M. (2000). “Study on Expert System of Irrigation Forecast and Decision Making for Water Saving.” Institute of Agricultural Soil & Water Engineering of Northwest SciTech University of Agriculture & Forestry, Yangling, Shaanxi. Wang, X., Zhao, C., and Chen, H. (1993). “Study on Water Saving Agriculture and Water Saving Techniques.” Meteorology Press, Beijing.

WATER-SAVING AGRICULTURE IN CHINA

171

Wright, G. C., Nageswara Rao, R. C., and Farquhar, G. D. (1994). Water use efﬁciency and carbon isotope discrimination in peanut under water deﬁcit conditions. Crop Sci. 34, 92–97. Wu, G., Wang, X., and Lou, C. (1993). Abscisic acid content and leaf conductance in relation to soil water content in. Pisum sativum L. Acta Phytophysiol. Sini. 19(4). Xi, D., Li, H., and Zhang, S. (1986). Study on the water-saving and determining fertilizer according to water condition cultivation of winter wheat in Heilonggang region. Irrigation and Drainage 5(2), 17–21. Xiao, G., Ren, W., Liu, Y., and Lu, Y. (1999). Research on technique of storing rainwater in pit for crop complementary irrigation. Agr. Res. Arid Areas 17(3), 81–86. Xing, X., and Wang, X. (1992). Study on ways for soil water resources development and utilization in Heilonggang region, Hebei Province. In “Study on Water-Saving Agriculture” (Y., Xu, Ed.), pp. 46–51. Science Press, Beijing. Xu, J., Qiu, L., Cheng, N., and Liu, P. (2000). Expert system on regional water resources planning and water-saving and yield-increasing irrigation in the irrigation districts—A key research project of Ministry of Water Resources. North China WaterPower College, Beijing. Xu, M. (1991). Water-saving agriculture and its physiological ecological basis. J. Appl. Ecol. 2(1). 70–76. Xu, Y. (1992). Trends of water-saving agriculture in China. In “Study on Water-Saving Agriculture” (Y. Xu, Ed.), pp. 1–13. Science Press, Beijing. Yin, G., Liu, Z., Lin, H., and Zhang, Z. (2000). Optimum period and amount of supplementary irrigation with catchment of rainfall on upland ﬁeld with different cropping patterns. Agr. Res. Arid Areas. 18(1), 85–90. Yuan, X., Wang, H., Zhang, X., and You, M. (1992). Relationship between winter wheat yield and water consumption. In “Study on Relationship between Crop and Water” (X. Xie and H. Yu, Eds.), pp. 10–18. Science and Technology Press of China, Beijing. Zhang, C. (1999). Speech on Green Forum ‘Problems on water saving agriculture in China.’ In “Problems on Water Saving Agriculture in China” (Dept. of Rural and Social Development, Ministry of Science and Technology, China, Ed.), pp. 5–10. China WaterPower Press, Beijing. Zhang, Q. (1999). Water, life and environment—Looking at the water saving irrigation revolution in China in the light of the international water issues. In “Problems on Water Saving Agriculture in China” (Dept. of Rural and Social Development, Ministry of Science and Technology, China, Ed.), pp. 58–76. China WaterPower Press, Beijing. Zhang, W. (1999). Brackish water and sewage utilization.Guangming Daily, 18 May. Zhang, X., Cheng, W., and Liu, E. (1992). Experimental study on the effects of ﬁeld mulching on water saving. In “Study on Water-Saving Agriculture” (Y. Xu, Ed.), pp. 200–203. Science Press, Beijing. Zhang, Z., Gong, S., and Wang, X. (Eds.), (1999). “Technology of Rainwater Catchment Project,” pp. 1–5. China WaterPower Press, Beijing. Zhang, H., and Liu, X. (1992). Quantity analysis on water-saving irrigation of winter wheat in the North China Plain. In “Study on Water-Saving Agriculture” (Y. Xu, Ed.), pp. 104–109. Science Press, Beijing.

This Page Intentionally Left Blank

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES E. Ben-Dor The Remote Sensing and GIS Laboratory Department of Geography and the Human Environment Tel-Aviv University PO Box 39040 Ramat Aviv, Tel-Aviv 69978 Israel

I. Introduction A. General Overview B. Quantitative Soil Spectroscopy in the Laboratory C. Spatial and Spectral Aspects D. The Hyperspectroscopy Approach II. Principles of Quantitative Remote Sensing of Soils A. Spectral Measurements B. Spectral Chromophores III. Mechanisms of the Soil –Radiation Interactions A. Chemical Processes B. Physical Processes IV. Problems in Quantitative Remote Sensing of Soil A. Atmospheric Attenuation B. Spectral Resolution and Number of Channels C. Signal to Noise D. Pixel Size and Sampling Techniques E. Measurement Geometry V. Parameters Affecting the Remote Sensing of Soil A. Vegetation Coverage B. Soil Crust and Surface VI. High-Spectral-Resolution Sensors A. Current and Future Sensors B. Cost and Availability VII. General Analytical Methods VIII. Closing Remarks and Recent Examples References

The remote-sensing approach, using satellite and airborne sensors, is rapidly entering the field of environmental sciences as a complementary tool for studying

173 Advances in Agronomy, Volume 75 C 2002 by Academic Press. All rights of reproduction in any form reserved. Copyright 0065-2113/02 $35.00

174

E. BEN-DOR natural processes. This is mainly because the approach enables an overview of large areas simultaneously, using multiple spectral information that correlates with most of the common land cover compositions, on a temporal basis and in a cost-effective way. In soil science, this technique has shown a potential for determining soil groups, the soil genesis process, and soil degradation, and reflects some soil environment interactions as well. Apparently, the limited spectral information provided by the former sensor did not allow the quantitative remote sensing of soils, and thus it could not be used to benefit the future of such endeavors as “precision agriculture.” Recent technological developments using remote sensing for monitoring the environment have produced a new approach that is able to provide quantitative rather than qualitative information regarding soil status. This approach, namely, hyperspectroscopy imaging (HSR), is characterized by many spectral channels, which expand the spectral information of the sensed material to be analyzed under quantitative approaches. This technique uses a image spectrometer that has been mounted onboard an aircraft and is able to receive discrete information regarding a sensed target from orbit. It has been successfully used in many disciplines, including geology and marine and vegetative studies. Because this technique holds new capabilities, it opens new frontiers in soil applications as well. The capability of soil spectral information to predict several important soil properties has already been demonstrated under laboratory conditions. Under noncontrolled (field) conditions, difficulties associated with the far-distant position of the sensors relative to the target and with the limited ability to sense only the upper soil crust currently prevent the HSR approach from being simply applied to soils. Only a welldesigned HSR approach will be able to provide quantitative soil property maps from such far distances. This paper provides a detailed description of the quantitative (spectral-based) approach for assessing soil properties, using the reflectance radiation across the sun’s illumination range along with an extensive discussion of the obstacles (and their possible solutions) preventing this approach from being a pure laboratory equivalent. Also provided is a detailed review of recent work that has concentrated on quantitative soil remote sensing along with a discussion of the future availability of this technology in terms of cost, physical specifications, and C 2001 Academic Press. possible applications.

I. INTRODUCTION A. GENERAL OVERVIEW Remote sensing (RS) of soils has become an attractive tool for assessing and mapping the soil environment from a far distance ever since the ﬁrst commercial satellite (ERTS-1, known as LANDSAT-1) was placed in orbit in 1972. Since then, tremendous progress has been made in both data acquisition technology and data processing techniques. The availability of advanced analog and digital data, taken from sensors both in the sky and in orbit, makes remote sensing of the Earth a useful

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

175

tool for many applications. Today, the remote-sensing ﬁeld has more applications than just scientiﬁc-oriented approaches, and many new users are discovering its potential for practical uses. It became known that the remote-sensing tool could provide not only nice color images but also quantitative information about every ground picture element (pixel) in the area, which indeed opened up a new frontier in spatial assessment. In general remote sensing is deﬁned as “the science and art of obtaining information about an object, area or phenomenon through the analysis of data acquired by a device that is not in contact with the object, area and phenomenon in the investigation” (Lillsand and Kiefer, 1992). In reality remote sensing refers to data acquisition from a far distance (e.g., sky or space), using electromagnetic radiation, which serves as an “agent” between the object and a distant sensor. This approach enables the rapid coverage of large areas and makes possible the enhancement of phenomenon never before considered. The remote-sensing ﬁeld brings information (heretofore invisible to the human eye) obtained from the interaction of electromagnetic radiation with matter into the region visible to the human eye and provides a spatial overview of “invisible” data. Active (e.g., RADAR-based source) and passive (e.g., SUN-based source) energy are the common sources for the remote sensing of objects. Although the spectral region of these sources covers a wide range (0.4 μm to 1 m), soil materials carry signiﬁcant information only across the 0.4- to 2.5-μm spectral region, which, with the overlapping sun radiation response, makes this region very important for the remote sensing of soil. It is important to note that the physical and analytical principles of remote sensing of soils under ﬁeld conditions are similar to those applied under laboratory conditions. However, in the remote-sensing environment, the interaction of electromagnetic radiation along the source–target–sensor path, makes the remote sensing of soils a more complicated task than it is in the laboratory. This is because more factors that affect the ﬁnal soil spectra are involved, and they can bias the results if not properly considered. In this regard, the atmosphere plays an important role in preventing direct terrestrial remote sensing. Also the sensor’s low signalto-noise ratio (SNR) and its low spectral resolution capabilities are signiﬁcant limitations, and the fact that we sense only the upper 50 μm of the soil surface is a problem in remote sensing of soils where the entire proﬁle is important. Other limitations are the bidirectional effect, the sensor’s geometrical distortions, and costs of data acquisition. These obstacles not only hinder the quantitative remote sensing of soils but also make it a more complicated task. Soil reﬂectance spectra in the VIS–NIR–SWIR (0.4- to 2.5-μm) region are known to be rather complex and often do not permit utilization of simple spectral analysis routines (Ben-Dor et al., 1998, Price, 1998). In the quantitative approach, in general, and in applying NIR analysis, in particular, to soils, it is most important to ensure that the analyzed spectrum is indeed composed of pure soil components and not of artifacts. As mentioned earlier, some factors prevent the purely quantitative remote sensing of soils; however, if these factors are known, the remote-sensing ﬁeld can offer a unique way to rapidly assess and monitor

176

E. BEN-DOR

new phenomenon over large areas. If the RS approach can rapidly and accurately provide quantitative information about large areas, it might be considered a perfect tool for improving the decision-making system for soil production, understanding environmental dynamics, and interpreting previously unseen phenomenon (such as soil fertility, soil water capacity, etc.). Quantitative remote sensing of soils might also open up new horizons for the developing practice of “precise farming” (see, for example, Carey, 1995). Here, RS could be used to bring the farmer in touch with the most reliable and updated chemical and physical information about his ﬁelds on a pixel-by-pixel basis. Since signiﬁcant achievements have been made in the remote-sensing ﬁeld in both data acquisition technology and data-processing techniques, it appears that quantitative remote sensing of soils could be the next step. Advanced studies that have used cutting-edge RS technology conﬁrm this statement and indicate high expectations from future RS activity. In this section we will discuss the problematic factors preventing a direct spectral analysis of the electromagnetic signals for precise soil mapping purposes. In this regard, we will highlight the new remote-sensing technology’s ability to acquire near-laboratory quality data, provide a comprehensive discussion on soil chromophores, discuss the common quantitative ways for pixel classiﬁcation and subpixel assessments, and review recent studies that reveal breakthroughs in quantitative remote sensing of soil properties. In this section we will limit our discussion to the passive radiation of the Sun’s illumination, which in the RS ﬁeld is often divided into three regions: the visible (VIS, [Visible] 0.4- to 0.7-μm), near infrared (NIR, [Near Infra Red] 0.7- to 1.1-μm), and short-wave infrared (SWIR, [Short Wave Infra Red] 1.1- to 2.5-μm) spectral regions. Parts of this section were taken from Ben-Dor et al., 1999 and the readers are referred to this citation for enlarging the envelope knowledge about the soil spectra.

B. QUANTITATIVE SOIL SPECTROSCOPY IN THE LABORATORY Soil and soil spectra are rather complex phenomena. This prevents a straightforward prediction of reﬂectance properties by physical theories or models. Using physical models there is a potential for quantitative conversion of a reﬂectance spectrum of a multimineral surface to actual mineral abundances (Clark and Roush, 1984). Nevertheless, because this situation is not simply applied to a soil system, where complex relationships between chromophores exist, and because theoretical models do not always agree with reality, the theoretical approach is not valid for soil properties assessments. Consequently, empirical quantitative approaches were developed to derive chemical–physical information from the soil spectra. Basically the empirical methods are based on the reﬂectance being equivalent to the transmittance, and on the idea that photons obey Beer’s law for a given path length within the surface studied and on absorption coefﬁcients and the concentration of the material. Under laboratory conditions where physical parameters

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

177

remain constant, no atmospheric attenuation exists, and spectral noise is minimal, a soil spectrum tends to vary with mineralogy. Under such conditions the empirical relationship between the chemistry and the reﬂectance properties of powders can provide quantitative information about unknown materials solely from their reﬂectance spectra (Condit, 1972). Manipulation of spectra using derivatives and transformation to log space enables enhancement of weak spectral features and minimizes physical effects (Demetriades-Shah et. al., 1990). A promising quantitative laboratory approach, in the NIR–SWIR regions, was developed as a rapid method to analyze moisture in grains (Ben-Gera and Norris, 1968). The method, currently termed near-infrared reﬂectance analysis (NIRA), is currently widely accepted and used in many disciplines (Davies and Grant, 1987; Norris, 1988; Stark et al., 1986; Williams and Norris, 1987) including remote sensing of vegetation from hyperchannel sensors (Curran et al., 1992; Gong and Miller, 1992; LaCapra et al., 1996; Wessman et al., 1988, 1989) and other complex mixtures (Honigs et al., 1984). Basically the NIRA method assumes that a concentration of a given constituent is proportional to the linear combination of several absorption features. The method is empirical, and although no physical or chemical assumptions are made the method has a strong spectroscopy foundation. The NIRA approach has two stages: (1) the calibration stage, where a prediction equation for evaluating a property is developed; and (2) the validation stage, where the previous stage is validated. The calibration stage uses “training samples” that represent the study population in terms of spectral and physical/chemical properties. Then a prediction equation based on multiple regression analysis between the soil chemistry data (determined in the laboratory) and the selected spectral bands is generated. This calibration equation is further validated in stage 2 against “unknown samples” and is statistically examined for its prediction performance. The NIRA concept has been successfully applied to soil, soil minerals, and soil organic materials in the laboratory. Dalal and Henry (1986) showed that organic carbon, nitrogen, and soil moisture can be simultaneously predicted from the reﬂectance spectra of Australian soils. Morra et al. (1991) applied the NIRA methodology on soil samples of 12 subgroups, and established a model to predict the total carbon and nitrogen content solely from the soil reﬂectance. Ben-Dor et al. (1997) showed that the nature and shape of the organic matter’s reﬂectance curve vary with time, and that organic matter age can be predicted using the NIRA approach on data in the VIS–NIR–SWIR region. Nutrients in the soil environment have also been shown by Malley et al. (1999) to be capable of rapid determination in the laboratory using the NIRA technology. In another study, Ben-Dor and Banin (1990a) applied a NIRA algorithm on SWIR data from several smectite minerals and predicted their total Al2O3, Fe2O3, MgO, and SiO2 content. In arid and semiarid soils from Israel, Ben-Dor and Banin (1990b) predicted the CaCO3 content from data in the SWIR region, and showed (Ben-Dor and Banin, 1995a) the ability to quantitatively predict soil properties such as clay content, speciﬁc surface area, cation-exchange capacity, hygroscopic moisture, and organic matter content.

178

E. BEN-DOR

The NIRA approach has shown that using a relatively low number of spectral channels (25–63) is sufﬁcient to predict the following soil properties: hygroscopic moisture (25 channels), organic matter (30 channels) clay content, cation-exchange capacity, and speciﬁc surface area (63 channels). However, prediction of the carbonate content in the soils required 3113 channels. In another study, Ben-Dor and Banin (1995a) showed that the quantitative approach provided by the NIRA routines enables the quantitative assessment of SWIR-featureless soil properties, such as total Fe2O3, Al2O3, SiO2, free iron oxides, average particle size, and total K2O. Figure 1 presents the calibration (A) and validation (B) results of the NIRA analysis applied to 91 soils from Israel for the previously described “featureless” properties. Reasonable standard error of calibration (SEC) and performance (SEP) were obtained for the soil properties examined. The authors concluded that the intercorrelation between feature and featureless soil properties is the major mechanism that enables the NIRA to successfully work under so complex a matrix which makes it a powerful quantitative approach. Examining the NIRA approach for the VIS–NIR spectral region illustrated that this concept is also applicable to monotonous spectral curves with minimal spectral variations (Ben-Dor and Banin, 1994). The NIRA approach was able to work in a complex soil system because of the relatively high signal-to-noise ratio, high spectral resolution capability, constant measurement conditions, and a detailed study of the soil population at hand. The high quality of the spectra used in this method allows sufﬁcient utilization of many mathematical manipulations methods such as derivation, conversion to other space, and subtractions. These manipulations obviously improve the general accuracy obtained by the empirical methods. The NIRA approach has limitations as it was based on “older” theories that did not completely represent the scattering process. However, in practice the NIRA model has had several successful applications. Ben-Dor and Banin (1995c) used the NIRA algorithm on laboratory reﬂectance data (processed to simulate six TM bands in the VIS–NIR–SWIR region). They found that carbonate concentration, loss-on-ignition, speciﬁc surface area, and total SiO2 in soil could be predicted solely from the simulated TM spectra. These studies show that an empirical approach and no a priori knowledge can quantitatively analyze soils from their reﬂectance spectra. Other quantitative methods for analyzing soil spectra include those which explain the spectral variability of a given population by vector analysis (Condit, 1972; Kimes et al., 1993; Price, 1990). However, in this approach, it is possible that highly correlated channels will be grouped into one broad band, and this may underestimate subtle but signiﬁcant spectral features meaningful to soil properties. Figure 1 Plots of the predicted versus measured values at the calibration (A) and validation (B) stages for several featureless soil properties using the NIRA approach in the laboratory (Fe2O3, Al2O3, SiO2, K2O are total constituents in soils expressed as oxides. LOI, lost on ignition (1000◦ C) values; Fed, dithionite extractable iron AVGR, mean aggregate size fraction; F1, aggregate size between 1.4 and 1 mm). (From Ben-Dor and Banin, 1995b).

179

180

E. BEN-DOR

C. SPATIAL AND SPECTR A L A SPECTS Basically, remote sensing is a tool for providing a broad overview of a given area in a short time and for different applications. For the soil NIRA approach, both spectral and spatial aspects are signiﬁcantly important. The spatial domain provides the ability to locate the property in question in two dimensions; whereas the spectral domain provides a third, physiochemical dimension, which is important for target classiﬁcation. The ﬁrst aspect in this context (spatial) answers the need for a reliable and convenient way to precisely register a point located in the ﬁeld with its image and then to represent it on a map. The second aspect (spectral) fulﬁlls the requirement of providing detailed and precise information about the natural characteristics of a sensed pixel. The procurement of high spatial information is not only an optical-dependent process—often achieved by analog or digital cameras, obtaining high spectral information, which is essential for NIR analysis—but also a sensor-dependent process, which is still under development. From a practical standpoint, obtaining high-spectral-resolution data from great distances is still the “bottleneck” for quantitative RS-based analysis. High spectral resolution can be achieved in the ﬁeld (and also in the air) using portable ﬁeld spectrometers or digital sensor devices that enable data acquisition in many continuous spectral channels. The measurement of soil targets by portable ﬁeld spectrometers requires good knowledge of the ground location, in order to precisely map the spatial variation of the property in question. Because of this, methods to precisely locate the ground points are essential for any calibration or validation steps of the NIR analysis. In this regard, a satellite-based system, known as ground position system (GPS), can be used as the optimal solution for merging the spatial and spectral characteristics of selected points in the ﬁeld. If the NIR approach is considered for use in conjunction with RS data, knowledge of the exact location of the ground points is critical to achieve an accurate ﬁnal result. The precise allocation of chemical values on the image and on the ground for spatial presentation is the key factor in obtaining reliable thematic maps. Information concerning the registered points is used in the case of RS data used with a spectral-based model that runs on a pixel-by-pixel basis or in the case of a GIS using geostatistical models that interpolate the information into maps. In both analytical approaches, a quantitatively rectiﬁed map of the soil property in question is the ﬁnal product intended to be delivered to the end users (e.g., farmer, government authority). A new and advanced remote-sensing technique, namely, image spectroscopy, also known as hyperspectral remote sensing (HSR) (more on this later), provides a spectrum representation of every pixel in the image, and therefore can be used together with NIR analysis. Although merging high spectral and spatial information appears to be an ideal combination for accurate remote sensing of soil properties, this may require greater computing resources (CPU and digital space) and is not always recommended for terrestrial applications. Recently, Goetz and Kindel

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

181

(1996) pointed out that high spectral resolution is a more important property than high spatial resolution, if quantitative mineral mapping application is done. They showed that the airborne-visible/infrared-imaging spectrometer (AVIRIS) sensor (20-m pixel size and 224 spectral channels across the VIS–NIR–SWIR region) can provide similar mapping results to what can be extracted by the HyDyce sensor (3-m pixel size and 50 spectral channels across the SWIR region). It should be pointed out that another argument with regard to the need for high spectral resolution of RS data exists (e.g., Price, 1995). This is because of the intercorrelation among closed channels, which in some sense, makes the spectral information repeatable. In this regard, Ben-Dor and Banin (1995c) have shown, under laboratory conditions, that it is possible to apply NIR analysis using the six channels of the TM sensor for determining four soil properties (CaCO3, SIO2, LOI, and SSA). To the best of our knowledge, no such work has been applied to a real remote-sensing data set, and thus it is still a question whether six spectral channels are enough for quantitative assessment of soil properties from a far distance. In summary it can be said that high spectral and spatial resolution can provide useful information regarding a soils’ surface properties, but the exact spatial/spectral combination is still subject to many factors, and the value has to be weighed against the system used, available budget, and speciﬁc questions asked.

D. THE HYPERSPECTROSCOPY APPROACH As mentioned earlier, it is obvious that for the NIRA analysis process, highspectral-resolution data having a good signal-to-noise ratio is a key factor for deriving chemical information from soil reﬂectance. Apparently, in the remote-sensing ﬁeld, most of the sensors are multi- and not hyperspectral systems, and thus it is assumed that quantitative approaches for soil applications are limited. Multisensor data may provide indirect qualitative information about a sensed area by using several common image-processing methods (see later discussion). Accordingly, in the past decade, new data acquisition and processing techniques have been developed to extract detailed spectra-based information for thematic mapping. This capability has enabled a more quantitative rather than qualitative view of large areas and has stretched remote-sensing capabilities to their outside limits. This technology (image spectroscopy, or hyperspectral remote sensing) and its concept are shown in Fig. 2. Instead of acquiring images in a few separate bands of various widths (as is done with common satellite mutisensors) HSR makes possible the simultaneous collection of images in 100 or more contiguous spectral bands. The result is then, a continuous spectrum for each pixel across and along the image. However, although the HSR technology is being developed and has shown signiﬁcant progress since the ﬁrst airborne-imaging spectrometer (AIS) was examined onboard a C-130 aircraft in 1983 (Conel et al., 1987; Goetz, 1987), some

182

E. BEN-DOR

Figure 2 Schematic overview of the imaging spectrometer principal. (After Vane and Goetz, 1988).

signiﬁcant drawbacks have been encountered with the actual quantitative analysis of the HSR data information as compared with real laboratory data. Signal-to-noise level, sensor motions and type, nontransparent atmosphere, and problems in sensor calibrating are some of the main concerns that must not be ignored. In addition, changes in soil particle size, subpixel composition, and bidirectional effects can introduce signiﬁcant problems, which make a simple spectral quantitative analysis using HSR more difﬁcult (Goetz, 1992a,b; Rast, 1991). It should also be pointed out that the passive remote-sensing technique only senses the upper 50 μm of soil; therefore, the entire proﬁle of information, which serves as the major tool for soil mapping, cannot be achieved. Nevertheless, based on knowledge of the previously described factors, the HSR technique is rapidly entering the ﬁeld of remote sensing as a tool for precise and quantitative analysis of atmospheric–terrestrial systems from air (Goetz, 1991). Since HRS provides a near-laboratory-quality reﬂectance spectrum of each pixel, it allows identiﬁcation of objects based on their well-known spectral absorption features (Clark and Roush, 1984; Goetz, 1985). Today, only a limited number of sensors are available worldwide for HRS applications, and they are rather expensive for routine operation. Nevertheless, it seems that new technology that is under development will reduce the sensor cost to an affordable value, and very soon the HSR technology will be available for end users on a cost-effective basis. In this regard, HSR from orbit will allow data acquisition of good spectral data on a routine basis, which can be used for temporal applications of many terrestrial systems, including soils. Basically, the current HSR sensors cover the visible (VIS, 0.4- to 0.7-μm), near infrared (NIR, 0.7- to 1.1-μm), and short-wave infrared (SWIR, 1.1- to

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

183

2.5-μm) spectral regions, e.g., AVIRIS (Vane, 1987), GERIS (Collins and Chang, 1990), MIVIS (Bianchi and Marino, 1994), and HyMap (Cocks et al., 1998). Some are only capable for use in the VIS–NIR region, e.g., CASI (Gwinner et al., 1995) and AISA (Bars et al., 1998; Okkonen et al., 1997), and some are only capable for use in the SWIR region [SFSI (Staenz et al., 1999)]. There are two HSR instruments that also cover the thermal region (TIR, 3.0–14 μm) along with the VIS–NIR– SWIR region, e.g., MIVIS (Bianchi and Marino, 1994) and DAIS-7915 (Strob et al., 1996). A good review of current and proposed HSR sensors can be found in http://www.geouniz.ch/∼schaep/research/apex/is list.html. Also relevant information and contact lists of sensor vendors and owners are provided in Kruse (1998) with some case study examples and in Joseph (1996). It should be mentioned that recently (April 2000, http://terra/gfc.nasa.gov) NASA announced that the Terra spacecraft sensors are “open for business.” This is after completion of an in-orbit check and veriﬁcation of the new orbit sensors’ data. Two sensors onboard the Terra spacecraft are HSR applicable: the MODIS, with 36 bands across the VIS–NIR– SWIR–TIR spectral region, having low spatial resolution (250–1000 m); and the ASTER, with 14 channels across the VIS–NIR–SWIR–TIR region, having relatively high spatial resolution (15–90 m). In the near future new HSR systems will be placed in orbit by both governmental and private sectors, such as PROPBA (of the European Space Agency [ESA http://www.estec.esa.nl/CONFANNOUN/proba/] launched on October 2001) and HyPERION (of NASA) http://eo1.gsfc.nasa.gov/ Technology/Hyperion.html launched on November 2000). This will promote the use of HSR technology for many applications, and it is anticipated that quantitative soil remote sensing will play a major role in this stage.

II. PRINCIPLES OF QUANTITATIVE REMOTE SENSING OF SOILS A. SPECTRAL MEASUREMENTS Soil reﬂectance data can be acquired in the laboratory or in the ﬁeld and from both sky and orbit. Whereas in the laboratory soil reﬂectance measurements are done under controlled conditions (sample geometry, illumination, no atmospheric interference) in the ﬁeld, reﬂectance measurements are encumbered by problems such as variations in viewing angle, illumination changes, soil roughness, and exact ground position. Soil reﬂectance data acquired from the air or orbit involve additional difﬁculties such as low signal-to-noise ratio and atmospheric interference. In this regard, the laboratory-based measurements enable an understanding of the chemical and physical principles of soil reﬂectance, but do not promise that all of the laboratory applications can be adopted for the remote

184

E. BEN-DOR

sensing of soils. An extensive body of work has been applied to soil–laboratory analysis (see later in this chapter), whereas very little has been conducted for soil–ﬁeld (i.e., using a portable spectrometer) or soil–air (or space) applications (i.e., using HSR technology). In general, ﬁeld measurements are characterized by better spectral performances than the HSR sensors provide (see later discussion), and thus, they are preferable for extracting soil properties using the NIR– SWIR region. However, using a ﬁeld (point) spectrometer rather than an image spectrometer (point-by-point) reduces the accuracy of the ﬁrst mapping process. This is because the interpolation of selected points is less accurate than the information obtained from a mosaic of hundreds of closed pixels having chemical–physical information. Some common portable point spectrometers for remote sensing are available, such as ASD (http://www.asdi.com/prod/ps2.html), PIMA (htpp://www.intspec.com/pima/pima.htmt), GER (htpp://www.ger.com), and LICOR (http://licor.alcavia.net/). A comprehensive technical review of ﬁeld spectrometers is given in http://www.themap.com.au/overview spectrometer.htm. Point spectrometers are characterized by a high signal-to-noise ratio and good stability. Their quick response time can make point spectrometers operable from aircraft, from where they can spectrally track kinetic processes (e.g., the soil, and vegetation drying process). Based on this ability, Karnieli et al. (1998) mounted an ASD spectrometer on a light aircraft equipped with a video camera and acquired measurements along a climate cross section in Israel (personal communication). They showed for the ﬁrst time that a laboratory/ﬁeld point spectrometer can be used as an airborne tool for assessing soil and vegetation status with very high signal-to-noise standards. It should be pointed out that measuring soil reﬂectance in the ﬁeld using sun illumination is still problematic since it involves atmospheric attenuation (e.g., at 1.4 and 1.9 μm, a strong absorption of water vapor exists). To solve this problem an artiﬁcial light source that directly illuminates the target is used. For example, in the ASD spectrometer an external device with a closed chamber (named POTETO) is used, whereas in the PIMA spectrometer a built-in illumination source is in effect.

B. SPECTRAL CHROMOPHORES In the NIRA approach it is extremely important that the wavelength entered into the prediction models will hold a signiﬁcant spectral assignment and not spectral noise (Ben-Dor et al., 1997). This makes the wavelength assignment an important stage in any NIRA analysis. For that purpose one has to explore all available chromophores and study the possible relationship between soil properties and spectral responses of these signiﬁcant spectral features. A chromophore is a parameter or substance (chemical or physical) that signiﬁcantly affects the shape and nature of a soil spectrum. A given soil sample consists of a variety of chromophores, which vary with environmental conditions.

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

185

In many cases the spectral signals related to a given chromophore overlap with signals of other chromophores, and thereby hinder the assessment of the affect of a given chromophore. Because of the complexity of the chromophores in soil, it is important to understand the chromophores’ physical activity as well as their origin and nature. The spectra of pure minerals are extensively discussed elsewhere, and readers are referred to those sources (e.g., Clark et al., 1990; Hunt, 1970–1980). In the following section, however, our discussion will focus primarily on factors affecting soil spectra, directly and indirectly, from both chemical and physical chromophores.

1. Soil Chemical Chromophores Soil chemical chromophores are those materials in soil system that absorb incident radiation in discreet energy levels. Usually the absorption process appears on a reﬂectance spectrum as troughs whose positions are attributed to speciﬁc chemical groups in various structural conﬁgurations. All features in the VNIR–SWIR spectral regions have a clearly identiﬁable physical basis. In soils, three major chemical chromophores can be roughly categorized as follows: minerals (mostly clay and iron oxides), organic matter (living and decomposing), and water (solid, liquid, and gas phases).

a. Clay Minerals Clay minerals (also referred to as phyllosilicate minerals) are crystalline aluminosilicate minerals organized in a layered structure. The crystal structure consists of two basic units: the Si tetrahedron, which is formed by a Si4+ ion surrounded by four O2− ions in a tetrahedral conﬁguration, and the Al octahedron, which formed by an Al3+ ion surrounded by four O2− and two OH−1 ions in an octahedral conﬁguration. These structural units are joined together into tetrahedral and octahedral sheets, respectively, by adjacent Si tetrahedral sharing all three basal corners and by Al octahedrons sharing edges. These sheets, in turn, form the clay mineral layer by sharing the optical O of the tetrahedral sheet. Layer silicates are classiﬁed into eight groups according to layer type, layer charge, and type of interlayer cations. The layer type designated 1 : 1 is organized with one octahedral and one tetrahedral sheet, whereas the 2 : 1 layer type is organized with two octahedral and one tetrahedral sheet. A one-layer octahedral sheet (1) is also found in highly leached acid soils. The layer silicate charge is a function of isomorphic substitution that occurs in both tetrahedral and octahedral positions during the weathering process. Charge density is one of the major factors governing soil behavior, and therefore mineral species and composition are considered to be keys to understanding soil behavior.

186

E. BEN-DOR

b. Origin of Clay Minerals in Soils All clay minerals are derived from the weathering of primary minerals. The occurrence of smectite, vermiculite, illite, or kaolinite is related to the degree of weathering and the chemical nature of the soil environment. Muscovite tends to produce illite; whereas biotite tends to produce vermiculite. Both illite and vermiculite are associated with slightly weathered materials. Vermiculite requires large amounts of magnesium during clay formation, which would most likely occur in neutral to slightly alkaline soils. Illite occurs to a greater extent than vermiculite in soils of moderate acidity. Illite tends to form smectite as surface potassium ions are removed by the weathering processes and new cation substitution occurs. Smectite minerals (2 : 1 conﬁguration) are an important component of slightly to moderately weathered soils, which are formed under relatively high pH values and speciﬁc Si and Al concentrations in the soil solution. Kaolinite minerals (1 : 1 conﬁguration) are predominant in highly weathered, leached soils that turn, under stronger weathering and acid conditions, into gibbsite (1 conﬁguration). Whereas gibbsite is quite rare, smectite and kaolinite are more commonly found in soils. Kaolinite may be formed from a 2 : 1 mineral during the weathering process and requires an environment where both silica and alumina are accumulated in a ratio favorable to its formation. Illite and vermiculite are associated with youthful materials. Smectite is formed in the middle stage of the weathering process, and therefore is most likely found in many soils as the major or secondary mineral. Similarly, the kaolinite component in soils tends to increase with increasing stages of weathering. Soils of warm temperate regions have a high percentage of kaolinite in the clay fraction, whereas cold areas tend to form more illitic- and smectitictype minerals. Of all the soil minerals discussed previously, smectite is thought to be the most active, because of its high speciﬁc surface area and electrochemical reactivity.

c. Other Minerals Figure 3 provides pure spectra of selected addition minerals found in soils. The nonclay minerals commonly found in soils are divided into ﬁve groups: namely, silicates, phosphates, oxides and hydroxides, carbonates, and sulﬁdes and sulfates. The fraction of each mineral in soils is dependent upon the environmental conditions and the parent materials. Primary minerals will most likely be found in young soils, where the weathering process is weak. Whereas silicates such as feldspars are rarely found in mature soils, quartz may be also found in some developed soils, depending on their environmental conditions and parent material. In general, the quartz mineral is spectrally inactive in the VIS–NIR–SWIR region and, therefore, diminishes other spectral features in the soil mixture. Other nonclay silicate minerals such as feldspars may have some diagnostic absorption features that make the soil spectrum less monotonous.

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

187

Figure 3 Reﬂectance spectra of representative pure non smectite clay minerals. (From Grove et al., 1992).

Oxide group minerals occur in highly weathered areas such as those associated with slopes, highly leached proﬁles, or in areas of “mature” soils. Phosphate and sulfate minerals can be found in soils as apatite and gypsum, respectively. Although both minerals have unique spectral features, their occurrence in soils may be relatively rare and even nondetectable. Other oxides, such as iron, are strongly spectrally active, mostly in the VIS region, because of the crystal ﬁeld and the charge transfer mechanism. The content of free oxides (both iron and aluminum) is low in young soils, but increases gradually as soil ages, just as happens with organic matter.

III. MECHANISMS OF THE SOIL–RADIATION INTERACTIONS A comprehensive description on the physical mechanism of the electromagnetic radiation with diverse minerals and rocks is provided by Clark (1998). This section, however, focuses on the most common chromophores in the soil environment and

188

E. BEN-DOR

its relationship with the electromagnetic radiation across the VIS–NIR–SWIR spectral region.

A. CHEMICAL PROCESSES 1. Clay Minerals Basically the spectral features of clay minerals in the NIR–SWIR region are associated with overtones and combination modes of fundamental vibrations of functional groups in the infrared region. Of all clay mineral elements, only the hydroxide group is spectrally active in the VIS–NIR–SWIR region. The OH group can be found either as part of the mineral structure (mostly in the octahedral position which is termed lattice water) or as part of a thin water molecule directly and indirectly attached to the mineral surfaces (termed adsorbed water). Three major spectral regions are active for clay minerals in general and for smectite minerals in particular: around 1.3–1.4, 1.8–1.9, and 2.2–2.5 μm. For Ca-montmorillonite (SCa-2), which represents a common clay mineral in the soil environment, the lattice OH features are found at 1.410 μm (assigned 2υ OH, where υ OH symbolizes the stretching vibration around 3630 cm−1), and at 2.206 μm (assigned υ OH + δ OH, where δ OH symbolizes the bending vibration at around 915 cm−1); whereas OH features of free water are found at 1.456 μm (assigned υ W + 2δ W, where υW symbolizes stretching vibration at around 3420 cm−1, and δW the bending vibration at around 1635 cm−1), and 1.910 μm (assigned υ ′ W + δ W, where υ ′ W symbolizes the high-frequency stretching vibration at around 3630 cm−1), and 1.978 μm (assigned for υ W + δ W). Note that the assignment positions discussed here can slightly change from one smectite to another depending upon their chemical composition and surface activity. The spectra of three smectite endmembers are given in Fig. 4 as follows: montmorillonite (dioctahedral, aluminuos), nontronite (dioctahedral, ferruginous) and hectroite (trioctahedral, manganese) minerals. The OH absorption feature of the ν OH + δ OH in combination mode at around 2.2 μm is slightly but signiﬁcantly shifted for each endmember. In highly enriched Al smectite, (montmorillonite) the Al–OH bond is spectrally, active at 2.16–2.17 μm. In highly enriched iron smectite (nontronite), the Fe–OH is effective at 2.21–2.24 μm, and in highly enriched magnesium smectite (hectorite) the Mg–OH bond is spectrally active at 2.3 μm. Based on these wavelengths, Ben-Dor and Banin (1990a) were able to ﬁnd a signiﬁcant correlation between the absorbance values derived from the reﬂectance spectra and the total content of Al2O3, MgO, and Fe2O3. Except for a signiﬁcant lattice OH absorption feature at around 2.2 μm in smectite, invaluable information about OH in free water molecules can be measured at around 1.4 and 1.9 μm. Because smectite minerals contributed to the soils’ relatively high

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

189

Figure 4 Reﬂectance spectra of three pure smectite endmembers in the NIR–SWIR region (nontronite = Fe smectite; hectorite = Mg smectite; montmorillonite = Al-smectite). Possible combination and overtone modes for explaining each of the spectral features are also given.

speciﬁc surface areas, which were covered by free and hydrated water molecules, these absorption features can be signiﬁcant indicators of the water content in soils. Kaolinite and illite minerals are also spectrally active in the SWIR region as they both consist of octahedral OH sheets. From Fig. 3, which presents pure spectra of non-smectite layer-clay minerals (kaolinite, chlorite, vermiculite, and illite), it is observed that different positions and spectral shapes of the lattice OH in the layer minerals affects soil spectra across the SWIR region. These changes are a result of the different structure and chemical composition of the minerals. In the case of kaolinite, a 1 : 1 mineral (one octahedral and one tetrahedral layer), the fraction of the OH group is higher than in 1 : 2 minerals (one octahedral and two tetrahedral layers), and, hence, the lattice OH signals at around 1.4 and 2.2 μm are relatively strong, and the signal at 1.9 μm is very weak (because of relatively low surface areas and adsorbed water molecules). In the case of gibbsite, an octahedral aluminum structure (1), the 1.4 μm is even stronger, but the signal at 2.2 μm is shifted signiﬁcantly to the IR (Infra-Red) region relative to the kaolinite. It should be noted that under relatively high signal-to-noise conditions, a second

190

E. BEN-DOR

overtone feature of the structural OH (3υOH) can be observed at around 0.95 μm in layer OH-bearing minerals as well (Goetz et al., 1991). Based on the previously described spectral features, Chabrillat et al. (1997) were able to use the AVIRIS sensor data to assess and map expansive clay soils in Colorado for urban planning and environmental applications. The afﬁnity of water molecules to clay mineral surfaces is correlated to their speciﬁc surface area. The speciﬁc surface area sequence of the previously described minerals is smectite > vermiculite > illite > kaolinite > chlorite > gibbsite, which usually provide a similar spectral sequence at the water absorption feature near 1.8 μm (area and intensity). As smectite and kaolinite are clay minerals often found in soils, they can also appear in a mixed-layer formation that overlap spectrally. Kruse et al. (1991) described a speciﬁc case at Paris Basin, France, where intrastratiﬁcation of smectite/kaolinite (a result of the alkaline weathering process of the ﬂint-bearing chalk) was identiﬁed. Figure 5 presents the spectra of mixed layer smectite, kaolinite, from the Paris basin area soils as examined by Kruse et al. (1991). The noticeable asymmetrical absorption feature of OH at 2.2 μm was further examined by Kruse to yield a graph that predicts the relative amount of kaolinite in the mixture (Fig. 6). 2. Carbonates Carbonates, especially calcite and dolomite, are found in soils that are formed from carbonic parent materials, or in a chemical environment that permits calcite and dolomite precipitation. Carbonates, especially those of ﬁne particle size, play a major role in many of the soil chemical processes most likely to occur at the root zone. A relatively high concentration of ﬁne carbonate particles may cause a ﬁxation of iron ions in the soil and consequently an inhibition of chlorophyll production. The absence of carbonate in soils on the other hand may affect the buffering capacity of the soil, and hence negatively affect the biochemical and physicochemical processes. The C–O bond, part of the –CO3 radical in carbonate, is the spectrally active chromophore. Hunt and Salisbury (1970, 1971a–c, 1971) pointed out that ﬁve major overtones and combination modes are available for describing the C–O bond in the SWIR region. In this table, υ1 accounts for the symmetric C–O stretching mode, υ2, for the out-of-plane bending mode, υ3 for the antisymmetric stretching mode, and υ4 for the in-plane bending mode in the infrared region. Gaffey (1986) has added two additional signiﬁcant bands centered at 2.23–2.27 μm (moderate) and at 1.75–1.80 μm (very weak); whereas Van der Meer (1995) summarized the seven possible calcite and dolomite absorption features with their spectral widths. It is evident that signiﬁcant differences occur between the two minerals. This enabled Kruse et al. (1990), Ben-Dor and Kruse (1995), and others to differentiate between calcite and dolomite formations using airborne spectrometer data with bandwidths of 10 nm. Except for the seven major C–O

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

191

Figure 5 Reﬂectance spectra of mixed-layer kaolinite/smectite from Paris Basin area. (After Kruse et al., 1991).

bands, Gaffey and Reed (1987) were able to detect impurities of copper in calcite minerals, as indicated by the broad band between 0.903 and 0.979 μm. However, such impurities are difﬁcult to detect in soils, because overlap with other strong chromophores may occur in this region. Gaffey (1985) showed that impurities of Fe in dolomite shift the carbonate’s absorption bands toward longer wavelengths, whereas Mg in calcite shifts the band towards shorter wavelengths. As carbonates in soils are most likely to be impure, it is only reasonable to expect that the carbonates’ absorption feature positions will be slightly different from one soil to another.

192

E. BEN-DOR

Figure 6 A correlation between the asymmetry of the 2.2-mm absorption band and percentage kaolinite form Paris Basin soil samples consisting of interstratiﬁcation of kaolinite/smectite. (After Kruse et al., 1991).

A correlation between the reﬂectance spectra and the carbonate concentration in soil was demonstrated by Ben-Dor and Banin (1990b) in Fig. 7. They used a calibration set of soil spectra from Israel and their chemical data to ﬁnd three wavelengths that best predict the calcite content in arid soil samples (1.8, 2.35, and 2.36 μm). They concluded that the strong and sharp absorption features of the C–O bands in the examined soils provide an ideal tool for studying the carbonate content in soils solely from their reﬂectance spectra. The best performance obtained for quantifying soil carbonate content ranged between 10 and 60%. 3. Organic Matter Organic matter plays a major role with respect to many chemical and physical processes in the soil environment, and has a strong inﬂuence on soil reﬂectance characteristics. Soil organic matter is a mixture of decomposing tissues of plants, animals, and secretion substances. The sequence of organic matter decomposition in soils is strongly determined by the soil’s microorganism activity. In the initial stages of the decomposition process only marginal changes occur within the chemistry of the parent organic material. The mature stage refers to the ﬁnal stage of microorganism activity, where new, complex compounds, often called humus, are formed. The most important factors affecting the amount of soil organic matter are those involved with soil formation, i.e., topography, climate, time, type of

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

193

Figure 7 Prediction of CaCO3 concentration is soils from Israel using a calibration curve of the reﬂectance values at 1.8, 2.35 and 2.36 mm against the measured chemical values. (After Ben-Dor and Banin, 1990b).

vegetation, and the oxidation state. Organic matter, especially humus, plays an important role in many of the soil properties such as soil aggregation, soil fertility, soil water retention, ion transformation, and soil color. Because organic matter has spectral activity throughout the entire VNIR–SWIR region, especially in the VIS region, workers have extensively studied organic matter from a remote-sensing standpoint (e.g., Kristof et al., 1971). Baumgardner et al. (1970) noted that if the organic matter in soils drops below 2%, it has only minimal effect on the reﬂectance property. Montgomery (1976) indicated that organic matter content as high as 9% did not appear to mask the contribution of other soil parameters to soil reﬂectance. In another study Baumgardner et al. (1985), indicated that organic matter content relates to soil reﬂectance by a curvilinear exponential function. Mathews et al. (1973) found that organic matter correlated with the reﬂectance values at the 0.5- to 1.2-μm range, whereas Beck et al. (1976) suggested that the 0.90- to 1.22-μm region is suited for mapping organic matter in soils. Krishnan et al. (1980) used a slope parameter at around 0.8 μm to predict organic matter content, and Da Costa (1979) found that simulated Landsat channels (bands 4, 5, and 6) yield reﬂectance readings that are signiﬁcantly correlated with organic carbon content in soils. Downey and Byrne (1986) have shown that

194

E. BEN-DOR

Figure 8 The spectral reﬂectance curves of pure organic matter isolated from Alﬁsol and its extracted humic acid.

it is possible to predict both moisture and bulk density of milled peat using the spectral information. The wide spectral range found by different workers to assess organic matter content suggests that organic matter is an important chromophore across the entire spectral region. Figure 8 shows a reﬂectance spectra of pure organic matter (in the NIR–SWIR region) as isolated from an Alﬁsol and the humus compounds extracted from this organic matter (OM). Numerous absorption features exist that relate to the high number of functional groups in the OM. These can all be spectrally explained by combination and vibration modes of organic functional groups (Chen and Inbar, 1994). Vinogradov (1981) developed an exponential model to predict the humus content in the upper horizon of plowed forest soils by using reﬂectance parameters between 0.6 and 0.7 μm for two extreme endmembers (humus-free parent material and humus-enriched soil). Schreier (1977) found an exponential function to account for the organic matter content in soil from reﬂectance spectra. Al-Abbas et al. (1972) used a multispectral scanner, with 12 spectral bands covering the range from 0.4 to 2.6 μm from an altitude of 1200 m, and showed that a polynomial equation will predict the organic matter content from only ﬁve channels. They implemented the equation on a pixel-by-pixel basis to generate an organic content map of a 25-ha ﬁeld. Dalal and Henry (1986) were able to predict the organic matter and total organic nitrogen content in Australian soils using wavelengths in the SWIR region (1.702–2.052 μm), combined with chemical parameters derived from the soils. Using similar methodology Morra et al. (1991) showed that the SWIR region is suitable for identiﬁcation of organic matter

195

Rs (0.1 bar) (%)

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

Wavelength (µm) Figure 9 Spectral curves of three organic soils exhibiting different levels of decomposition. a = ﬁbric; b = hemic; and c = sapric. (After Baumgardner et al., 1985).

composition between 1.726 and 2.426 μm. Evidence that organic matter assessment from soil reﬂectance properties is related to soil texture, and more likely to soil’s clay, was given by Leger et al. (1979) and Al-Abbas (1972). Aber et al. (1990) noted that the organic matter, including its decomposition stage, affects the reﬂectance properties of mineral soil. Baumgardner et al. (1985) demonstrated that three organic soils with different decomposition levels yielded different spectral patterns (Fig. 9). A study by Ben-Dor et al. (1997), using a controlled decomposition process over more than a year, revealed signiﬁcant spectral changes across the entire VIS–NIR–SWIR region as the organic matter aged. Figure 10 shows a typical spectrum of grape marc (CGM) organic matter during a decomposition process that lasted 392 days. Signiﬁcant changes can be seen in the slope values across the VIS-NIR region and within the spectral features across the entire spectrum. Ben-Dor et al. (1997) postulated that some of the analyses traditionally used to assess organic matter content in soils from reﬂectance spectra may be biased by the age factor. As many soils consist of dry vegetation in one degradation stage or another, assessment of the organic matter using reﬂectance spectra should consider the vegetation’ aging status. Although mineral soil consists of a relatively low content of organic matter (around 0–4%), accurate assessment of organic matter content in soils requires high spectral resolution data across the entire VIS–NIR–SWIR region. 4. Water The various forms of water in soils are all active in the VIS–NIR–SWIR region (based on the vibration activity of the OH group) and can be classiﬁed into three major categories: (1) hydration water where it is incorporated into the lattice of

196 Figure 10 The reﬂectance spectra of two endmembers that represent two extreme composting stage t0 = 0 days and t8 = 378 days for grape marc material (CGM). Major wavelengths are annotated on the fresh organic matter (t0 = 0), where the small box shows the spectra of all intermediate decomposition. (From Ben-Dor and Banin, 1995a).

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

197

the mineral (e.g., limonite (Fe2O3·3H2O) and gypsum (CaSO4·4H2O), (2) hygroscopic water which is adsorbed on soil surface areas as a thin layer, and (3) free water which occupies soil pores. Each of these categories inﬂuences the soil spectra differently, providing the capability of identifying the water condition of the soil and well be treated separately below. Three basic fundamentals in the IR regions exist for water molecules, particularly the OH group: υw1-asymmetric stretching, δw-bending, and υw3-symmetric stretching vibrations. Theoretically, in a mixed system of water and minerals, combination modes of these vibrations can yield OH absorption features at around 0.95 μm (very weak), 1.2 μm (weak), 1.4 μm (strong), and 1.9 μm (very strong) related to 2ω1 +υ w3, υ w1+υ w3+υ w, υ w3+2υ w, υ w3+υ w, respectively. (1) The hydration water can be seen in minerals such as gypsum as strong OH absorption features at around 1.4 and 1.9 μm (Hunt et al., 1971b). (2) The hygroscopic (adsorbed) water in soils is adsorbed on the surface areas of clay minerals (especially smectite) and organic matter (especially humus). Early results by Obukhov and Orlov (1964), in the VIS region, showed that the slope of the spectral curve for soils is not affected by wetting, and that the ratio of the reﬂectance of moist soil to that of dry soil remained practically constant. Shields et al. (1968) also pointed out that “moisture has no significant effect on the hue or chroma of several soils.” Peterson (1979) observed linear relationships between bidirectional reﬂectance factors at 0.71 μm of oven-dried soil samples that consisted of water tension between 15 and 0.33 bar. These ﬁndings actually suggest that soil albedo is the ﬁrst factor in the soil spectrum that is altered upon soil wetting (Idso et al., 1975). The primary reason for this is the change of the medium surrounding the particles from air to water which decreases their relative refractive index (Ishida et al., 1991; Twomey et al., 1986). Based on this idea, Ishida et al. (1991) developed a quantitative theoretical model to estimate the effect of soil moisture on soil reﬂection. The shape of soil reﬂectance curves are strongly affected by the presence of water absorption bands at around 1.4 and 1.9 μm, and occasionally weaker absorption bands at around 0.95 and 1.2 μm. Because the hygroscopic water in soil is governed by the atmospheric conditions, the signiﬁcant spectral changes are related to changes in the adsorbed water molecules on the mineral’s surfaces. It is interesting to note that a similar observation was demonstrated by Bowers and Hanks (1965) with soils that consisted of different moisture values (ranging from 0.8 to 20.2%). This observation demonstrates that the gas phase (water vapor in this case) in the soil environment plays a major role in the quantitative assessment of both structural and free water OH. Further insight to this problem was provided by Montgomery and Baumgardner (1974) and Montgomery (1976), who pointed out that it was not possible to quantitatively assess water content in soils because of different dry state conditions under which the soils were measured. Using reﬂectance spectra of several treated smectite

198

E. BEN-DOR

minerals, Cariati et al. (1983) examined shifts of the OH absorption features at 1.4, 1.9, and 2.2 μm. They found that vibration properties of the adsorbed water strongly depend upon the composition of the smectite structure. In another study, Cariati et al. (1981) pointed out that several kinds of interactions are responsible for the vibration properties of the hygrsocopic molecules, where sometimes this may even change with the water content. Because smectite is the most effective clay mineral in the soil environment that affects the reﬂectance spectrum at the major water absorption features, Carieti’s observations may help us to understand the spectral activity of hygroscopic moisture in soils. Further work, however, is still required in order to implement the results obtained for pure smectite in the complex soil system. (3) Free pore water (wet condition) is water that is not in the hygroscopic phase or ﬁlling the entire pore size (saturated condition). The rate of movement of this water into the plant is governed by water tension or water potential gradients in the plant soil system. Water potential is a measure of the water’s ability to do work compared to pure free water, which has zero energy. In soils, water potential is less than that of pure free water due in part to the presence of dissolved salts and the attraction between soil particles and water. Water will ﬂow from areas of high potential to lower potential, and hence ﬂow from the soil to the root and up the plant occurs along potential gradients. In agricultural systems plant growth occurs with soil water potentials between 15 and 0.3 bar tension (note these are actually negative water potentials); however, water tensions in dessert environments are far greater. Baumgardner et al. (1985) studied the reﬂectance spectra of a representative soil (Typic Haplludalf by the USDA) with various water tensions (Fig. 11). As expected, when water tension decreased (and, hence, water content increased) the general albedo decreased, and the area under the strong 1.4- and 1.9-μm water absorption features also decreased. Clark (1981) examined the reﬂectance of montmorilloinie at room temperature for two different water conditions (Fig. 12). Clark (1981) showed that albedo decreased dramatically from dry to wet material. Other changes related to the water and lattice OH can be observed across the entire spectrum as well. Some of these changes are directly related to the total amount of free and adsorbed water and some, to the increase of the spectral reﬂectance fraction of the soil (wet) surface. In kaolinite minerals, a similar trend was observed in two moisture conditions, however, the changes around the water OH absorption features were less pronounced than those in montmorillonite. In montmorillonite, adding water to the sample enhanced the water OH features at 0.94, 1.2, 1.4, and 1.9 μm, because of the relatively high surface area and a corresponding high content of adsorbed water. In kaolinite, the relatively low speciﬁc surface area obscured a similar response, and hence only small changes are noticeable. In the montmorillonite, the lattice–OH features diminished at 2.2 μm, suggesting that the hygrsocopic moisture is a major factor affecting the clay minerals’ (and soil’s) spectra. In soils where the entire

199

Rs (%)

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

Wavelength (µm)

0. 24

Figure 11 Spectra curve of Typic Haplludalf soil at four different moisture tensions: a; oven dry; b; 15 bar; c, 0.3 bar; d, 0.1 bar. (After Baumgardner et al., 1985).

0. 18

MONTMORILLONITE / HALON T = 292 K

0. 06

REFLECTANCE 0. 12

A

B

0. 80

1. 20

2. 00 1. 62 WAVELENGTH IN MICRONS

2. 40

Figure 12 Reﬂectance spectra of montmorillonite with 50% (A) and 90% (B) water mixed in the sample (by weight) at room temperature. (After Clark, 1981).

200

E. BEN-DOR

pore size (or more) is ﬁlled with water [in saturated (or ﬂooded) conditions, respectively], it is more likely that the soil reﬂectance consists of more specular than lambertian components. It should be noted that under remote sensing conditions the water vapor absorptions overlap the soil water signals, and hence, using the previously described relationship, may be questionable. 5. Iron Iron is the most abundant element on the earth as a whole and the fourth most abundant element in the Earth’s crust. The average Fe concentration in the Earth’s crust is 5.09 mass, and the average Fe3+/Fe2+ ratio is 0.53 (Ronov and Yaroshevsky, 1971). The geochemical behavior of iron in the weathering environment is largely determined by its signiﬁcantly higher mobility in the divalent than in the trivalent state. Change in the oxidation state, and consequently in mobility, tends to take place at different soil conditions. Major Fe-bearing minerals in the Earth’s crust are the maﬁc silicates, Fe-sulﬁdes, carbonates, oxides, and smectite clay minerals. All Fe3+ oxides have striking colors ranging among red, yellow, and brown due to selective light absorption in the VIS range caused by transitions in the electron shell. It is well known that even a small amount of iron oxides can change the soil color signiﬁcantly. The red, brown, and yellow “hue” values, all caused by iron, have been widely used in soil classiﬁcation systems in almost all countries and languages. A representative soil spectra with various amounts of total Fe2O3 is presented in Fig. 13. The iron’s feature assignments in the VIS–NIR region result from the electronic transition of iron cations (3+, 2+), either as the main constituent (as in iron oxides) or as impurities (as in iron smectite). Hunt et al. (1971a) summarized the physical mechanism that allows Fe2+ (ferrous) and the Fe3+ (ferric) to be

Figure 13 Reﬂectance spectra of soils consisting of different textures but exhibiting iron absorption bands. a, ﬁne sand, 0.20% Fe2 O3 ; b, sandy loam, 0.64% Fe2 O3; c, silty loam, 0.76% Fe2 O3 ; d, clay, 25.6% Fe2 O3 . (After Baumgardner et al., 1985).

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

201

spectrally active in the VIS–NIR region as follows: The ferrous ion typically produces a common band at around 1 μm due to the spin allowed during transition between Eg and T2g quintet levels into which the D ground-state splits into an octahedral crystal ﬁeld. Other ferrous bands are produced by transitions from 5T2g to 3T1g at 0.55 μm; to 1A1g at around 0.51 μm; to 3T2g at 0.45 μm; and to 3T1g at 0.43 μm. For the ferric ion, the major bands produced in the spectrum are the result of the transition from the 6A1g ground state to 4T1g at 0.87 μm: to 4T2g at 0.7 μm and to either 4A1g or 4Eg at 0.4 μm. Just as organic matter is an important indicator for soils, iron oxides provide signiﬁcant evidence that soil is being formed at a given Earth crust (Schwertmann, 1988). Iron oxide content and species are strongly correlated with the soil weathering process in both the short term and the long term. Transformation of iron oxide in soil often occurs during natural soil conditions. Hematite and goethite are common iron oxides in soils, and their relative content in soils is strongly controlled by soil temperature, water, organic matter, and annual precipitation. Hematitic soils are reddish and goethitic soils are yellowish brown. Their reﬂectance spectra also differ, as can be seen in Fig. 14. Hematite (α-Fe2O3) has Fe3+ ions in octahedral coordination with oxygen. Goethite (α-FeOOH) also has Fe3+ in octahedral coordination, but different site distortions along with oxygen ligand (OH) provide the main absorption features that appear near 0.9 μm.

Figure 14 et al., 1992).

Reﬂectance spectra of representative iron oxide minerals in soils. (From Grove

202

E. BEN-DOR

Lepediocrocite (γ -FeOOH), which is associated with goethite but rarely with hematite, is another common unstable iron oxide found in soils. It mostly appears in the subtropic regime and is often found in the upper subsoil position (Schwertmann, 1988). Maghemite (γ -Fe2O3) is also found in soils, mostly in subtropical and tropical regions and occasionally has been identiﬁed in soils in the humid temperature areas. Ferrihydrite is a highly disordered Fe3+ oxide mineral found in soils in cool or temperate, moist climate areas, characterized by young iron oxide formations and soil environments relatively rich with other compounds (e.g., organic, silica etc.). Iron associated with the structure of clay minerals is also an active chromophore in both the VNIR and the SWIR spectral regions. This can be seen in the nontronite type of mineral already presented in Fig. 4. Based on the structural OH–Fe features of smectite in the SWIR, region Ben-Dor and Banin (1990a) were able to generate a prediction equation to account for the total iron content in a series of smectite minerals. The wavelengths selected automatically by the method they used were 2.2949, 2.2598, 2.2914, and 1.2661 μm. Stoner (1979) also observed a higher correlation between reﬂectance at the 1.55- to 2.32-μm region and the iron content in soils; whereas Coyne et al. (1989) found a linear relationship between total iron content in montmorillonite and the absorbance measured at the 0.6- to 1.1-μm spectral region. Ben-Dor and Banin (1995a) used spectra of 91 arid soils and showed that total iron content in soils (both free and structural iron) can be predicted by multiple linear regression analysis and wavelengths at 1.075, 1.025, and 0.425 μm. Obukhov and Orlov (1964) generated a linear relationship between the reﬂectance values at 0.64 μm and the total percentage of Fe2 O3 in other soils. Taranik and Kruse (1989) were able to show that a binary encoding technique of the spectral slope values across the VIS–NIR spectra region is capable of differentiating a hematite mineral from a mixture of hematite–goethite–jarosite minerals. It is important to mention that an indirect inﬂuence of the iron on the overall spectral characteristics of soils can often occur. In the case of free iron oxides, it is well known that soil particle size is strongly related to the absolute iron oxide content (Ben-Dor and Singer, 1967; Soileau and McCraken, 1967; Stoner and Baumgardner, 1981). As the iron oxide content increases, the size fraction of soil particles increases as well, because of the cementation effects of the free iron oxides. As a result, problems resulting from different scattering effects are introduced within the soil being examined. Moreover, free iron oxides, mostly in their amorphous state, may coat the soil particles with a ﬁlm that prevents natural interaction between the soil particle (clay or nonclay minerals) and the sun’s photons. Karmanova (1981) found that well-crystallized iron compounds had the strongest effect on the spectral reﬂectance of soil and that removal of nonsilicate iron (mostly iron oxides) helps to enhance other chromophores in the soil. In this respect, Kosmas et al. (1984) have demonstrated that a second derivative technique in the VIS region is a feasible approach for differentiating even small features of synthetic goethite from clays and have suggested that such a method may

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

203

be adopted to assess quantities of iron oxide in mixtures. Based on these spectral characteristics Alexander et al. (1999) have shown a possibility to spectrally assess the alteration of soil properties, and Gerbermann and Neher (1979) showed that mixtures of clay and sand of soils can be predicted from the reﬂectance spectra. It can be concluded from the previous discussion that iron in soil is a very strong chromophore, and the determination of its content in clay and soils from the reﬂectance spectra in the entire VIS–NIR–SWIR region is feasible. Based on the complexity of the iron component in the soil environment as well as on the intercorrelation between iron and other soil components, sophisticated methods and relatively high spectral resolution data are absolutely needed to determine iron content from the reﬂectance spectra. 6. Soil Salinity Soil salinity is one of the major factors affecting biomass production and is the principle cause for soil degradation (Csillag et al., 1993). Salt-affected areas cover about 7% of Earth’s land surface (Toth et al., 1991) and are located mostly in arid and semiarid regions (Verma et al., 1994). However, salt-affected soils can also be found in subhumid and coastal zone areas associated with hydrogeological structures. Salts in soils are reported to be Na2CO3, NaHCO3, and NaCl, which are very soluble and mobile components in the soil environment. Typically saline soils have poor structure, are highly erosive, have low fertility, low microbial activity, and other attributes not conducive to plant growth. The spectral signature of saline soils can be a result of the salt itself, or indirectly, from other chromophores related to the presence of the salt (e.g., organic matter, particle size distribution). Hunt and Salisbury (1971c) reported an almost featureless spectrum of halite (NaCl 433B from Kansas). Although salt is spectrally a featureless property, Hick and Russell (1990) raised a hypothesis that there are certain wavelengths, in the VIS–NIR–SWIR region that can provide more accurate information about saline-affected areas. Dwievdi and Sreenivas (1998) applied an image manipulation tool for the study of soil salinity from remote sensing means, whereas Rao et al. (1995) investigated the spectral reﬂectance of salt-affected soils and found some spectral variations. Vegetation is an indirect factor that facilitates detection of salt soils from reﬂectance measurements (Hardisky et al., 1983; Wiegand et al., 1994). Gausman et al. (1970), for example, pointed out that cotton leaves grown in saline soils had a higher chlorophyll content than leaves grown in low-salt soil. Hardisky et al. (1983) used the spectral reﬂectance of a Spartina alterniflora canopy to show a negative correlation between soil salinity and spectral vegetation indices. In the absence of vegetation, the major inﬂuence of salt is on the structure of the upper soil surface. Because no direct signiﬁcant spectral features are found in the VNIR–SWIR region for identifying sodic soil, indirect techniques are thought to be more appropriate for classifying salt-affected areas (Sharma and Bhargava, 1988; Verma

204

E. BEN-DOR

et al., 1994). Salt in water is most likely to affect the hydrogen bond in water molecules causing subtle spectral changes, and based on this Hirschfeld (1985) suggested that high spectral resolution data are required. Support for this idea is given by Szilagyi and Baumgardner (1991), who reported that characterizing the salinity status in soils was feasible with high-resolution laboratory spectra. A relatively high number of spectral channels is also important in order to identify an indirect relationship between salinity and other soil properties that appear to consist of chromophores in the VIS–NIR–SWIR regions. Csillag et al. (1993) analyzed high-resolution spectra taken from about 90 soils in the United States and Hungary against chemical parameters: including clay and organic matter content, pH, and salt. They state that because salinity is such a complex phenomenon, it cannot be attributed to a single soil property. While studying the capability of commercially available Earth-observing optical sensors, they were able to point out that six broad bands in the VIS–NIR–SWIR region best discriminated soil salinity. These six channels were selected solely on the basis of their overall spectral distribution, which provided complete information about salinity status. In another study, Metternicht and Zinck (1996) showed that by using six reﬂective combined Landsat bands, it is possible to discriminate salt- and sodium-affected soil with varying conﬁdent limits. In their study they discussed the nondirect salt effect on the spectral responses of the soils and suggest the fusion of more electromagnetic radiation to spot more shade on this problem. Thus, it can be concluded that it is necessary to look at the entire spectral region in order to evaluate the salinity level in different environments and unknown soil systems. Mougenot et al. (1993) noted that in addition to an increase in reﬂectance with salt content, high-salt content may mask ferric ion absorption in the VIS region. Mougenot concluded that salts are not easily identiﬁed in proportions below 10 or 15%. One more important factor about saline soils is the fact that in modern agriculture, farmers are adding gypsum to sodic soils for soil reclamation (Singh, 1994). The artiﬁcial increase of the gypsum content in such soils may alter the soil reﬂectance spectra signiﬁcantly, and, hence, requires attention. Recently Ben-Dor et al. (1999, 2001) were able to demonstrate a favorable soil mapping capability using a HSR airborne sensor, in which soil salinity was spatially emerged via properties such as electrical conductivity (EC) and pH of the soil pasta. In summary, although salt is not a strong and direct chromophore, its interaction with other soil components (water, structure, iron, and organic matter) makes its assessment possible but complicated. 7. Chemical Chromophores: Summary In order to summarize the overall chemical chromophore activity in soils, we provide a summary of chromophores associated with soil and geological matter from the literature (Fig. 15). Also given are the intensities of each chromophore in the VNIR–SWIR spectral regions as appear in these studies. The current review demonstrates that high-resolution spectral data can provide additional

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

205

Figure 15 The active groups of the soil chromphores, spectrum. For each possible mechanism possible wavelength range and absorption feature intensity are given. The spectrum was generated using information presented in the literature.

information, sometimes quantitative, about soil properties strongly correlated to the chromphores: e.g., water, primary and secondary minerals, organic matter, iron oxides, water, and salt.

B. PHYSICAL PROCESSES The reﬂectance of light from the soil surface is dependent also upon numerous physical processes. Reﬂection, or scattering, is clearly described by Fresnel’s equation and depends upon the angle of incidence radiation and upon

206

E. BEN-DOR

the index of refraction of the materials in question. Generally, physical factors are those parameters which affect soil spectra with regard to Fresnel’s equation, but which do not cause changes in the position of the speciﬁc chemical absorption. These parameters include particle size, sample geometry, viewing angle, radiation intensity, incident angle, and azimuth angle of the source. Changes in these parameters are most likely to affect the shape of the spectral curve through changes in baseline height and absorption feature intensities. In the laboratory, measurement conditions can be maintained constant. In the ﬁeld, several of these parameters are unknown and may introduce problems in accurate assessment of the affect of these parameters on soil spectra. Many studies covering a wide range of materials have shown that particle size differences alter the shape of soil spectra (powdered material) (Baumgardner et al., 1985; Hunt, 1970; Pieters, 1983). Speciﬁcally, Hunt and Salisbury (1970) quantiﬁed affects of about 5% in absolute reﬂectance due to particle size differences and that these changes occurred without altering the position of diagnostic spectral features. Under ﬁeld conditions, aggregate size rather than particle size distributions may be more important in altering soil spectra (Baumgardner et al., 1985; and Orlov, 1966). In the ﬁeld, aggregate size may change over a short time frame due to tillage, soil erosion, eolian accumulation, or physical crust formation (e.g., Jackson et al., 1990). Basically the aggregate size, or more likely roughness, plays a major role in the shape of ﬁeld and airborne soil spectra (e.g., Cierniewski, 1987, 1989). Escadafal and Huete (1991) showed that ﬁve soils with a rough surface exhibited strong anisotropy reﬂectance properties. In this regard macro and micro topography effects may play an important role in the measured reﬂectance, and thus has to be strongly considered for further analysis. A practical solution for evaluating the affects of physical parameters is to evaluate the reﬂectance of a given target relative to a perfect reﬂector measured at the same geometry and viewing angle of the target in question. In reality such conditions are impossible to achieve in the ﬁeld, and complex effects such as particle size effect cannot be absolutely removed by this method. It is postulated that more effort should be expended to account more precisely for physical effects under ﬁeld conditions (from both a spectroscopy and an imaging spectroscopy point of view), such as Pinty et al. (1989) were trying to do with simulating the bidirectional effect over bare soils.

IV. PROBLEMS IN QUANTITATIVE REMOTE SENSING OF SOIL As already mentioned, the remote sensing of soil (and its chromophores) from a far distance introduces signiﬁcant problems that limit a classical quantitative

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

207

assessment. Under natural (ﬁeld) conditions, the reﬂectance spectrum of soil is affected not only by the chemical–physical variation of the sample but also by other external parameters that can make the quantitative (NIRA) assessment of this soil quite questionable (from either a point or an image spectrometer). The factors are variation in soil particle size, limitation of the Sun’s radiation, atmospheric attenuation, low spectral resolution (in HSR only), low signal-to-noise ratio (in HSR only), change in viewing geometry, partial coverage of soil with vegetation, pixel size, subpixel problems, and sensing of only the upper soil surface. If not properly considered, these factors could signiﬁcantly affect the results of the quantitative assessment of the soil in question; hence, a clear understanding of these effects is required. The objective of this section is to review each of these factors and to explain how they can be reduced to allow accurate assessment of soil chromophores from remote-sensing means.

A. ATMOSPHERIC ATTENUATION The atmosphere’s gases and aerosols play a major role in the VIS–NIR–SWIR spectral regions. Across these regions, absorption and scattering of electromagnetic radiation take place. Water vapor, oxygen, carbon dioxide, methane, ozone, nitrous oxides, and carbon monoxide are the components that are spectrally active across approximately half of the VNIR–SWIR regions. Some good models for retrieving gas aerosol interference exist and are widely used by many workers, e.g., MODTRAN-4 (Berk et al., 1999), 5S and 6S codes (Tanre et al., 1986), and ATRAM (Gao et al., 1993) and ACORM (ACORM, 2001). As is to be expected, view and incidence angles as well as adjunctive effects of close terrain are very important in the retrieval of surface reﬂectance. Also good estimation of the solar isolation is important for all atmospheric model corrections (Green and Gao, 1993). Recently Richter (1997) included all of the previously described in one package known as ATCOR-4, and was able to provide the best available (model-base) algorithm for atmospheric correction. It is beyond the scope of this chapter to discuss these models; however, one should be aware that in many cases, the models do not perfectly remove all atmospheric attenuation and may alter the soil spectrum. For example, see Fig. 16a, which provides AVIRIS–ATREM corrected data and ﬁeld spectra of the same area (Gao et al., 1993). The larger number of features found in the AVIRIS spectrum is presumably artifacts of the correction routine. This problem is most likely to appear in hyperchannel data, where discrete absorption features are more pronounced relative to multichannel data, which virtually average small features into one wide value. This demonstrates that NIRA analysis using a hyperspectral remotesensing process may be biased if the atmospheric attenuation is not perfectly removed.

208

E. BEN-DOR 0.6 RETRIEVED MEASURED

a

0.4

Reflectance

.2

0.0 0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

2.2

2.4

1.0

0.8

b

0.6

0.4

0.2

0.0 0.5

1.0

1.5

2.0

2.5

Wavelength (µm) Figure 16 (a) A retrieved (ATREM) reﬂectance spectrum (solid line) from AVIRIS data acquired over an area covered by the mineral sericite in the Northern Grapevine Mountains, California, and a measured reﬂectance spectrum (dotted line) in the ﬁeld using a portable spectrometer (after Gao et al., 1993). (b) Retrieved reﬂectance (EMPIRICAL LINE) spectrum (solid line) from DAIS-7915 data acquired over an area covered by mineral calcite in southern Israel, and a measured reﬂectance spectrum (dotted line) in the ﬁeld using a portable spectrometer.

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

209

The use of empirical methods to correct the atmosphere attenuation to remove such signals is shown in Fig. 16b. In this ﬁgure a calcareous soil spectrum extracted from the DAIS 79 channel scanner data (after applying an empirical line correction technique) is plotted against the ﬁeld spectrum of this exact target. The good match obtained between the two presented spectra demonstrates that it is possible to polish the remaining artifacts into a stage that near-laboratory spectra can be generated from HSR sensors. A recent study by Boardman and Huntington (1997) showed that it is possible to polish these artifacts out from the corrected data (and especially from those obtained by the ATREM code) by applying a new approach, namely, EFFORT that works on a basis of a spectral-averaging technique. Nevertheless, it is still important to take precaution, no matter what method is used, in order to avoid artifacts from entering into any soil classiﬁcation algorithm applied to the atmospheric-corrected HSR data. To illustrate the spectral regions under which atmospheric attenuation could affect the soil spectrum, we provide in Fig. 17 a reﬂectance spectrum of a playa taken by an AVIRIS sensor over Rogers Dry Lake, California, with a correction to the solar effect but without removing atmospheric attenuation. The VIS region

Figure 17 An AVIRIS spectrum of a playa target at Rogers Dry Lake after removing the solar effect. Across the spectrum the major gas absorption’s absorption features are annotated to show the area where atmospheric attenuations might overlap with soil features. (From Ben-Dor et al., 1998).

210

E. BEN-DOR

is affected by aerosol scattering (monotonous decay from 0.4 to 0.8 μm) and absorption of ozone (around 0.6 μm), water vapor (0.73, 0.82 μm), and oxygen (0.76 μm). The NIR–SWIR region is affected by absorption of water vapor (0.94, 1.14, 1.38, and 1.88 μm), carbon dioxides (around 1.56, 2.01, and 2.08 μm), and methane (2.35 μm). As discussed previously, even weak spectral features in the soil spectrum may consist of very useful information. Therefore, great caution must be taken before applying any quantitative models to soil reﬂectance spectra derived from air- or space-borne hyperchannel sensors. The validation of the corrected data is an essential criterion for ensuring that the reﬂectance spectrum consists of reliable soil information.

B. SPECTRAL RESOLUTION AND NUMBER OF CHANNELS Spectral resolution refers to the number of channels the sensor has and to the function that explains each of the channel’s responses. Assuming that this function is a gaussian-like spectrum, the width at the half-maximum height [full halfmaximum width (FHMW)] may represent the channel resolution. The lower this value the higher the resolution of a given channel. Figure 18 provides a theoretical signal channel response function that represents the FHMW values, whereas Fig. 19 provides the spectral response plots of AVIRIS and the Thematic Mapper (LANDSAT-5) channels. The lower FHMW values of the AVIRIS, along with its

Figure 18 A scheme showing the full half-maximum width (FHMW) principle for estimating the spectral resolution of an imaging spectrometer.

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

211

Figure 19 A schematic detailed spectrum and the band response function of AVIRIS (224 bands) and the Thematic Mapper LANDSAT-5 (6 bands) sensors. (From Goetz et al., 1993).

relatively high number of channels (224), make this sensor more accurate in terms of spectral resolution than the LANDSAT-5 sensor. Goetz (1987) concluded that across the 0.4- to 2.5-μm spectral region, a 10-nm sampling interval is sufﬁcient for describing salient features in the reﬂectance spectra of rocks, minerals, organic matter plants, and suspended matter in water bodies. The previous ﬁgure actually demonstrates the need for high-spectral-resolution reﬂectance data in the VIS–NIR–SWIR region. This need was recognized soon after the ﬁrst ERTS-1 mission in 1972. Satellite sensors were designed to have channels at spectral regions that are generally active for Earth targets. The HSR strategy was developed because of the requirement to get more channels across the spectrum so as to “no longer have to choose the best possible bands” for each application. The effect of losing spectral information by using low spectral resolution sensors can be seen in Fig. 20 [taken from Ben-Dor and Banin (1995a)]. In this spectrum, a selected soil from Israel (Haploxeralf) is represented by high (3113 channels) and low (6 channels) spectra. Although these spectra were composed under laboratory conditions, it can be clearly seen that the spectral information is dramatically diminished going from high to low spectral resolution (notice the spectral absorption of OH in the clay lattice (2.2 μm) and in the soil surface (1.4, 1.9 μm). These

212

E. BEN-DOR

Figure 20 A laboratory spectrum of an Israeli soil (Lithic Rhodoxeralfs) after band degradation to simulate lower spectral resolution. (After Ben-Dor and Banin, 1995a).

observations suggest that other invisible absorption signatures (e.g., of organic matter) may be lost by the spectra resolution degradation, which may cause problems in applying the NIRA strategy. It is, however, questionable what level of spectral resolution one should expect for NIRA soil remote-sensing purposes. Ben-Dor and Banin (1995a–c) have found that for each soil property, a different number

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

213

of channels is required for optimal spectral prediction of each property (ranging from 3113 to 25 spectral channels). It is assumed that under real remote-sensing conditions, this restriction would still be valid, and a combination of a sensor–soil property and external conditions–soil property chain might be important. In other words, since there is still no information regarding NIR and real remote-sensing analysis, this issue has to be examined in future studies. For soil applications, airborne- and space-borne-imaging spectrometers should consist of a reasonable number of spectral channels across the entire VNIR–SWIR region, which covers the spectrally active regions of all chromophores with a reasonable bandwidth (Carrere, 1991). Price (1991) believes that relatively low spectral channels (15–20) with 0.04- to 0.10-μm bandwidths and high signal-tonoise ratios are those that promise better remote-sensing capabilities of soils. Goetz and Herring (1989) preferred more spectral channels (192) but a wider bandwidth (about 10 nm) to permit diagnostic evaluation of speciﬁc features across the entire VNIR–SWIR region. We believe that for quantitative analysis of soil spectra, the optimal bandwidth and number of channels may be strongly dependent on the soil population and the property examined. There is no doubt, however, that high signalto-noise ratio is a crucial factor in quantitative analysis of soil spectra derived from both air and space measurements.

C. SIGNAL TO NOISE Signal-to-noise ratio is the major parameter by which the received spectral information can be judged. A typical laboratory reﬂectance spectrum (that has had noise artiﬁcially added) of a selected soil mineral is shown in Fig. 21. It can be clearly seen that as the noise factor increases (and hence the signal-to-noise ratio decreases) the quality of the data gets poorer. In the NIR analysis where every small spectral signature is important, such noises may make reliable analysis impossible. In reality, remote-sensing devices provide SNR values far beyond any that can be achieved in the laboratory. Several methods are known to estimate the SNR values of remote-sensing images (Gao, 1993). A common and very simple method was proposed by Kaufmann et al. (1991). In this method, a uniform area is selected from the image in question, and the mean values (simulated signals) for every wavelength are divided by the standard deviation of the selected pixels (simulated noise) to yield an SNR spectrum. Ben-Dor et al. (1994) showed values of 10–100 for the GER sensor (63 channels across the VIS–NIR–SWIR spectral region) for desert areas, and Ben-Dor et al. (2000) showed values of 15–150 for the CASI (48 spectral channels across the VIS–NIR spectral region) in urban areas. The equivalent laboratory SNR values are at the range of 2000 and above. Figure 22 shows an example of an SNR spectrum extracted from a soil in a GER image. It is postulated that the SNR value is not a constant parameter across the spectrum, and

214

E. BEN-DOR

Figure 21 The effect of adding noise to a pure reﬂectance spectrum in a pure kaolinite mineral laboratory spectrum. (From Goetz et al., 1993).

for each channel a different value exists. In general, the noise sources are composed of photon noise (Np: obtained from the target, background, and atmosphere), the detector noise (Nd: obtained from the photon counting device), and postdetector electronic noise (Ne: obtained from the sensor’s electronic arrangement). The total noise is calculated by noise = N2p + N2d + N2e . The more the photons interact with the sensor, the more the SNR value increases. In this regard Np can result from atmospheric attenuation (block of the Sun’s radiation in the Sun–target–sensor path) and a low dwell time over the target

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

215

Figure 22 The signal-to-noise ratio of the GER airborne imaging sensor over soil targets as estimated from the data based on Kaufmann et al. (1991) method. (From Ben-Dor and Levin, 2000).

(e.g., whiskbroom sensor). A lower SNR can also be a result of detector noise (Nd) and of electronic noise (Ne). For conducting quantitative analysis in general and of NIR in particular, the SNR values must be on a high enough level to provide the analysis of reliable spectral signals. Recently there has been tremendous progress in achieving high signal-to-noise ratios from airborne HSR sensors. Today the JPL AVIRIS sensor (Green et al., 1999) and the Hi-Vista HyMap sensor (Cocks et al., 1998) can be declared the new generation of HSR sensors, which can provide near-laboratory-quality data from a far distance having SNR values close to 1000.

D. PIXEL SIZE AND SAMPLING TECHNIQUES For a favorable NIR remote-sensing-based analysis, there is a strong need for an accurate and representative soil-sampling technique. This is because the soil samples that are brought to the laboratory for analytical analysis must be precisely allocated both on the image and on the ground. Misallocation of the soil samples at the calibration stage can signiﬁcantly bias the validation stage and hence prevent any practical utilization. This suggests that in addition to the regular NIR technique developed in the laboratory for well-known samples, a tremendous effort must be given to the geometrical positioning of ﬁeld samples, if remote-sensing/GIS

216

E. BEN-DOR

attitude is considered. Doing so may be the key factor in being able to apply NIR analysis to remote-sensing data. If this is not properly done, there will always be the question of whether the low performance is a result of the incorrect selection of spatial and spectral pairs or of some other mentioned potential problems. Because of sensor motion during the data acquisition process, a given pixel in an image cannot be represented by exact ground pixel geometry. Furthermore, because of the sensor movement in the yaw, pitch, and role axis, the acquired image could be geometrically distorted and hence prevented from producing orthophoto projections for thematic map missions. An image pixel consists of information from surrounding ground areas that make its spectral information rather complex. The image pixel contains much more information than its represented ground pixel, and hence makes quantitative analysis quite complicated. The conclusion drawn from this is that soil ground sampling for NIRA analysis (either for calibration or validation purposes) must be carefully done. It is generally accepted that an area measuring 4 × 4 pixels around a selected point represents a favorable ground area for an image pixel to later represent the ground. This requires, therefore, the collection of soil samples of one point from a 4 × 4 pixel size area. It is strongly recommended that the upper 2 cm of an area be randomly sampled, making complete coverage of the area and taking soil from a good number of points in the area and then mixing them together to yield one soil sample. Geometric rectiﬁcation of a distorted image for map projection can be done in several ways. The common methods are an image-to-map or image-to-image rectiﬁcation process, where several ground points from one image (distorted) are registered on another (rectiﬁed) data set (image or map) and a mathematical algorithm is applied to rectify the distorted image (Ben-Dor et al., 1997). New techniques, using GPS and INS systems to rectify image data during the ﬂight are being used today (Haalan et al., 1997). These devices allocate on real time the exact coordination of every pixel on the image and provide a near-orthophoto projection of the image rapidly and accurately. In this regard, the use of a ground GPS device is extremely important as well; and the better the resolution, the better the results. GPS devices can provide varied accuracy, depending on the instrument and the data collection technique used. In reality it is possible to get lower than a 1-m resolution. In fact the ground resolution for the soil environment may cover four pixels. This is because soil is a low-spatial-frequency environment. In highspatial-frequency domains, such as an urban environment, higher resolution on the ground is required.

E. MEASUREMENT GEOMETRY Measuring ﬁeld reﬂectance based on Sun illumination is a problematic issue. This is because ﬁeld roughness may introduce non-lambertian reﬂectance behavior,

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

217

which results in nonstandardized ﬁeld measurements, which are nonrepresentative as compared to the known spectral library. (This problem is similar to the sample preparation done in the laboratory.) This problem exists for ﬁeld point spectrometers as well as for imaging devices. In point spectrometry there is always the problem of the small ﬁeld of view (FOV), which can yield a nonrepresentative spectrum of the soil in question. Although soil reﬂectance is not only a function of the wavelength but also a function of the incident and viewing angle of the illumination, only limited studies have investigated the soil’s bidirectional reﬂectance effects in conjunction with multiple illumination or viewing directions (Irons et al., 1989). Irons et al. (1989) remotely sensed bare soil’s bidirectional reﬂectance using a pointable airborne sensor. A rough, recently plowed bare soil surface was found to scatter light back in the antisolar direction more strongly than did a smooth soil surface. This study demonstrated the importance of surface roughness in determining the directional distribution of soil reﬂectance in the ﬁeld. Valeriano et al. (1995) recently demonstrated the same thing using 14 soils from Brazil. The bidirectional effect applies to image data as well, and there is a strong need to quantify it and apply its effect to real remote-sensing data. Cierniewski (1987) developed a model to account for soil roughness based on the soil reﬂectance parameter, illumination properties, and viewing geometry for both forward and backward slopes. The model shows that the shadowing coefﬁcient of the soil surface decreases with a decrease in soil roughness. For soils on forward slopes of more than 20◦ the shadowing coefﬁcient also decreases when the solar altitude increases in the full interval of the Sun’s altitude, ranging from 0 to 90◦ . The model indicates that this relationship for soil slopes having a surface roughness lower than 0.5 for a speciﬁed range of solar altitude may be the opposite. Using empirical observations of smooth soil surfaces, Cierniewski showed that the model closely agreed with ﬁeld observations. A brief and good summary on the multipleand single-scattering models of soil particles with respect to the roughness effect is given by Irons et al. (1989). Viewing angle is also a parameter that signiﬁcantly affects the reﬂectance spectra of any object on the Earth’s surface (Escadafal and Huete, 1991); however, there are various models to correct for changes in this parameter (Egbert and Ulaby, 1972; Hapke, 1993; Liang and Townshend, 1996). In soils there is a relationship between a shadowing coefﬁcient (simulated), soil roughness factor, solar altitude, and the slope of the soil surface. As the solar altitude decreases, the shadow parameter increases for any rough soil surface factor parameter and for both forward and backward slopes. It is also agreed that as the soil becomes rougher, the shadow parameter becomes signiﬁcantly magniﬁed for almost any slope direction. Various authors have attempted to model the geometry of the soil surface to evaluate the effect of different viewing angles. The models considered the soil surface as being composed of spheres (Cierniewski, 1987, 1989), cylinders (Den Dulk, 1989), cubes (Escadafal, 1989) or blocks, ripples, and paraboloids (Mulders et al., 1992). The predictions from these models agreed rather well with empirical observations.

218

E. BEN-DOR

The validity of these models depends on their ability to accurately quantify the soil surface in terms of the various shapes, the reﬂectance differences of the surface constituents, and the reﬂectance attributed to shadow. Epema (1992) found differences of about 10% in nadir reﬂectance of moderately rough soils with solar zenith angles of 30 and 60◦ . Kimes (1983) and Kimes et al. (1985) presented results of the directional distribution of reﬂectance of bare and vegetated surfaces. For the reﬂection of bare surfaces, back scattering in the illumination direction dominated. Coulson (1966) found the same under laboratory conditions for most minerals, but he observed a dominant forward scattering for low-absorbing materials like gypsum and quartz. Under ﬁeld conditions, however, even for gypsum and quartz surfaces, back scattering dominates.

V. PARAMETERS AFFECTING THE REMOTE SENSING OF SOIL A. VEGETATION COVERAGE Soil is a growing environment for green plants (natural and agricultural) and a sink for decomposing tissues of vegetation and fauna. Because large parts of the world’s soils are vegetated (green or dry), the problem of deriving soil spectra from the mixture of soil–vegetation signals is complex (Huete and Escadafal, 1986). Siegal and Goetz (1977) postulated that “the effect of naturally occurring vegetation on spectral reflectance of earth materials is a subject that deserves attention.” At one extreme are situations where the canopy cover is so dense that reﬂectance from soils is too difﬁcult to interpret. In such cases approaches that apply geostatistical methods to vegetation-free soils in the area to infer the possible soil variation beneath the canopy are used (Ben-Dor et al., 2001). In situations where vegetation cover is only partial, a mixed signal from soil and vegetation occurs, and to some extent the chemical and physical components can be resolved (Murphy and Wadge, 1994). In soil–vegetation mixtures, nonlinear models are typically used to resolve issues of the soil spectra (Goetz, 1992; Ray and Murray, 1996). Otterman et al. (1995) noted that the relationship between the amount, type, and architecture of a vegetation cover and the reﬂectance properties of the underlying soil are important issues (e.g., low albedo soils are those most signiﬁcantly affected by vegetation). The 0.68- to 1.3-μm spectral region, of soils is the region most severely affected by green vegetation as a result of the steep reﬂectance increase caused by vegetation (see, e.g., Ammer et al., 1999). Dry vegetation does not alter the spectrum in the VNIR region, except for changing the albedo; whereas in the SWIR region, signiﬁcant vegetation effects are related to cellulose, lignin, and water. The low reﬂectance of green vegetation beyond 1.4 μm

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

219

indicates that if a soil–vegetation mixture exists, most of the spectral information relates to rock and soil types (Siegal and Goetz, 1977). Two chromophores—water and organic matter—which exist in both plant and soil material, can complicate the interpretation of spectra, particularly in the SWIR region. In the green vegetation– soil mixture, liquid water of green and dry vegetation can overlap with the soil water forms. Also signatures of lignin, cellulose, and protein can signiﬁcantly affect the soil components in the soil–vegetation mixture. Murphy and Wadge (1994) showed in one case that although live vegetation has a greater impact on the SWIR region of soil spectra, dead vegetation had a greater impact on the 2.2-μm absorption features (see, for example, the reﬂectance spectra of pure organic matter given in Fig. 9). Murphy and Wadge (1994) concluded that dead vegetation tissues had a greater impact on soil spectra than live vegetation, and they suggested greater consideration by workers regarding the effect of dead vegetation on soil’s spectra. From a vegetation point of view, Tucker and Miller (1977) postulated that “remotely sensed data of vegetated surfaces could be analyzed more accurately if the contribution of the underlying soils spectra are known.” Tueller (1987), and Smith et al. (1990) noted that it is difﬁcult to extract vegetation information when the coverage is less than 30–40%. The normalized differential vegetation (NDVI) is a parameter commonly used to estimate the cover of green vegetation from satellite and airborne data. The index, which is based on the normalized difference between the NIR and the VIS reﬂectance values, is very sensitive to soil background, atmosphere, and Sun angle conditions. Based on that background, Huete (1988) developed a new index called soil-adjusted vegetation index (SAVI), which accounts for soil brightness and shadows, and more recently (Liu and Huete, 1995) presented another index, the modiﬁed NDVI (MNDVI), which accounts for atmospheric attenuation as well. The SAVI has been shown to signiﬁcantly minimize soil-related problems in nadir measurements over a variety of plant canopies and densities and in data derived from canopy radiant transfer models (Huete and Escadafal, 1991). More precise models take into account the vegetation architecture (Otterman et al., 1995) or contain additional correction factors (Rondeaux et al., 1996). Richardson et al. (1975) developed three plant canopy models for extracting plant, soil, and shadow reﬂectance components of a cropped ﬁeld. Using such models, Murphy and Wadge (1994) were able to separate soil and vegetation spectra by using GER 63-channel imaging spectrometer data (Ben-Dor and Kruse, 1995). Roberts et al. (1993) also incorporated an unmixing procedure to discriminate vegetation, litter, and soils using AVIRIS 224-channel imaging spectrometer data (Vane et al., 1993) and were able to account for different soil types using a residual spectrum technique. Another study that used unmixing approach with AVIRIS data for soil information extraction has been conducted by Accioly et al. (1998). It can be concluded that soil spectral signatures can be extracted from areas that are partially covered by decaying or live vegetation, however, caution must be taken when assessing the “true” soil reﬂectance spectra in a vegetation–soil

220

E. BEN-DOR

mixture. Using a less informative imaging sensor (GER with 63 channels) Drake (1991) showed that it is feasible to discriminate rocks soil and vegetation communities in large areas. Analytical solutions can be applied in order to extract soil signals from soil– vegetation mixtures (Beirwirth, 1990). However, in cases where the vegetation coverage is complete, these solutions cannot be applied. In this case, a synergy between spatial models (GIS-based) and remote-sensing methods can provide indirect information about the underlying soil. In this regard, Ben-Dor et al. (2001) were able to map organic matter, soil salinity, and soil moisture in a heavily vegetated area. This was done by using points in the soil environment that surrounded the vegetated area. Each point was determined by NIRA analysis as applied to HSR data. For this procedure, more than 100 points were used and an interpolation geostatistical method, namely, the inverse distance weighting interpolation (IDW). This technique serves as a model in the MapInfo software (MapInfo User’s Guide, 1997).

B. SOIL CRUST AND SURFACE In the remote sensing of the Earth using Sun radiation, only the upper 50 μm of the soil is sensed. Accordingly, all processes that take place in this thin layer must be signiﬁcantly considered in quantitative remote sensing. This is very important with regard to soils, where the entire soil body (0–2 m) (strongly required for soil mapping classiﬁcation) cannot be sensed. Two crusts that form on the upper soil horizon and signiﬁcantly change the spectral reﬂectance of the soils will be reviewed here. 1. Biogenic Crust A major thin-layer component in arid soil areas, usually ignored by workers, is the biogenic crust. This issue has received more and more attention recently, and its importance to the explanation of anomalies in ﬁeld soil spectra and satellite data has been shown (Pinker and Karnieli, 1995). The biogenic crust mainly consists of lower, nonvascular plants (microphytic) covering the upper soil surface in a thin layer (Rogers and Langer, 1972; West, 1990). The microphytic community consists of mosses, lichens, algae, fungi, cyanobacteria, and bacteria. Each of these groups has pigments that are spectrally active in the VIS region under certain environmental conditions and thus can mask soil features and, more seriously, can be interpreted as the soil signature. O’Neill (1994) showed that spectral features (between 2.08 and 2.10 μm) of a soil could be attributed to the microphytic crust and speculated that this was due to cellulose. Karnieli and Tsoar (1994) showed that the microphytic crust caused a decrease in the overall albedo in the soils, which

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

221

led to the false identiﬁcation of anomalies in arid soils. The spectral response related to the biogenic crust permits linear mixing models, unlike the complex architecture of higher vegetation, which requires nonlinear models to analyze mixed signals. Karnieli et al. (1998) have shown that the biogenic crust consists of organic matter characteristics in the SWIR spectral region and can bias the remote-sensing composition of crusted soils. It is obvious that the biogenic crust status has to be more carefully considered by workers, especially when trying to assess the vegetation or organic matter coverage over a given area. In this regard, more quantitative studies are required in order to fully account for the biogenic crust’s effect on the soil spectra. 2. Physical Crust As Baumgardner (1985) stated “early remote sensing researchers of soils recognized the fact that soils often formed surface crust that could make a soil appear dry when it was actually wet.” Cipra et al. (1980) observed higher reﬂectance values in the VIS region from crusted soil relative to the same soil with the crust broken. Soil crust and cover can be formed by different processes. The biogenic crust, as discussed earlier, is one example of such interference. Eolian material and desert varnish are others. A lithosphere crust that is often found in soil is called “rain crust.” This crust is formed by raindrops (Morin et al., 1981), which cause a segregation of ﬁne particles at the surface of the soil. This can increase runoff and lead to soil erosion. The crusting effect is more pronounced in saline soils and has been well studied in relation to the mineralogical and chemical changes of the soil surface (Shainberg, 1992). The immediate observation after a rainstorm is an enhancement of “hue” and “value” values of the soil color because of an increase in the ﬁne particle fraction on the surface. One can assume that the reﬂectance spectrum of the “rain crust” would be totally different from that of the original soils, because it contains a greater clay fraction with a different textural component. In the literature, the issue of “rain crust” as it affects the spectral signature of soils has not received considerable attention; however, we encourage workers to consider this problem in their studies. Recently Goldshlager (2001a,b) were able to show, under controlled rainstorm conditions, that in three different soils from Israel the reﬂectance spectra changes dramatically in the NIR–SWIR region in two mechanisms: (1) the albedo increases as the rain amount increases (similar to what has been clearly observed by other workers); (2) the clay spectral signatures are enhanced as the rain amount increases. The latter actually shows that the physical crust may produce a spectral error of about 15%. This suggests that remote sensing of crusted soils might provide biased information regarding the soil composition. In this regard sandy soils may be interpreted as clay-like soils, and if NIR analysis is applied, wrong conclusions could be drawn. This observation actually stresses the view that in remote sensing, the entire proﬁle cannot be assessed, and thus

222

E. BEN-DOR

a classical soil mapping, where all of the soil’s proﬁle properties are taken into account cannot be done via remote sensing means.

VI. HIGH-SPECTRAL-RESOLUTION SENSORS A. CURRENT AND FUTURE SENSORS A detailed overview of airborne-imaging spectrometers (past and present) and their performance for use in the remote sensing of soils can be found in Schaepman’s Comprehensive List (mentioned earlier in this chapter: http://www/geo/unizh.ch/∼schaep/research/apec/is list.html). Another important review of visible and infrared sensors, presented along with several detailed case studies of HSR sensors is provided by Kruse et al. (1998). In this review Kruse presents the most popular sensors used in orbit and provides a comprehensive description of present airborne HSR sensors, including vendors’ addresses and information channels for potential users. Kruse also provides information about planned systems (for 1999), some of which are already available today (for 2000). In this regard ASTER and MODIS, two of NASA’s sensors onboard the Terra spacecraft, were declared by NASA as “ready for business” after a careful examination and validation of the data transmitted from orbit. These sensors provide invalid HSR information to the ground, and it is anticipated that new studies will soon prove this. There are only a few types of point spectrometers, and they were reviewed earlier (ASD, GER, PIMA, LICOR). Although not frequently used, point spectrometers can be operated from the air. For that purpose, however, a short response time and large-capacity output are required. This is basically because of the aircraft’s motion, which dictates data acquisition at high-frequency domains. Another problem is the need for an accurate registration tool for the enormous number of spectral points acquired by airborne devices. As noted earlier, Karnieli and his colleagues (personal communication) were able to combine a video camera with a point ASD spectrometer and ﬂy it on a Cesna light aircraft for use in mapping soils from the air. They showed that it is possible to use this technique for soil vegetation applications and demonstrated that it is a cost-effective technique for soil applications. Solar-based remote sensing in general and remote sensing that relies on point or imaging spectrometers in particular suffer from the fact that some photons are absorbed almost totally into the atmosphere (e.g., at around 1.4 and 1.9 μm where water molecules are active). Further, in the SWIR region the Sun’s energy is weak, and important spectral information might be contaminated with noise (Fig. 23). The Earth’s rotation, which causes the illumination source to be nonconstant from one measurement to another, and cloud coverage, which prevents photons from interacting with the ground, are also factors that require attention before applying

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

Figure 23 et al., 1992).

223

The sun spectrum and the associated gas absorption in the atmosphere. (From Goetz

traditionial analyses to data acquired through solar-based remote-sensing of soils. Using artiﬁcial illumination can be a solution for ﬁeld measurements only. This is because it is impossible to illuminate large areas with a Sun simulator. Artiﬁcial illumination sources are used in several spectrometers, such as in the PIMA (sensitive to the SWIR region) and in the ASD (sensitive to the VIS–NIR–SWIR). Both spectrometers are portable and, via the artiﬁcial illumination sources, provide new ﬁeld capabilities.

B. COST AND AVAILABILITY An HSR remote sensing of soils is still a cost-effective mission. This is largely because of the investment in advanced technology (development and maintenance). Many potential users still have not been exposed to this medium and wrongly conclude that traditionial methods are always costly missions. This conclusion does not take into account the vast amount of information one can get in a matter of seconds using remote-sensing tools or the tools’ ability to cover wide areas and to provide spatial overviews. To illustrate this point, a calculation was made

224

E. BEN-DOR

using the most sophisticated and accurate HSR sensor known—the HyMAP. This calculation was provided by Dr. L. de Gasparis from the HiVISTA company, which manufactures and operates the sensor worldwide (personal communication for 2000). The cost for one HyMAP ﬂight could reach as much as $40,000, which is an unrealistic cost to be beared by a single farmer or even a group of farmers. However, if the ﬂight were planned properly and the end users were well organized, the price could be drastically decreased, making HSR use feasible. A calculation example: The scanner in the 5-m resolution mode can cover up to 2000 km2 in a day if the ﬂight is well planned. This would cost about U.S. $25,000, not including the aircraft. If one just covered half of this area, it would come to about $25.00/km2 (or 25 cents/acre!), the least costly for geophysical data or any other agricultural data. However, since there are only a few scanners worldwide, the cost of getting the scanner to any place on the globe would probably run $40,000 to $50,000. If this cost could be distributed over a large area, it would not greatly impact the cost per square kilometer. Thus, large surveys, multinational missions, and wellplanned organization of such activities are indeed key factors for making the use of the airborne HSR feasible. Also massive advertising of the HSR capabilities over potential end users is important. Another solution for reducing the data acquisition cost would be to put an HSR sensor in orbit and routinely acquire data at a low cost. An example of this is the cost of LANDSAT images, which has been signiﬁcantly reduced after the successive launch of LANDSAT-7 in September 1999. Over the last decade, LANDSAT-5 data cost about $3500 per scene, whereas the new LANDSAT-7 cost per scene has been established to be about $700. The per-scene cost of LANDSAT-5 has been reduced to about $1500, and it is anticipated that soon the entire market will drift downward accordingly. It will require reasonable cost to cause potential end users to use remote-sensing data on a routine basis for their speciﬁc needs. In this regard, the accuracy and the additional information one can get from the remote sensing tool play a secondary role. Comparisons of satellite data to airborne data can also be made on a temporal basis. Whereas airborne mechanisms can practically acquire data on any (clear) day, polar Sun synchronic satellites can acquire data only on certain (clear) days. This limitation is based on the satellite’s path and viewing angle, and is measured by the amount of time that passes between two data acquisition events of the exact geographical location (known also as a “revisit time”). A typical revisit time is about 14 days, which could be a limitation for monitoring the environment. However, for mapping soil properties, this period is more than sufﬁcient, because soil properties do not tend to change rapidly. The 1998 and 2001 launch failures to mount an HSR sensor in orbit Lewis and ObView-4 has not eliminated efforts to place another HSR sensor in orbit. At this moment three HSR devices are being operated from orbit (MODIS and ASTER), and they have recently been joined by other new sensors that also orbit the Earth (e.g., HyPERION of the United States and PROBA of the ESA, see Section I). Although these instruments are more scientiﬁc

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

225

than commercial, there is no doubt that they will serve as invaluable prototypes for future HSR-soil applications from space (Vincent 1997). Recently, a considerable amount of effort has been made to make HSR remote sensing an effective tool for commercial utilization (see Kruse, 1998). The advances achieved in HSR technology and the knowledge of how to process data to correct for previous obstacles, for the ﬁrst time, from a practical standpoint, provide remote-sensing users a tool that can provide a reliable, near-laboratory-quality spectrum of every pixel from a far distance. As has already been discussed, understanding soil spectra principles and their limitations is a crucial step in applying quantitative applications (such as soilNIRA) to HSR data. As discussed earlier, information about soils from reﬂectance spectra can be derived in the visible-near-infrared (VIS–NIR, 0.4–1.1 μm) and short-wave-infrared (SWIR, 1.1–2.5 μm) spectral regions. However, also available is the thermal emission across the TIR (3–5 μm, 8–12 μm), which is also an HSR-operative region and hence can provide additional information for the thematic–analytical processes performed on the physical–chemical data of soils.

VII. GENERAL ANALYTICAL METHODS Methods for quantitative remote-sensing applications rely on the spectral information received by the sensor. The more detailed this information, the more informative the results extracted. The lowest spectral resolution consists of three broad channels (with low spatial resolution) or one panchromatic channel (with high spatial resolution). Spectral information can be visually enhanced by coupling the screen guns of three computers with three spectral channels, coded red, green, and blue (RGB). This color composite image can yield a false color representation of the area in question, and often depicts spatial variation never picked up by the naked eye. The band ratio technique is another common method in which two images (each referring to a spectral channel) are divided on a pixel-by-pixel basis. This method has proven to be an effective tool for discrimination of surface units and also for minimizing shadowing effects (Lillesand and Kiefer, 1994). These examples (color composites and band ratio techniques) represent qualitative image enhancement; however, they cannot provide quantitative information. For quantitative remote-sensing applications more channels have to be incorporated in the analysis, and the classiﬁcation methods become more complicated. Principle and canonical component analysis (PCA) (Lillesand and Kiefer, 1994) or minimum noise fraction (MNF) transformations (Boardman and Kruse, 1994) are techniques designated to remove or reduce spectral redundancy and enhance the important spectral information. The procedures compress all of the information contained in an original n-channel data set into fewer regions, “new channels,” or components. The components are then used in lieu of the original data, and the color

226

E. BEN-DOR

composites from this process yield new and useful information that cannot be extracted in other described ways. Whereas the previously-mentioned methods can overestimate spectral information and are not capable of dealing with speciﬁc absorption features; methods for feature extraction and comparison are available. The spectral methods can be divided, according to Mustard and Sunshine (1999), into three categories: (1) characterization of the position, strength, and shape of absorption features; (2) comparison of the complete shape of the absorption features from a remote-sensing data set against those in a known spectral library; and (3) quantitative deconvolution of overlapping superimposed absorptions. The last category represents the most sophisticated quantitative approach applied to HSR data. In this regard mineral combinations and abundances are computed with and without prior knowledge of the possible components that appear in the examined area. In this category its is assumed that an individual pixel often contains more than one land cover type. Pixels with more than one land cover type are referred to as “mixed pixels,” while those containing only one type are called “pure pixels.” Subpixel analysis is therefore a leading methodology for quantitative remote-sensing analysis, especially in the area of HSR (Merickle et al., 1984). Greater accuracy can be achieved in the estimation of the land cover composition of a scene if each mixed pixel can be decomposed and the proportion of its component cover types (known as “endmembers”) can be determined in a process known as “unmixing” (Frans and Schowengerdt, 1999). Over the past decades the unmixing process has been developed intensively, and many related methods have been produced. Several types of models have been proposed, notably, linear, probabilistic, geometric or geometric–optical, stochastic geometric, and fuzzy models. In the linear models the reﬂectance of a pixel in each spectral band is expressed as a linear combination of the reﬂectances of its component endmembers in that spectral band, weighed by their respective surface proportions (Bierwirth, 1990; Cross et al., 1991; Duncan et al., 1990; Huete, 1986; Merickel et al., 1984; Novo and Shimabukuro, 1994). Probabilistic models are based on several techniques of probability, such as maximum likelihood procedures (Horowitz et al., 1975) and approximate maximum likelihood techniques (Marsh et al., 1980). In the geometric or geometric–optical models, the geometry of three crowns, their distribution, and the direction of solar illumination (Neckel and Labs, 1984) are taken into account in order to evaluate the relative proportions of crowns, shadow, and background in the pixel (Gilabert et al., 1994; Hiernaux, 1991). The stochastic geometric model developed by Jasinski and Eagleson (1989, 1990) is a special case of the geometric model in which the scene’s geometric parameters are treated variably in order to absorb the random variables in their spatial structure. The fuzzy models employed the principles of the fuzzy set theory, which makes it possible for an element to have partial membership in more than one set (Foody and Cox, 1994; Wang, 1990). A comprehensive review of mixture modeling can be found in Ichoku and Karnieli (1996). As Schot et al. (1999) noted, today’s remote-sensing algorithms attempt to extract higher

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

227

and higher levels of information from increasingly complex images. In this respect we are attempting to quantitatively identify the fraction of a material that is present in a pixel, using tens of hundreds of bands sequence data. As a result the algorithm becomes very complex. The latter suggests an alternative process, namely, SIG (synthetic image generation), in which a model is used to generate a high-ﬁdelity representation of what an actual sensor would see by modeling the scene’s physics, including all intervening phenomena associated with the image formation and image capture process. Recently progress has been made toward converting the unmixing technique into a more unsupervised (rather than supervised) classiﬁcation technique. In this manner, endmember selection, based on spectral signatures, is done without having a priori knowledge of the scene’s components. Extracting pure (endmembers) pixels from a given area is therefore a signiﬁcant step in the process. Pure Purity Index (PPI) algorithm and endmember selection using a dimensionality visualization technique are part of the ﬁve-step method, proposed by Kruse and Boardman (1997), which has become a leading method for quantitative analysis of HSR data. Efforts to improve the unmix technique are being made by many workers, such as Frans and Schowengerdt (1999) and Milton (1999), because the technique promises to have the feasibility to account for spatial variations based on spectral response. As already discussed, all of the available methods for quantitative remote-sensing applications rely on spectral information (spectral shape and nature) such as spectral matching, spectral peak extraction, and spectral peak modeling (Mustard and Sunshine, 1999). In this regard, another method, which compares a spectrum of a pixel with a spectrum of a selected endmember (Boardmann, 1993), is also known to be a quantitative tool for classifying HSR data. Several methods for extracting band shape are available (Clark et al., 1990; Kruse et al., 1991; Kimes et al., 1993). Three methods for modeling the absorption band response are also known (Burns, 1970; Farr et al., 1980; Singer, 1981). All three categories require a chromophoric material with signiﬁcant absorption peaks and a signiﬁcant spectral response. In the soil environment, however, these methods tend to be problematic, because soil is characterized as a very spectrally monotonous material. In this regard Condit (1972) has suggested a vector analysis method to classify soils based on their spectral characteristics, but still showed that more study is required in order to fully understand the soil spectral behavior. A comprehensive review of statistical methods for quantitative applications using HSR data has been recently published in a monograph edited by Van der Meer (1999). Also Mustard and Sunshine (1999) have reviewed the spectral analytical methods for remote sensing of the Earth. Apparently the NIRA approach for soil applications, which tends to ﬁnd an empirical relationship between spectral and pure chemical–physical information, has not occupied the place in the hierarchy of methods that may have been expected. This is mainly because of the obstacles reviewed earlier in this chapter, and because further investigation is needed in order

228

E. BEN-DOR

to establish practical steps to be followed in the ﬁeld, which are totally different from those to be followed in the laboratory (Preissler and Loercher, 1995). It is important to note, however, that for vegetation applications, the NIRA remotesensing technique has been used successfully by many workers (e.g., Curran et al., 1992, 1997; Dawson et al., 1999; Martin and Aber, 1997). Nevertheless, although soil surface is more complex than vegetation canopy, laboratory success regarding spectral information (chromophores) and new HSR data may suggest that the NIRA approach is a potential method for remote sensing of soil. In this regard, and very recently, Ben-Dor et al. (2001) and Udelhoven et al. (1998) applied the NIRA approach to real HSR data in order to check its potential. Both studies employed a complete range of NIR steps, which included precise soil sampling, wet chemistry analysis, spectral manipulation, extraction of the empirical relationship between the spectral response, and the chemistry information of selected areas and ﬁeld validation. In both studies it was proven that the NIR strategy is indeed feasible for use in soil surface mapping. Ben-Dor et al. (1999) concluded that although many difﬁculties occurred (most of them discussed in this chapter), the NIRA technique is a promising method for quantitative soil mapping applications. Moreover, they assumed that had better sensors and treatments been used, better results could have been achieved. In summary it could be said that the HSR technique promises to provide unique spectral information that gives a new overview of large areas.

VIII. CLOSING REMARKS AND RECENT EXAMPLES Many studies have been conducted with the intention of classifying soil and soil properties, using optical sensors onboard orbital satellites, such as LANDSAT MSS and TM, SPOT and NOAA-AVHRR (e.g., Agdu et al., 1990; Cipra, 1980; Frazier and Cheng, 1989; Kirein-Young and Kruse, 1989; Morran et al., 1992; Mulders, 1987; Westin and Franzee 1976). Qualitative classiﬁcation approaches have traditionally been used to analyze multichannel data in cases where the spectral information was relatively low. Nevertheless, it has also been possible to obtain useable sets of information about soil type, soil degradation, and soil conditions from “broad” channel sensors by applying sophisticated classiﬁcation approaches (Ben-Dor and Banin, 1995c; Price, 1990). Over the years, soil spectra have been collected and analyzed in the laboratory both quantitatively and qualitatively by many workers (e.g., Latz et al., 1981; Price, 1995). According to Price (1995), a comprehensive literature survey of these data sets is impractical, because most collections are documented only through internal reports and are not easily obtained. Price (1995) reviewed several spectral data sets, and for soils and minerals, those of Condit (1970), Biehl et al. (1984), Stoner et al. (1980a,b), Satterwhite and Henley (1991), and Grove et al. (1992) are the most appropriate. Another

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

229

limitation of soil libraries is that in some cases no chemical–physical information is available with the spectra. Because soil spectra represent rather complicated sets of data, it is critical to have both the soil spectra and the chemical information in a given data set. In general it can be concluded that if all of the obstacles reviewed in this chapter can be overcome, analysis of the spectra can yield useful information about the chemical characteristics of the soil, even though many of the channels are intercorrelated and the soil matrix is complex. Because of the unknown interactions between soil chromophores, it is impossible to predict the most appropriate wavelengths for explaining a given constituent in a given situation (illumination, viewing angle, optimal mathematics manipulation etc.). The complex interactions between components in soils may make the theoretical models impractical, and, hence, empirical models need to be incorporated. It is true that spectral variability can be explained by relatively small and broad spectral bands, but there is no doubt that additional information would make for better, more detailed explanations. The development of a sophisticated analytical method and a synergy between physical and empirical models could be the keys for retrieving quantitative information about soil properties solely from their airborne reﬂectance spectra. This option should be at the center of the agendas of today’s workers, as new spectral imaging systems, with greater near-laboratory spectral capabilities, are entering the ﬁeld of remote sensing. Recent studies have already proven that this statement is correct (Ben-Dor et al., 2001; and Udelhoven et al., 1998), using airborne HSR data. In general, imaging spectrometers should consist of a reasonable number of spectral channels across the entire VNIR–SWIR region that will cover the spectrally active regions of all chromophores with a reasonable bandwidth. To the best of our knowledge, the NIRA strategy has not yet been used with real remote-sensing data for soil property applications. This is in spite of the fact that for vegetation, the methodology has been successful at all levels. In the scientiﬁc literature, however, a considerable amount of effort has been made toward determining important soil properties, such as soil particle size (Zhang et al., 1992), soil organic matter (Henderson et al., 1992; Ishida and Ando, 1999), salt (Dwivedi et al., 1999; Mougenot et al., 1993), iron content (Coleman and Montgomery 1987, moisture (Seyler et al., 1998) and soil erosion (Metternicht and Fermont, 1998). The most sophisticated spectral-based technique that is used for soil classiﬁcation is the spectral unmixing technique that accounts for the linear combination of the chromophores that make up a given ground pixel (e.g., Accioly et al., 1998; Fox et al., 1990; McCubbin et al., 1998). A recent study by Ben-Dor et al. (2000) used the DAIS-7915 sensor data with 72 channels in order to map several soil properties over a problematic area in Israel. This study used a complete NIRA concept along with attempts to solve most of the obstacles discussed in this chapter and showed that the NIRA concept is indeed a promising technique for use outside of the laboratory. In their study, a full range of NIRA steps, including calibration and validation stages, sample selections, wet-chemistry analysis, and

230

E. BEN-DOR

spectral manipulations, were performed. Using this concept they were able to map organic matter, soil moisture, soil salinity, and a property that is correlated with soil texture (soil-saturated moisture). Using the quantitative soil property images, they applied a GIS–NIR combined technique to estimate a spatial distribution of the soil property under heavy canopy coverage. This was done by randomly and automatically selecting points out of the vegetation areas and interpolating all of these points onto an iso-contour soil property map. Figure 24 (see color inserts) provides the resulting NIR image of organic matter and soil salinity (EC) as an example; whereas Fig. 25 (see color inserts) shows the interpolation results of the GIS–NIR analysis for these properties. Recently, using HyMAP scanner data, Taylor and Dehhan (2000) were able to map soil salinity based on distinctive spectral features across the VIS–NIR spectral region based on features related to combined water in hydrated evaporite minerals. Another project that applied a neuron-network algorithm to derive soil properties using HSR data was done by Udelhoven et al. (1998). They were able to predict organic carbon content, using simulated AVIRIS and DAIS-7915 spectra. Applying their algorithm on a pixel-by-pixel basis yielded the spatial variation of organic carbon over large areas. Organic carbon, or matter, tends to be the most chromophoric property for soil mapping purposes because it consists of signiﬁcant spectral features across the entire spectrum (Ben-Dor et al., 1997), even when using a low-resolution spectrum. Therefore it is understandable why many studies, including those that use multispectral sensors, provide good predictions of organic matter on a large spatial scale. Coops et al. (1998) used CASI data for determining many soil properties, using an internal statistical algorithm in order to come up with the best spectral response for each property. Although this approach is not completely an NIR strategy, it employs several similar steps, such as the intercorrelation of spectral and chemical data. In this regard they found poor (but signiﬁcant) correlations between several spectral response channels and soil moisture, total P, exchangeable cation capacity, and Ca Mg ratio. Coleman et al. (1991) studied the determination of several soil properties using eight spectral bands across the VIS–NIR–SWIR–TIR spectral region. In this regard they were able to predict silt, clay, and organic matter content is soils. Using a similar approach, Coleman and Montgomery (1987) were able to predict organic mater, soil moisture, and iron oxides using several multiple regression equations. Cwick et al. (1998) used spectral information to correlate soil K with spectral responses. Quantitative information regarding indirect soil information can be found in Kaufman and Gao (1992) with regard to quantitative water vapor mapping. This was done by using the water vapor peak depths at 0.94 and 1.14 μm and applying a quantitative algorithm on a pixel-by-pixel basis that accounts for the peak depth. Although it is not exactly a soil property, it might provide valuable information for quantifying soil moisture or evaporation rates of soils. In summary it can be said that qualitative remote sensing of soils is a feasible and applicable process, but quantitative remote sensing of soils is a

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

231

more problematic issue. However, taking into consideration the new technology that is under development, and good understanding of all received difﬁculties, it is promising that NIRA or some similar technique will play a signiﬁcant role in future remote sensing activities. This evidence strongly demonstrates that soil NIRA analysis is a unique approach for a precise spatial soil monitoring concept using a cutting-edge remote-sensing technology in the years to come.

ACKNOWLEDGMENT We acknowledge support for this research by the Israel Science Foundation. Part of this material was originally published in a signiﬁcant portion of a chapter in Remote Sensing for the Earth Sciences: Manual of Remote Sensing, 3e, Vol III (Rencz, ed.) entitled “Soil Reﬂectance” by Ben-Dor et al., 1998 copyright. It is reproduced herein with the permission of John Wiley and Sons, Inc.

REFERENCES Aber, J., Wessman, C. A., Peterson, D. L., Mellilo, J. M., and Fownes, J. H. (1990). Remote sensing of litter and soil organic matter decomposition in forest ecosystems. In “Remote Sensing of Biosphere Functioning” (R. J. Hobbs and H. A. Mooney, Eds.), pp. 87–101. springer-Verlag, New York Berlin Heidelberg London Paris Tokyo Hong Kong. Accioly, L., Huete, A. R., and Batchily, K. (1998). Using mixture analysis for soil information extraction from an AVIRIS scene at the Walnut Gulch experimental watershed—Arizona. In “Proceedings of the 7th Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) Workshop.” JPL Pub. 90–54. Pasadena, CA. ACORM, 2001, Quick Start Guide Version 3.12 Analytical Imaging and Geophysics LLC, Boulder CO. 17p. Adams, J. B., Smith, M. O., and Gillespie, A. R. (1993). Imaging spectroscopy: Interpretation based on spectral mixture analysis, in remote geochemical analysis: Elemental and mineralogical composition. In “Topics in Remote Sensing, 4” (C. M. Pieters, and P. A. J. Englert, Eds.), pp. 145–166. Cambridge University Press, Cambridge. Adams, J. B., Smith, M. O., and Johnson, P. E. (1996). Spectral mixture modeling: A new analysis of rocks and soil types at the Viking Lander 1 site. J. Geoph. Res. 91(B8), 8098–8112. Agdu, P. A., Fehrenbacher, J. D., and Jansen, I. (1990). Soil property relationships with SPOT satellite digital data in the East central Illinois. Soil. Sci. Soc. Am. J. 5, 807–812. Al-Abbas, H. H., Swain, H. H., and Baumgardner, M. F. (1972). Relating organic matter and clay content to multispectral radiance of soils. Soils. Sci. 114, 477–485. Alexander, J., Dematte, M., and Garcia, G. J. (1999). Alteration of soil properties through a weathering sequence as evaluated by spectral reﬂectance. Soil Sci. Soc. Am. J. 63, 327–342. Ammer, U., Koch, B., Schneider, T., and Wittmeier, H. (1991). High resolution spectral measurements of vegetation and soil in ﬁeld and laboratory. In “Proceedings of the 5th International Colloquium, Physical Measurements and Signatories in Remote Sensing.” I, pp. 213–218. Courchevel, France. Bach, H., and Mauser, W. (1991). The applications of imaging spectroscopy data in the agriculture and hydrology the EISAC-89 campaign in the Freiburg test site. EARSeL Adv. Rem. Sens. 1, 34– 42. Banin, A., and Amiel, A. (1970). A correlation of the chemical physical properties of a group of natural soils of Israel. Geoderma 3, 185–198.

232

E. BEN-DOR

Bars, R., Okkonen, J., and Bernard, W. (1998). Fast hyperspectral information production by ASIA airborne imaging spectrometer. In “Proceedings of the 1ST EARSeL Workshop on Imaging Spectroscopy.” pp. 173–182. Zurich, 6–8 October 1998. Baumgardner, M. F., Kristof, S. J., Johannsen, C. J., and Zachary, A. L. (1970). Effects of organic matter on multispectral properties of soils. In “Proceedings of the Indian Academy of Science,” Vol. 79, pp. 413–422. Baumgardner, M. F., Silva, L. F., Biehl, L. L., and Stoner, E. R. (1985). Reﬂectance properties of soils. Adv. Agro. 38, 1–44. Beck, R. H., Robinson, B. F., McFee, W. H., and Peterson, J. B. (1976). Information Note 081176. Laboratory Application of Remote Sensing, Purdue University, West Lafayette, IN. Bedidi, A., Cervelle, B., Madeira, J., and Pouget, M. (1990). Moisture effects on spectral characteristics (visible) of lateritic soils. Soil Sci. 153, 129–141. Beirwirth, P. N. (1990). Mineral mapping and vegetation removal via data-calibration pixel unmixing, using mutlispectral images. Int. J. Rem. Sens. 11, 1999–2017. Ben-Dor, E., and Banin, A. (1990a). Diffuse reﬂectance Spectra of smectite minerals in the near infrared and their relation to chemical composition. Sciences Geologiques Bull. 43(2–4),117–128. Ben-Dor, E., and Banin, A. (1990b). Near infrared reﬂectance analysis of carbonate concentration in soils. Appl. Spec. 44(6), 1064–1069. Ben-Dor, E., and Banin, A. (1994). Visible and near infrared (0.4-1.1 μm) analysis of arid and semiarid soils. Rem. Sens. Env. 48, 261–274. Ben-Dor, E., and Banin, A. (1995a). Near infrared analysis (NIRA) as a rapid method to simultaneously evaluate, several soil properties. Soil Sci. Soc. Am. J. 59, 364–372. Ben-Dor, E., and Banin, A. (1995b). Near infrared analysis (NIRA) as a simultaneously method to evaluate spectral featureless constituents in soils. Soil Sci. 159, 259–269. Ben-Dor, E., and Banin, A. (1995c). Quantitative analysis of convolved TM spectra of soils in the visible, near infrared and short-wave infrared spectral regions (0.4–2.5 μm). Int. J. Rem. Sens. 18, 3509–3528. Ben-Dor, E., Banin, A., and Singer, A. (1991). Simultaneous determination of six important soil properties by diffuse reﬂectance in the near infrared region. In “Proceedings of the 5th International Colloquium on Physical Measures and Signatures in Remote Sensing.” Vol. 1, pp. 159–163. Courchevel, France. Ben-Dor, E., Kruse, F. A., Lefkoff, A. B., and Banin, A. (1994). Comparison of three calibration techniques for the utilization of GER 63 channel scanner data of Makhtesh Ramon, Negev, Israel. Photogrammetric Engineering and Remote Sensing, 60, 1339–1354. Ben-Dor, E., Inbar, Y., and Chen, Y. (1997). The reﬂectance spectra of organic matter in the visible near infrared and short wave infrared region (400–2500 nm) during a control decomposition process. Rem. Sens. Env. 61, 1–15. Ben-Dor, E., Irons, J. R., and Epema, G. (1999). Soil reﬂectance studies in manual of remote sensing. In “Remote Sensing for the Earth Sciences” (A.N. Rencz, Ed.), Manual of Remote Sensing, 3rd edition, Vol. 3, pp. 111–189. Wiley, New York. Ben-Dor, E., and Kruse, F. A. (1994). Mineral mapping of Maktesh Ramon Negev, Israel, using GER 63 channel scanner data and linear unmixing procedures. In “Proceedings of the 10th Thematic Conference on Geologic Remote Sensing,” pp. 215–225. San Antonio, TX. Ben-Dor, E., and Kruse, F. A. (1995). Surface mineral mapping of Makhtesh Ramon Negev, Israel using GER 63 channel scanner data. Int. J. Rem. Sens. 18, 3529–3553. Ben-Dor, E., and Levin, N. (2000). Determination of surface reﬂectance from raw hyperspectral data with out symoltenouse ground truth measurmenst. A case study of the GER 63-channle sensor data acquiered over Naan Israel. Int. J. Rem. Sens. 21, 2053–2074. Ben-Dor, E., Patkin, K., and Banin, A. (1999). Detection of several soil properties using the DAIS-7915 hyperspectral scanner data over a saline area. In “Proceedings of the 13th International Conference on Applied Geology Remote Sensing,” Vol. I, pp. 316–323. Vancouver, Canada.

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

233

Ben-Dor, E., Patkin, K., Richter, R., Muller, A., and Koufmann, H. (2000). Mapping of several soil properties using DAIS-7915. In A Decade of Trans-European Remote Sensing Cooperation (ed. M. F. Buchroithner), Balkema Publisher, Lisse, Abingdon, Exton, Tokyo 385–390. Ben-Dor, E., and Singer, A. (1987). Optical density of vertisol clays suspensions in relation to sediment volume and dithionite-citrate-bicarbonate extractable iron. Clays. Clay. Min. 35, 311–317. Ben-Gera, I., and Norris, K. H. (1968). Determination of moisture content in soybeans by direct spectrophotometry. Israeli J. Agr. Res. 18, 124–132. Berk, A., Anderson, G. P., Acharya, P. K., Chetwynd, J. H., Bernstein, L. S., Shettle, E. P., Matthew, M. W., and Adler-Golden, S. M. (1999). MODTRAN-4 User Manual, Air Force Material Laboratory Hanscom, AFB, MA 94 p. Bianchi, R., and Marino, C. M. (1994). CNR LARA Project, Italy: MIVIS/MIDAS environmental airborne hyperspectral Remote Sensing System. In “Proceeding of the First International Air borne Conference and Exhibition,” Vol. 3, pp. 613–619. Strasbourg, France. Biehl, L. L, Daughtry, C. S. T, and Bauer, M. E. (1984). “Vegetation and Soils Field Research Data Base: Experiment Summaries.” LARS Technical Report 042832. Purdue University, West Lafayette IN. Binaghi, E., Brivo, P. A., Ghezzi, P., Rampini, A., and Zilioli, E. (1999). Investigation the behavior of neural and fuzzy-statistical classiﬁcation in sub-pixel land cover estimation. Can. J. Rem. Sens. 25, 171–188. Boardman, J. W. (1993). Automating spectral unmixing of AVIRIS data using convex geometry concept Summaries of the 4th Annual JPL Airborne Geosciences Workshop (ed. Green R. O.) JPL Pub. 93-26, I Pasadena CA, 11–14. Boardman, J. W., and Huntington, F. J. (1997). Mineralogic and geochemical mapping at Virginia City, Nevada using 1995 AVIRIS data. In “The 12th International Conference and Workshops on Applied Geologic Remote Sensing” Vol. I, pp. 191–198. Denver, CO. Boardman, J. W., and Kruse, F. A. (1994). Automated spectral analysis: a geological example using AVIRIS data, north Grapevine Mountains, Nevada. In “Proceedings of the ERIM Tenth Thematic Conference on Geologic Remote Sensing.” Vol. I, pp. 407–418. Bowers, S., and Hanks, R. J. (1965). Reﬂectance of radiant energy from soils. Soil Sci. 100, 130–138. Burns, R. (1970). Mineralogical Applications of Crystal Field Theory, Cambridge University Press, Cambridge, 551 p. Carey, J. (1995). Tilling the soil by satellite, Bus. Week 11, 62–63. Cariati, F., Erre, L., Micera, G., Piu, P., and Gessa, C. (1981). Water molecules and hydroxyl groups in montmorillonites as studied by near infrared spectroscopy. Clays Clay Min. 29, 157– 159. Cariati, F., Erre, L., Micera, G., Piu, P., and Gessa, C. (1983). Polarization of water molecules in phyllosylicates in relation to exchange cations as studied by near infrared spectroscopy. Clays Clay Min. 31, 155–157. Carrere, V. (1991). Use of high spectral resolution airborne visible/infrared imaging spectrometer data for geologic mapping: An overview. In “Proceedings of the 5th International Colloquium- Physical Measurements and Signatures in Remote Sensing,” pp. 137–143. Curchevel, France. Carrere, V., and Conel, J. E. (1993). Recovery of atmospheric water vapor total column abundance from imaging spectrometer data around 940 nm—Sensitivity analysis and application to airborne visible/infrared imaging spectrometer (AVIRIS), data. Rem. Sens. Env. 44, 179–204. Chabrillat, S., Goetz, F. H. A., Olsen, H. W., Krosley, L., and Noe, D. (1997). Use of AVIRIS hyeprspectral data to identify and map expansive clay soils in the front range urban corridor in Colorado. In “Proceedings of the 3rd International Airborne Remote Sensing Conference and Exhibition,” Vol. I, pp. 390–397. Copenhagen. Chen, Y., and Inbar, Y. (1994). Chemical and spectroscopical analysis of organic matter transformation during composting in relation to compost maturity. In “Science and Engineering of Composting: Design, Environmental, Microbiology and Utilization Aspects” (H. A. J. Hoitink and H. M. Keener, Eds.), pp. 551–600. Renaissance Publications, Worthington, OH.

234

E. BEN-DOR

Cierniewski, J. (1987). A model for soil surface roughness inﬂuence on the spectral response of bare soils in the visible and near infrared range. Rem. Sens. Env. 23, 98–115. Cierniewski, J. (1989). The inﬂuence of the viewing geometry of bare soil surfaces on their spectral response in the visible and near infrared. Rem. Sens. Env. 27, 135–142. Cipra, J.E., Franzmeir, D. P., Bauer, M. E., and Boyd, R. K. (1980). Comparison of multispectral measurements from some nonvegetated soils using Landsat digital data and a spectroradiometer. Soil Sci. Soc. Am. J. 44, 80–84. Clark, R. N. (1981). The reﬂectance of water-mineral mixtures at low temperatures. J. Geo. Res. 86, 3074–3086. Clark, R. N. (1997). Spectroscopy. of Rocks and Minerals and Principle of Spectroscopy. In “Remote Sensing for the Earth Sciences” (A. N. Rencz Ed.), Manual of Remote Sensing, 3rd edition, Vol. 1, pp. 3–59. Wiley New York Chichester Weinheim Brisbane Singapore Toronto. Clark, R. N., Gallagher, A. J., and Swayze, G. A. (1990). Material absorption band depth mapping of imaging spectrometer data using a complete band shape least-squares ﬁt with library reference spectra. In “Proceedings of the 2nd Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) Workshop,” pp. 176–186, JPL Pub. 90-54. Pasadena, CA. Clark, R. N., and Roush, T. L. (1984). Reﬂectance spectroscopy: Quantitative analysis techniques for remote sensing applications. J. Geophys. Res. 89, 6329–6340. Cocks, T., Jrnssen, R., Stewart, A., Wilson, I., and Shields, T. (1998). The HyMapTM airborne hyperspectral sensor: The system, calibration and performance. In “Proceedings of the 1st EARSeL Workshop on Imaging Spectroscopy,” pp. 37–42. Zurich, Switzerland. Coleman, T. L., Agbu, P. A., Montgomery, O. L., Gao, T., and Prasad, S. (1991). Spectral band selection for quantifying selected properties in highly weathered soils. Soil Sci. 151, 355–361. Coleman, T. L., and Montgomery, O. L. (1987). Soil moisture, organic matter and Iron content effect on the spectral characteristics of selected vertisols and alﬁsols in Alabama. Photo. Eng. Rem. Sens. 53, 1659–1663. Collins, W., and Chang, S. (1990). The geophysical environmental research corp. 63 channel airborne imaging spectrometer and 12 band thermal scanner. SPIE 1298, 62–71. Condit, H. R. (1970). The spectral reﬂectance of American soils. Photo, Eng, Rem. Sens. 36, 955–966. Condit, H. R. (1972). Application of characteristic vector analysis to the spectral energy distribution of daylight and the spectral reﬂectance of American soils. App. Opt. 11, 74–86. Conel, J. E. (1969). Infrared emissivities of silicates: experimental results and a cloudy atmosphere model of spectral emission from condensed particulate mediums. J. Geo. Res. 74, 1614–1634. Conel, J. E., Green, R. O., Vane, G., Bruegge, C. J., and Alley, R. E. (1987). AIS-2 radiometry and a comparison of methods for the recovery of ground reﬂectance. In “Proceedings of the 3rd, Airborne Imaging Spectrometer Data Analysis Workshop,” pp. 18–47, JPL-Pub. 87-30. Jet Propulsion Laboratory, Pasadena, CA. Coops, N., Ryan, P., and Bishop, A. (1998). Investigating CASI responses to soil properties and disturbance across Australian eucalypt forest. Can. J. Rem. Sens. 24, 153–168. Coulson, K. L. (1966). Effects of reﬂection properties of natural surfaces. J. App. Mete. 10, 1285–1295. Coyne, L. M., Bishop, J. L., Sacttergood, T., Banin, A., Carle, G., and Orenberg, J. (1989). Nearinfrared correlation spectroscopy: Quantifying iron and surface water in series of variably cationexchanged montmorillonite clays. In “Spectroscopic Characterization of Mineral and Their Surfaces” (L. M. Coyne, S. W. S. McKeever and D. F. Blake, Eds.), pp. 407–429. Washington D.C., American Chemical Society. Csillag, F., Pasztor, L., and Biehl, L. L. (1993). Spectral band selection for the characterization of salinity status of soils. Rem. Sens. Env. 43, 231–242. Curran, P. J., Dungam, J. L., Macler, B. A., Plummer, S. E, and Peterson, D. L. (1992). Reﬂectance spectroscopy of fresh whole leaves for the estimation of chemical concentration. Rem. Sens. Env. 39, 153–166.

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

235

Curran, P. J., Kupiec, J. H., and Smith, G. M. (1997). Remote sensing the biochemical composition of a slash pipe canopy. IEEE Trans. Geosci. Rem. Sens. 35, 415–420. Cwick, G. J., Aide, M. T., and Bishop, M. P. (1998). Use of hyperspectal and biochemical data from black spruce needles to map soils at forest site in Manitoba. Can. J. Rem. Sens. 24, 187–192. Da Costa, L. M. (1979). Surface soil color and reﬂectance as related to physicochemical and mineralogical soil properties. Ph.D. dissertation, University of Missouri, Columbia. Dalal, R. C., and Henry, R. J. (1986). Simultaneous determination of moisture, organic carbon and total nitrogen by near infrared reﬂectance spectroscopy. Soil Sci. Soc. Am. J. 50, 120–123. Davies, A., and Grant, A. (1987). Review: Near infrared analysis of food. Int. J. Food. Sci. Technol. 22, 191–207. Dawson, T. P., Curran, P. J., North, P. R. J., and Plummer, S. E. (1999). The propagation of foliar biochemical absorption features in forest canopy reﬂectance: A theoretical analysis. Rem. Sens. Env. 67, 147–159. Demetriades-Shah, T. H., Steven, M. D., and Clark, J. A. (1990). High resolution derivative spectra in remote sensing. Rem. Sens. Env. 33, 55–64. Den Dulk, J. A. (1989). The interpretation of remote sensing, a feasibility study. Ph.D. Dissertation, Wageningen Agricultural University, The Netherlands. Downey, G., and Byrne, P. (1986). Prediction of moisture and bulk density in milled peat by near infrared reﬂectance. J. Food. Agr. Sci. 37, 231–238. Drake, N. A. (1991). Mapping rocks soils vegetation communities and vegetation density with GERIS using linear mixture modeling and post processing techniques. EARSeL Adv. Rem. Sens. 1, 48–57. Dwivedi, R. S., and Sreenivas, K. (1998). Image transforms as a tool for the study of soil salinity and alkalinity dynamics. Int. J. Rem. Sens. 19, 605–619. Dwivedi, R. S., Srenivas, K., and Ramana, K. (1999). Inventory of salt-affected soils and waterlogged areas: a remote sensing approach. Int. J. Rem. Sens. 20, 1589–1599. Egbert, D. D., and Ulaby, T. F. (1972). Effect of angles on reﬂectivity. Photo. Eng. Rem. Sen. 348, 556–564. Epema, G. F. (1992). Spectral reﬂectance in the Tunisian desert. Ph.D. Dissertation, Wageningen Agricultural University, The Netherlands. Escadafal, R., Girard, M., and Courault, D. (1988). La couleur des sols: appreciation, mesure et relation avec les proprieties spectrales. Agronomy 8, 147–154. Escadafal, R., and Huete, A. R. (1991). Inﬂuence of the viewing geometry on the spectral properties (high resolution visible and NIR) of selected soils from Arizona. In “Proceedings of the 5th International Colloquium, Physical Measurements and Signatures in Remote Sensing,” Vol. I, pp. 401– 404. Courchevel, France. Everitt, J. H., Escobar, D. E., Gerbermann, A. H., and Alaniz, M. A. (1988). Detecting saline soils with video imagery. Photo. Eng. Rem. Sens. 54, 1283–1287. Farr, T. G., Bates, B. A., Ralph, R. L., and Adams, J. B. (1980). Effects of overlapping optical absorption bands of pyroxene and glass on the reﬂectance spectra of lunar soils. Proceedings of the 11th Lunar Planetary Science Conference, 719–729. Foody, G. M., and Cox, D. P. (1994). Sub-pixel land cover composition estimation using linear mixture model and fuzzy members function. Int. J. Rem. Sens. 15, 619–631. Fox, L., Fischer, A. F., Gillespie, A. R., and Smith, M. R. (1990). Using spectral unmixing analysis of AVIRIS high dimensional data for distinguishing soil chronosequences. In “Proceedings of the 2nd Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) Workshop,” JPL Pub. 90-54. Pasadena, CA. Franse, E. P., and Schowengerdt, R. A. (1999). Improving spatial-spectral unmixing with the sensor spatial response function. Can. J. Rem. Sens. 25, 131–151. Frazier, B. E., and Cheng, Y. (1989). Remote sensing of soils in the Eastern Palouse region with Landsat Thematic Mapper. Rem. Sens. Env. 28, 317–325.

236

E. BEN-DOR

Gaffey, S. J. (1985). Reﬂectance spectroscopy in the visible and near infrared (0.35–2.55 μm): Applications in carbonate petrology. Geology 13, 270–273. Gaffey, S. J. (1986). Spectral reﬂectance of carbonate minerals in the visible and near infrared (0.35–2.55 μm): Calcite, aragonite and dolomite. Am. Min. 71, 151–162. Gaffey, S. J., and Reed, K. L. (1987). Copper in calcite: Detection by visible and near infra-red reﬂectance. Econ. Geo. 82, 195–200. Gao, B. C. (1993). An operational method for estimating signal to noise ratio from data acquired with imaging spectrometer. Rem. Sens. Env. 43, 23–33. Gao, B-C., Heidebrecht, K. B., and Goetz, A. F. H. (1993). Derivation if scaled surface reﬂectance from AVIRIS data. Rem. Sens. Env. 44, 165–178. Gausman, H. W., Allen, W. A., Cardenas, R., and Bowen, R. L. (1970). Color photos, cotton leaves and soil salinity. Photo. Eng. Rem. Sens. 36, 454–459. Gerbermann, A. H., and Neher, D. D. (1979). Reﬂectance of varying mixtures of a clay soil and sand. Photo. Eng. Rem. Sens. 45, 1145–1151. Gilabert, M. A., Conese, C., and Maselli, F. (1994). An atmospheric correction method for automatic retrieval of surface reﬂectance from TM images. International Journal of Remote Sensing 15, 2065–2086. Goetz, A. F. H. (1985). Imaging spectrometry for Earth remote sensing. Science 228, 1147–1153. Goetz, A. F. H. (1987). HIRIS—High-resolution imaging spectrometer: Science opportunities for the 1900s. Instrument Panel Report for NASA. Earth Observing System IIc. Goetz, A. F. H., and Herring, M. (1989). A high resolution imaging spectrometer (HIRIS) for EOS. IEEE Tran. Geo. Rem. Sens. 27, 136–144. Goetz, A. F. H. (1991). Imaging spectrometry for studying earth, air, ﬁre and water. EARSeL Adv. Rem. Sens. 1, 3–15. Goetz, A. F. H. (1992a). Imaging spectrometry for Earth remote sensing In “Imaging Spectroscopy: Fundamentals and Prospective Applications” (F. Toselli and J. Bodechtel, Eds.), pp. 1–19. ECSE, EEC, EAEC, Brussels/Luxembourg. Goetz, A. F. H. (1992b). Principles of narrow band spectrometry in the visible and IR: Instruments and data analysis. In “Imaging Spectroscopy: Fundamentals and Prospective Applications” (F. Toselli and J. Bodechtel, Eds.), pp. 21–32. ECSE, EEC, EAEC, Brussels/Luxembourg. Goetz, A. F. H., Curtis, B., Kruse, F. A., and Boardman, J. W. (1993). “Imaging Spectrometer Data Analysis Techniques.” Center for the Study of the Earth from Space. CSES/CIRES, University of Colorado, Denver. Goetz, F. A. H., Hauff, P., Shippert, M., and Maecher, A. G. (1991). Rapid detection and identiﬁcation of OH-bearing minerals in the 0.9-1.0 μm region using new portable ﬁeld spectrometer. In “Proceeding of the 8th Thematic Conference on Geologic Remote Sensing,” Vol. I, pp. 1–11. Denver, CO. Goetz, A. F. H., and Kindel, B. (1996). Comparison of unmixing results derived from AVIRIS, high and low resolution, and HYDICE images at Cuprite Nevada. In “Proceedings of the AVIRIS Workshope,” pp. 1–10, JPL Pub. 90-54. Jet Propulsion Laboratory, Pasadena, CA. Goldshlager, N., Ben-Dor, E., Benyamini, Y., Agassi, M., and Blumberg, D. (2001a). The spectral reﬂectance of soil’s structural crust in the SWIR region (1.2–2.5 μm). Terra-Nova (in press). Goldshlager, N., Ben-Dor, E., Benyamini, Y., Agassi, M., and Blumberg, D. (2001b). Soil crusting and inﬁltration processes as monitored by soil reﬂectance spectroscopy in the SWIR region. Remote Sensing of Environment Journal (in press). Gong, P., Pu, R., and Mille, J. R. (1992). Correlation leaf area index of ponderosa pine with hyperspectral CASI data. Can. J. Rem. Sens. 18, 275–282. Green, R. O., and Gao, B. C. (1993). A proposed update to the solar irradiance spectrum used in LOWTRAN and MODTRAN. In “Summaries of the Fourth Annual JPL Airborne Geoscience Workshop,” Vol. I, 81–84, JPL Pub. 93-26, Jet Propulsion Laboratory, Pasadena, CA.

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

237

Green, R., Pavri, B., Faust, J., Chovit, C., and Williams, O. (1999). AVIRIS In-ﬁght radiometric calibration results for 1998 airborne visible/infrared imaging spectrometer (AVIRIS). In “Summaries of the 8th Annual JPL Airborne Geoscience Workshop,” Vol. I, JPL Pub. 93-26. Jet Propulsion Laboratory, Pasadena, CA. Grove, C. I., Hook, S. J., and Paylor, E. D. (1992). “Laboratory Reﬂectance Spectra of 160 Minerals, 0.4 to 2.5 Micrometers,” JPL Pub. 92-2. Pasadena, Jet Propulsion Laboratory. Haalan, N., Stallmann, D., and Cramer, M. (1997). Geomteric processing of high resolution airborne scanner imagery using GPS-INS and ground control points. In “Proceedings of the 3rd International Airborne Remote Sensing Conference and Exhibition,” Vol. I, pp. 327–378. Copenhagen, Denmark. Hapke, B. W. (1981). Bidirectional reﬂectance spectroscopy I. Theory. J. Geophys. Res. 86, 3039–3054. Hapke, B. W. (1993). “Theory of Reﬂectance and Emittance Spectroscopy.” Cambridge University Press, New York. Hapke, B. (1996). A model of radiative and conductive energy transfer in planetary regoliths. J. Geophys. Res. 101, 16817–16831. Hardisky, M. A., Klemas, V., and Smart, R. M. (1983). The inﬂuence of soil salinity, growth form and leaf moisture on the spectral radiance of Spartina Alterniﬂora canopies. Photo. Eng. Rem. Sens. 49, 77–83. Henderson, T. L., Baumgardner, M. F., Franzmeierm, D. P., Stott, D. E., and Coster, D. C. (1992). High dimensional reﬂectance analysis of soil organic matter. Soil Sci. Soc. Am. J. 56, 865– 872. Henderson, B. G., and Jakosky, B. M. (1994). Near-surface thermal gradients and their effects on mid-infrared emission spectra of planetary surfaces. J. Geoph. Res. 99/E9, 19063–19073. Hick, R. T., and Russell, W. G. R. (1990). Some spectral considerations for remote sensing of soil salinity. Aust. J. Soil. Res. 28, 417–431. Hirschfeld, T. (1985). Salinity determination using NIRA. Appl. Spec. 39, 740–741. Huete, A. R. (1988). Soil Adjusted Vegetation Index (SAVI). Rem. Sens. Environ. 25, 47–57. Honigs, D. E., Hieftje, G. M., and Hirschfeld, T. (1984). A new method for obtaining individual component spectra from those of complex mixtures. Appl. Spec. 38, 317–322. Horowitz, H. M., Lewis, J. T., and Pentland, P. (1975). Estimating proportion of Objects from Multispectral Scanner Data, Final Report NASA-CR-141862, U.S. Governmental Printing Ofﬁce, Washington D.C, pp. 108. Huete, A. R., and Escadafal, R. (1986). Separation of soil-plant spectral mixtures by factor analysis. Rem. Sens. Env. 19, 237–251. Huete, A. R., and Escadafal, R. (1991). Assessment of biophysical soil properties through spectral decomposition techniques. Rem. Sens. Env. 35, 149–159. Hunt, G. R. (1980). Spectoscopic Properties of Rock and Minerals In: “Handbook of Physical Properties Rocks”. (C. R. Stewart, Ed.), pp. 295. CRC Press, Boca Raton, FL. Hunt, G. R., and Salisbury, J. W. (1970). Visible and near infrared spectra of minerals and rocks: I: Silicate minerals. Mod. Geol. 1, 283–300. Hunt, G. R., Salisbury, J. W., and Lenhoff, A. (1971a). Visible and near-infrared spectra of minerals and rocks: III Oxides and hydroxides. Mod. Geol. 2, 195–205. Hunt, G. R., Salisbury, J. W., and Lenhoff, C. J. (1971b). Visible and near-infrared spectra of minerals and rocks: Sulﬁdes and sulfates. Mod. Geol. 3, 1–14. Hunt, G. R., Salisbury, J. W., and Lenhoff, C. J. (1971c). Visible and near-infrared spectra of minerals and rocks: Halides, phosphates, arsenates, vandates and borates. Mod. Geol. 3, 121–132. Hunt, G. R., and Salisbury, J. W. (1971). Visible and near infrared spectra of minerals and rocks: Carbonates. Mod. Geol. 2, 23–30. Hunt, G. R., and Salisbury, J. W. (1976). Visible and near infrared spectra of minerals and rocks: XI Sedimentary rocks. Mod. Geol. 5, 211–217.

238

E. BEN-DOR

Ichoku, C., and Karnieli, A. (1996). A review of mixture modeling techniques for sub-pixel land cover estimation. Rem. Sens. Rev. 13, 161–186. Idso, S. B., Jackson, R. D., Reginato, R. J., Kimball, B. A., and Nakama, F. S. (1975). The dependence of bare soil albedo on soil water content. J. Appl. Meteor. 14, 109–113. Irons, J. R., Weismiller, R. A., and Petersen, G. W. (1989). Soil reﬂectance In “Theory and Application of Optical Remote Sensing” (G. Asrar, Ed.), pp. 66–106. Wiley Series in Remote Sensing, Wiley, New York. Ishida, T., and Ando, H. (1999). Use of disjunctive cokriging to estimate soil organic matter from Lndast Thematic image. Int. J. Rem. Sen. 20, 1549–1565. Ishida, T., Ando, H., and Fukuhara, M. (1991). Estimation of complex refractive index of soil particles and its dependence on soil chemical properties. Rem. Sens. Env. 38, 173–182. Jackson, R. D., Teillet, P. M., Slater, P. N., Fedosjsvs, G., Jasinski, M. F., Aase, J. K., and Moran, M. S. (1990). Bidirectional measurements of surface reﬂectance for view angle corrections of oblique imagery. Rem. Sens. Env. 32, 189–202. Joseph, G. (1996). Imaging sensors for remote sensing. Rem. Sen. Rev. 13, 257–342. Karmanova, L. A. (1981). Effect of various iron compounds on the spectral reﬂectance and color of soils. Sov. Soil. Sci. 13, 63–60. Karnieli, A., Kidron, G., Ghassler, C., and Ben-Dor, E. (1998). Spectral characteristics of Cynobacteria soil crust in semiarid environment Rem. Sens. Env. 69, 67–75. Karnieli, A., and Tsoar, H. (1994). Spectral reﬂectance of biogenic crust developed on desert dune sand along the Israel-Egypt border. Int. J. Rem. Sens. 16, 369–374. Kaufman, Y. J., and Gao, B. C. (1992). Remote sensing of water vapour in the Near IR from EOS/MODIS IEEE Tran. Geo. Sci. Rem. Sens. 30, 871–884. Kaufmann, H., Weisbrich, W., Beyth, M., Bartov, Y., Itamar, A., Mazor, E., Ronen, S., and Kafri, U. (1991). Mineral identiﬁcation using GER-II data acquired from Makhtesh Ramon/Negev, Israel. In “EARSeL-Advances in Remote Sensing—Imaging Spectroscopy,” 1/1, pp. 82–92. BoulogneBillancourt, France. Kierein-Young, K., and Kruse, F. A. (1989). Comparison of Landsat Thematic Mapper images and geophysical and environmental reassert imaging spectrometer data for alteration mapping. In “Proceedings of the 7th Thematic Conference on Remote Sensing for Exploration Geology,” Vol. I, pp. 349–359. Calgary, Alberta, Canada. Kimes, D. S. (1983). Dynamics of directional reﬂectance factor distributions vegetation canopies. Appl. Opt. 22, 1364–1372. Kimes, D. S., Irons, J. R., Levine, E. R., and Horning, N. A. (1993). Learning class discriminations from a data base of spectral reﬂectance of soil samples. Rem. Sens. Env. 43, 161–169. Kimes, D. S., Newcomb, W. W., Tucker, C. J., Zonneveld, I. S., Van Wijngaarden, J., de Leeuw, J., and Epema, G. F. (1985). Directional reﬂectance factor distributions for cover types of northern Africa. Rem. Sens. Env. 18, 1–19. Kneizys, F. X., Abdersen, G. P., Shettle, E. P., Gallery, W. O., Abreu, L. W., Selby, J. E. A., Chetwynd, J. H., and Clough, S. A. (1988). “Users Guide to LOWTRAN-7. Air Force Geophysics Laboratory,” Hanscom AFB, AFGL-TR-88–0177, MA. Kosmas, C. S., Curi, N., Bryant, R. B., and Franzmeier, D. P. (1984). Characterization of iron oxide minerals by second derivative visible spectroscopy. Soil. Sci. Soc. Am. J. 48, 401–405. Krishnan, P., Alexander, J. D., Bulter, B. J., and Hummel, J. W. (1980). Reﬂectance technique for predicting soil organic matter. Soil. Sci. Soc. Am. J. 44, 1282–1285. Kristof, S. F., Baumgardner, M. F., and Johannsen, C. J. (1971). Spectral mapping of soil organic matter. Journal Paper No. 5390, Agricultural Experiment Station, Purdue University, West Lafayette, IN. Kruse, F. A. (1998). Visible and infrared: Sensors and case studies. In “Remote Sensing for the Earth Sciences Manual of Remote Sensing” (A. N. Rencz, Ed.), 3rd edition Vol. 3, 11, pp. 613–643. Wiley, New York/Chichester/Weinhein/Brisbane/Toronto.

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

239

Kruse, F. A., and Boardman, J. W. (1997). Characterization and mapping of Kimberlites and related diatremes in Utha, Colorado and Wyoming, USA, using the airborne visible/infrared imaging spectrometer (AVIRIS). In “Proceedings of the 12th International Conference and Workshops on Applied Geologic Remote Sensing,” Vol. I, pp. 21–28. Denver, CO. Kruse, F. A., Kierein-Young, K. S., and Boardman, J. W. (1990). Mineral mapping of Cuprite, Nevada with a 63-channel imaging spectrometer. Photo. Eng. Rem. Sens. 56, 83–92. Kruse, F. A., Thiry, M., and Hauff, P. L. (1991). Spectral identiﬁcation (1.2–2.5 μm) and characterization of Paris basin kaolinite/smectite clays using a ﬁeld spectrometer In “Proceedings of the 5th International Colloquium, Physical Measurements and Signatures in Remote Sensing,” Vol. I, pp. 181–184. Courchevel, France. Kruse, F. A. (1999). Chapter 11: Visible-Infrared Sensors and Case Studies In: “Remote Sensing for the Earth Sciences” (Rencz A. N., ed.), Manual of Remote Sensing 3rd Edition, pp. 567–612. John Wiley & Sons. LaCapra, V. C., Melack, J. M., Gastil, M., and Valeriano, D. (1996). Remote sensing of foliar chemistry of inundated rice with imaging spectrometry. Rem. Sens. Env. 55, 50–58. Latz, K., Weismiller, R. A., and Van Scoyoc, G. E. (1981). “A Study of the Spectral Reﬂectance of Selected Eroded Soils of Indiana in Relationship to Their Chemical and Physical Properties.” LARS Technical Report 082181. Purdue University, West Lafayette, Ind. Leger, R. G., Millette, G. J. F., and Chomchan, S. (1979). The effects of organic matter, iron oxides and moisture on the color of two agricultural soils of Quebec. Can. J. Soil Sci. 59, 191–202. Liang, S., and Townshend, R. G. (1996). A modiﬁed Hapke model for soil bi-directional reﬂectance. Rem. Sens. Env. 55, 1–10. Lillesand, T. M., and Kiefer, R. W. (1994). Remote Sensing and Image Interpretation. John Wiley & Sons, Inc., New York 750 p. Liu, H. Q., and Huete, A. (1995). A feedback based modiﬁcation of the NDVI to minimize canopy background and atmospheric noise. IEEE Trans. Geosci. Rem. Sens. 33, 457–465. Malley, D. F., Yesmin, L., Wray, D., and Edwards, S. (1999). Application of near-infrared spectroscopy in analysis of soil mineral nutrients. Comm. Soil. Sci. Plant Anal. 30, 999–1012. MapInfo User’s Guide (1996). Vertical mapper contour modeling and display software. Desktop mapping software. Martin, M. E., and Aber, J. D. (1997). High spectral resolution remote sensing of forest canopy lignin, nitrogen and ecosystem process. Ecol. Appl. 7, 431–443. Mathews, H. L., Cunningham, R. L., and Peterson, G. W. (1973). Spectral reﬂectance of selected Pennsylvania soils. Soil. Sci. Soc. Am. J. 37, 421–424. McCubbin, I, Green, R., Lang, H., and Roberts, D. (1998). Mineral mapping using partial unmixing at Ray mine AZ. In “Proceedings of the 7th Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) Workshop,” JPL Pub. 90-54. Pasadena, CA. Merickle, M. B., Lundgren, J. C., and Shen, S. S. (1984). A spatial processing algorithm to reduce the effect of mixed pixels and increase the reparability between classes. Pattern Recog. 17, 525–533. Metternicht, G. I., and Fermont, A. (1998). Estimating erosion surface features by linear mixture modeling. Rem. Sens. Env. 64, 254–265. Metternicht, G. I., and Zinck, J. A. (1996). Modeling salinity-alkalinity classes for mapping salt-affected topsoils in the semiarid valleys of Cochabamaba (Bolivia) ITC J. 44, 1282–1285. Miltom, E. J. (1999). Image endmembers and the scene model. Can. J. Rem. Sens. 25, 112–120. Montgomery, O. L. (1976). An investigation of the relationship between spectral reﬂectance and the chemical, physical and genetic characteristics of soils. Ph.D. Thesis, Purdue University, West Lafayette, IN (Libr. Congr. no 79-32236). Montgomery, O. L., and Baumgardner, M. F. (1974). “The Effects of the Physical and Chemical Properties of Soil and the Spectral Reﬂectance of Soils.” Information Note 1125 Laboratory for Applications of Remote Sensing, Purdue University, West Lafayette, IN.

240

E. BEN-DOR

Morin, Y., Benyamini, Y., and Michaeli, A. (1981). The dynamics of soil crusting by rainfall impact and the water movement in the soil proﬁle. J. Hydrol. 52, 321–335. Morra, M. J., Hall, M. H., and Freeborn, L. L. (1991). Carbon and nitrogen analysis of soil fractions using near-infrared reﬂectance spectroscopy. Soil. Sci. Soc. Am. J. 55, 288–291. Morran, S. M., Jackson, R. D., Slater, P. N., and Teillet, P. M. (1992). Evaluation of simpliﬁed procedures for retrieval of land surface reﬂectance factors from satellite sensor output. Rem. Sens. Env. 41, 169–184. Mougenot, B. (1993). Effect of salts on reﬂectance on reﬂectance and remote sensing of salt affected soils. Cahiers Orstom Serie Pedologie XXVIII 1; 45–54. Mougenot, B., Epema, G. F., and Pouget, M. (1993). Remote sensing of salt-affected soils. Rem. Sens. Rev. 7, 241–259. Mulders, M. A. (1987). “Remote Sensing in Soil Science. Developments in soil science,” Vol. 15. Elsevier, Amsterdam. Mulders, M. A., Den Dulk, J. A., and Uijlenhoet, R. (1992). Description of land surfaces, reﬂectance measurements and modeling for correlation with remote sensing data. Sci. du sol 30(3), 169–184. Murphy, R. J., and Wadge, G. (1994). The effects of vegetation on the ability to map soils using imaging spectrometer data. Int. J. Rem. Sens. 15, 63–86. Mustard, John, F., and Sunshine Jessica, M. (1999). Chapter 5: Spectral analysis for earth science investigation. In: “Remote Sensing for the Earth Sciences” (Rencz A. N., ed.), Manual of Remote Sensing 3rd Edition, pp. 251–306, John Wiley & Sons. ˚ Solar Phys. 90, Neckel, H, and Labs, D. (1984). The solar radiation between 3300 and 12500 A. 205–258. Norris, K. H. (1988). History, present state, and future prospects for near infrared spectroscopy. In “Analytical Application of Spectroscopy” (C. S. Creaser and A. M. C Davise,) Eds., pp. 3–8. Royal Society of Chemistry, London. Obukhov, A. I., and Orlov, D. C. (1964). Spectral reﬂectance of the major soil groups and the possibility of using diffuse reﬂection in soil investigations. Sov. Soil Sci. 2, 174–184. Okkonen, J., Hyvarinen, T., and Herrala, E. (1997). AISA airborne imaging spectrometer— on its way from hyperspectral research to operative use. In “Proceedings of the 3rd International Airborne Remote Sensing Conference and Exhibition,” Vol. I, pp. 189–1967, Copenhagen. O’Neill, A. L. (1994). Reﬂectance spectra of microphytic soil crusts in semi-arid Australia. Int. J. Rem. Sens. 15, 675–681. Orlov, D. C. (1966). Quantitative patterns of light reﬂectance on soils. I. Inﬂuence of particles (aggregate) size on reﬂectivity. Sov. Soil Sci. 13, 1495–1498. Orna, M. V. (1978). The chemical origins of color. J. Chem. Ed. 55, 478–484. Otterman, J., Brakke, T., and Marshak, A. (1995). Scattering by Lamberian-leaves canopy: Dependents of leaf-area projections. Int. J. Rem. Sens. 16, 1107–1125. Palacios-Orueta, A., and Ustin, S. L. (1998). Remote sensing of soil properties in the Santa Monica mountains. I: Spectral analysis. Remote Sensing of Environment 65, 170–183. Peterson, J. B. (1979). Use spectral data to estimate the relationship between soil moisture tension and their corresponding reﬂectance. Annual Report OWRT Purdue University. Pieters, C. M. (1983). Strength of mineral absorption features in the transmitted component of nearinfrared reﬂected light. First results from RELAB. J. Geophys. Res. 88, 9534–9544. Pinker, R. T., and Karnieli, A. (1995). Characteristics spectral reﬂectance of semi-arid environment. Int. J. Soil Sci. 16, 1341–1363. Pinty, B., Verstraete, M. M., and Dickson, R. E. (1989). A physical model for prediction bi-directional reﬂectance over bare soil. Rem. Sens. Env. 27, 273–288. Preissler, H., and Loercher, G. (1995). Extraction of soil properties from laboratory and imaging spectrometer data. IEEE 1968–1970. Price, J. C. (1990). On the information content of soil reﬂectance spectra. Rem. Sens. Env. 33, 113–121. Price, J. C. (1991). On the value of high spectral resolution measurements in the visible and

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

241

near-infrared. In “Proceedings of the 5th International Colloquium, Physical Measurements and Signatures in Remote Sensing,” Vol. I, pp. 131–136. Courchevel, France. Price, J. C. (1995). Examples of high resolution visible to near-infrared reﬂectance spectra and a standardized collection for remote sensing studies. Int. J. Rem. Sens. 16, 993–1000. Price, J. C. (1998). An approach for analysis of reﬂectance spectra Rem. Sens. Env. 64, 316–330. Rao, B. R. M., Ravi Sankar, T., Dwivedi, R. S., Thammappa, S. S., and Venkataratnam, L. (1995). Spectral behavior of salt-affected soils. Int. J. Rem. Sens. 16, 2125–2136. Rast, M. (1991). “Imaging Spectroscopy and Its Application in Spaceborne Systems.” ESA publication Division esa SP-1144. Ray, T. W., and Murray, B. C. (1996). Nonlinear spectral mixing in desert vegetation. Rem. Sens. Env. 55, 59–79. Richardson, A. J., Wiegand, C. L., Gausman, H. W., Cullar, J. A., and Gerbermann, A. H. (1975). Plant, soil and shadow reﬂectance components of raw crops. Photo. Eng. Rem. Sens. 41, 1401–1407. Richter, R. (1996). A spatially adaptive fast atmospheric correction algorithm. Int. J. Rem. Sens. 17, 1201–1214. Richter, R. (1997). Correction of atmospheric and topographic effects for high spatial resolution satellite imagery. Int. J. Rem. Sens. 18, 1099–1111. Roberts, D. A., Smith, M. O., and Adams, J. B. (1993). Green vegetation, nonphotosynthetic vegetation, and soils in AVIRIS data. Rem. Sens. Env. 44, 255–269. Rogers, R. W., and Langer, R. T. (1972). Soil surface lichens in arid and subarid south-eastern Australia. Introduction and ﬂoristics. Austr. J. Botany 20, 197–213. Rondeaux, G., Steven, M., and Baret, F. (1996). Optimization of soil-adjusted vegetation indices. Rem. Sens. Env. 55, 95–107. Ronov, A. A., and Yaroshevsky, A. A. (1971). Chemical composition of the earth’s crust In “The Earth’s Crust and Upper Mantle” (P. J. Hart Ed.), pp. 37–57. American Geophysical Union, Washington DC. Satterwhite, M. B., and Henley, J. P. (1991). “Hyperspectral Signatures (400–2,500 nm) of Vegetation, Minerals, Soils Rocks and Cultural Features: Laboratory and Field Measurements.” U.S. Army Corps of Engineers, ELT-0573, Fort Belvoir, VA. Schot, J. R., Brown, S. D., Raqueno, R. V., Gross, H. N., and Robison, G. (1999). An advance synthetic inage generation model and its application to multi/hyperspectral algorithm development. Can. J. Rem. Sens. 25, 99–111. Schwertmann, U. (1988). Occurrence and formation of iron oxides in various pedoenvironment. In “Iron in Soils and Clay Minerals” (J. W. Stucki, B. A. Goodman, and U. Schwertmann, Eds.), pp. 267–308. NATO ASI Series, Reidel Publishing Company, Dordrecht/Boston/Lancaster/Tokyo. Seyler, F., Bernoux, M., and Cerri, C. C. (1998). Landsat TM image texture and moisture variations of the soil surface under the rainforest of the Rondonia state, Brazil. Int. J. Rem. Sens. 19, 1299–1317. Shainberg, I. (1992). Chemical and mineralogical components of crusting In “Soil Crusting ” M. E. Sumenr and B. A. Stewart, Eds.), Lewis Publications, Ann Arbor, MI. Sharma, R. C., and Bhargava, G. P. (1988). Landsat imagery for mapping saline soils and wetlands in north-west India. Int. J. Rem. Sens. 9, 39–44. Shields, J. A., Paui, E. A., Arnaud, R. J., and Head, W. K. (1968). Spectrophtometric measurement of soil color and its relation to moisture and organic matter. Can. J. Soil. Sci. 48, 271–280. Siegal, B. S., and Goetz, A. F. H. (1977). Effect of vegetation on rock and soil type discrimination Photo. Eng. Rem. Sens. 43, 191–196. Singer, R. B. (1982). Near Infrared spectral reﬂectance of mineral mixtures: systematic combinations of pyroxenes, olivine and iron oxides. Journal of Geophysical Research 86, 7967–7982. Singer, A., and Ben-Dor, E. (1987). Origin of red clays layer interbedded with basalt of Golan Heights. Geoderma 39, 283–306. Singh, A. N. (1994). Monitoring change in the extent of salt-affected soils in northern India. International Journal of Remote Sensing 16, 3173–3182.

242

E. BEN-DOR

Smith, M. O., Ustin, S. L., Adams, J. B., and Gillespie, A. R. (1990). Vegetation in desert: I. A regional measure of abundances from multispectral images. Rem. Sens. Env. 31, 1–26. Soileau, J. M., and Craken, R. J. (1967). Free iron and coloration in certain well-drained Costal Plain soils in relation to their other properties and classiﬁcation. Soil. Sci. Soc. Am. J. 31, 248– 255. Staenz, K., Neville, R. A., Levesque, J., Szeredi, T., Singhroy, V., Borstad, G. A., and Hauff, P. (1999). Evaluation of CASI and SFSI hyperspectral data for environmental and geological applicationsTwo case studies. Can. J. Rem. Sen. 25, 311–322. Stark, E., Luchter, K., and Margoshes, M. (1986). Near-infrared analysis (NIRA): A technology for quantitative and qualitative analysis. Appl. Spec. Rev. 24, 335–339. Stenberg, B., Nordkvist, E., and Salomonsson, L. (1995). Use of near infrared reﬂectance spectra of soils for objective selection of samples Soil. Sci. 159, 109–114. Stoner, E. R. (1979). Physicochemical, site and bi-directional reﬂectance factor characteristics of uniformly moist soils. Ph.D. Thesis, Purdue University. Stoner, E. R., and Baumgardner, M. F. (1981). Characteristic variations in reﬂectance of surface soils. Soil. Sci. Soc. Am. J. 45, 1161–1165. Stoner, E. R., Baumgardner, M. F., Biehl, L. L., and Robinson, B. F. (1980). Atlas of Soil Reﬂectance Properties. “Research Bulletin 962. Agricultural Experiment Station, Indian Research.” Purdue University, West Lafayette, IN. Stoner, E. R., Baumgardner, M. F., Weismiller, R. A., Biehl, L. L., and Robinson, B. F. (1980). Extension of laboratory measured soil spectra to ﬁeld conditions. Soil. Sci. Am. J. 44, 572–574. Strobl, P., Richter, R., Muller, A., Lehmann, F., Ortel, D., Tischler, S., and Nielsen, A. (1996). DAIS system performance ﬁrst results from the 1995 evaluation campaigns, Proceedings of the Second International Airborne Remote Sensing Conference and Exhibition, 11, 325–334. San Francisco, Ca. Szilagyi, A., and Baumgardner, M. F. (1991). Salinity and spectral reﬂectance of soils. In “Proceedings of ASPRS Annual Convention, Baltimore,” pp. 430–438. Tanre, D., Deroo, C., Duhaut, P., Herman, M., Morcrette, J. J., Perbos, J., and Deschamps, P. Y. (1986). “Simulation of the Satellite Signal in the Solar Spectrum (5S), User Guide.” UST de Lille, 59655 Villenueve D’asc, France, Laboratory d’Optique Atmospherique. Taranik, D. L., and Kruse, F. A. (1989). Iron minerals reﬂectance in geophysical and environmental research imaging spectrometer (GERIS) data. In “Proceedings of the 7th Thematic Conference on Remote Sensing for Exploration Geology,” Vol. I, pp. 445–458. Calgary, Alberta, Canada. Taylor, G. R., and Dehaan, R. L. (2000). Mapping soil salinity with hyperspectral imagery. In “Proceedings of the 14th International Conference on Applied Geologic Remote Sensing,” pp. 512–520. Las Vegas, Nevada. Todd, A. W., and Hoffer, R. M. (1998). Responses of spectral indices to variations in vegetation cover and soil background. Photo. Eng. Rem. Sens. 64, 915–921. Toth, T., Csillag, F., Biehl, L. L., and Micheli, E. (1991). Characterization of semi-vegetated saltaffected soil by means of ﬁeld remote sensing. Rem. Sens. Env. 37, 167–180. Tucker, C. J., and Miller, L. D. (1977). Soil spectra contributions to grass canopy spectral reﬂectance. Photo. Eng. Rem. Sens. 43, 721–726. Tueller, P. T. (1987). Remote sensing science application in arid environment. Rem. Sens. Env. 23, 143–154. Twomey, S. A., Bohren, C. F., and Mergenthaler, J. L. (1986). Reﬂectance and albedo differences between wet and dry surfaces. Appl. Opt. 25, 431–437. Udelhoven, T., Hill, J., Schutt, B., Imeson, A., and Cammerat, L. H. (1998). A neural network approach for the identiﬁcation of the organic carbon of soils in a degraded semiarid ecosystem (Guadalentin, Spain) based on hyperspectral data from the DAIS-7915 sensor. In “Proceedings of the 1st EARSeL Workshop on Imaging Spectroscopy,” pp. 437–444, Zurich, Switzerland.

QUANTITATIVE REMOTE SENSING OF SOIL PROPERTIES

243

Valeriano, M. M., Epiphano, J. C. N., Formaggio, A. R., and Oliveira, J. B. (1995). Bi-directional reﬂectance factor of 14 soil classes from Brazil. Int. J. Rem. Sens. 16, 113–128. Van der Meer, F. (1994). Extraction of mineral absorption features from high-spectral resolution data using non-parametric geostatistical techniques. Int. J. Rem. Sens. 15/11, 2193–2214. Van der Meer, F. (1995). Spectral reﬂectance of carbonate mineral mixture and bidirectional reﬂectance theory: Quantitative analysis techniques for application in remote sensing. Rem. Sens. Rev. 13, 67–94. Van der Meer, F. (1999). “Remote Sensing and Digital Image Processing,” Kluwer Press, New York. Vane, G. (1987). “Airborne Visible/Infrared Imaging Spectrometer (AVIRIS),” JPL publication, pp. 87–138. Van, G., and Goetz, F. H. (1998). Terrestrial imaging spectroscopy, Remote Sensing of Environment 24, 1–29. Vane, G., Reimer, J. H., Chrien, T. G., Enmark, H. T., Hansen, E. G., and Porter, W. M. (1993). Airborne visible/infrared imaging spectrometer (AVIRIS). Rem. Sens. Env. 44, 127–143. Verma, K. S., Saeena, R. K., Barthwal, A. K., and Deshmukh, S. N. (1994). Remote sensing technique for mapping salt affected soils. Int. J. Rem. Sens. 15, 1901–1914. Vinogradov, B. V. (1981). Remote sensing of the humus content of soils. Sov. Soil. Sci. 11, 114–123. Wang, F. (1990). Fuzzy supervised classiﬁcation of remote sensing images. IEEE Trans. Geoph. Rem. Sens. 28, 194–200. West, N. E. (1990). Structure and function of microphytic soil crust in wildland ecosystems of arid to semi-arid regions. Advances in Ecological Research 20, 179–223. Westin, F. C., and Franzee, C. J. (1976). Landsat data in a soil survey program. Soil. Sci. Soc. Am. J. 40, 81–89. Zhang, R., Warrick, A. W., and Myers, D. E. (1992). Improvement of the prediction of soil particle size fraction using spectral properties Geoderma 52, 223–234. Wessman, C. A., Aber, J. D., Peterson, D. L., and Meliloo, J. M. (1988). Foliar analysis using near infrared spectroscopy. Can. J. Forest. Res. 18, 6–11. Wessman, C. A., Aber, J. D., and Peterson, D. L. (1989). An evaluation of imaging spectrometry for estimating forest canopy chemistry. Int. J. Rem. Sens. 10, 1293–1316. Wiegand, C. L., Rhoades, J. D., Escobar, D. E., and Everitt, J. H. (1994). Photographic and videographic observations for determining and mapping the response of cotton to soil salinity. Rem. Sens. Env. 49, 212–223. Williams, P. C., and Norris, K. H. (1987). Near-Infrared Technology. In “The Agricultural and Food Industries.” American Association of Cereal Chemists Inc., St. Paul, MN.

This Page Intentionally Left Blank

Index A ABA, see Abscisic acid Abscisic acid, water-saving in China, 157 Adsorbed water, soil–radiation interactions, 197–198 Aeollanthus biformifolius, copper and cobalt hyperaccumulation, 8 Alyssum bertolonii, nickel hyperaccumulation, 13 Alyssum lesbiacum, metal tolerance, 20 Alyssum murale, rhizosphere pH, 24 Amaranthus cruentus, radionuclide phytoextraction, 35 Amaranthus retroflexus, radionuclide phytoextraction, 34 Arabidopsis halleri metal hyperaccumulation, 14, 17 metal tolerance, 20–22 root to shoot metal translocation, 17 zinc and cadmium hyperaccumulation, 8 Arabidopsis thaliana mercury phytovolatilization, 42–43 metal tolerance, 22 Arrhenatherum elatius, heavy metal phytoextraction, 32 Arsenic hyperaccumulators, soil phytoextraction, 9–10 Astragalus, selenium and arsenic hyperaccumulation, 9 Atmospheric attenuation, remote sensing of soil, 207–210 B Berkheya coddii nickel hyperaccumulation, 13 root to shoot metal translocation, 17 Beta vulgaris, radionuclide phytoextraction, 35 Bicarboxylic acids, humic substance effects, 95 Biogenic crust, remote sensing, 220–221 Biological water-saving technology, China, 156–158

Biscutella laevigata, metal hyperaccumulation, 10, 14 Brassica chinensis, radionuclide phytoextraction, 35 Brassica juncea chemically induced phytomining, 36 heavy metal phytoextraction, 32–33 lead hyperaccumulation, 9 lead phytoextraction, 31–32 radionuclide phytoextraction, 34–35 selenium phytovolatilization, 40 Brassica napus, root exudates, 24–25 C Cadmium hyperaccumulators mechanism, 14–16 phytoextraction, 8, 10–13 Carbonates, soil–radiation interactions, 190–192 Carbon dioxide, soil organic matter, 121 Cardaminopsis halleri, see Arabidopsis halleri Chelating agents, chemically induced phytoextractions, 38–39 Chemical chromophores, remote sensing of soil, 185–187 China agricultural production, 136 biological water-saving technology, 156–158 ﬁve-water interaction systems, 146–147 groundwater use, 147–149 irrigation, 137–138 low-quality water use, 149–150 population, 136 precipitation, 136–137 rainwater utilization, 144–146 soil fertility, 162–163 soil moisture conservation, 161–162 SPAC interface processes, 146–147 surface water use, 147–149 water conveyance structures, 151–152 water-matched agricultural structure, 154–155 water-matched planning, 155–156 water resources management, 138

245

246

INDEX

China (continued) water-saving irrigation techniques, 152–154, 158–161 water-saving management administrative systems, 163–164 expert systems, 166 laws and regulations, 165 prices and fees, 164–165 products and marketing, 166–167 technical training and services, 165–166 Chromophores remote sensing of soil, 184–187 soil–radiation interactions, 204–205 Clay minerals remote sensing of soil, 185 soil–radiation interactions, 188–190 Cobalt hyperaccumulators, soil phytoextraction, 8–9 Columns, self-assembling humic molecule chromatography, 78–81 Compartmentation, metal tolerance in hyperaccumulators, 18–21 Complexation, metal tolerance in hyperaccumulators, 21–22 Conformational structure, humic substances, 63–66 Contaminated soils chemically induced phytoextraction applications, 27–28 chelating agent concerns, 38–39 heavy metals, 32–33 lead, 28–32 natural hyperaccumulation comparison, 37–38 overview, 26–27 radionuclides, 33–35 chemically induced phytomining, 35–36 clean-up needs, 3–4 hyperaccumulator plants, metal pool use, 25–26 metal phytoextraction arsenic, 9–10 cadmium, 8 cobalt, 8–9 copper, 8–9 deﬁnition, 6–7 lead, 9 nickel, 7–8, 13 other metals, 10, 14

selenium, 9–10 zinc, 8 zinc and cadmium, 10–13 metals and metalloids removal technologies, 4–6 risks, 2–3 phytovolatilization mercury, 41–43 selenium, 39–41 Controlled alternative irrigation, China, 160–161 Copper hyperaccumulators, soil phytoextraction, 8–9 Crop yield, China irrigation, 158–159 D Dichapetalum gelonioides, metal tolerance, 21 Diffuse reﬂectance infrared Fourier transform spectroscopy, humic substance polymerization, 112, 115 DRIFT, see Diffuse reﬂectance infrared Fourier transform spectroscopy E EDTA chemically induced phytoextractions, 38–39 heavy metal phytoextraction, 32 lead phytoextraction, 28–31 F FAs, see Fulvic acids Field mulching techniques, China, 161–162 Five-water interaction systems, China, 146–147 Free pore water, soil–radiation interactions, 198–200 Fulvic acids, chemical properties, 60 G Gel permeation, humic substance behavior, 69–70 Geometry, remote sensing of soils, 216–218 Grevillea exul var. exul, root to shoot metal translocation, 17 Groundwater China, 147–149 North China Plain, 137–138

247

INDEX H HAs, see Humic acids Haumaniastrum katangense, copper and cobalt hyperaccumulation, 8 HBED, lead phytoextraction, 30 Heavy metals, chemically induced phytoextraction, 32–33 HEDTA, lead phytoextraction, 28 Helianthus annuus, radionuclide phytoextraction, 34 High-pressure size-exclusion chromatography, humic substances basic substances, 70–74, 96–104 preparative chromatography, 96–99 self-assembling molecules, 77–95 solute–gel–eﬂuent, 100–104 High-spectral-resolution sensors, soils basic approach, 181–183 cost and availability, 223–225 general analytical methods, 226–228 sensor types, 222–223 Horse radish peroxidase, humic substance polymerization, 110, 112, 115 HPSEC, see High-pressure size-exclusion chromatography HRP, see Horse radish peroxidase HS, see Humic substances HSR sensors, see High-spectral-resolution sensors Humic acids, chemical properties, 60 Humic substances conformational structure, 63–66 future research, 125–126 hydrophobic superstructures, role in soil, 115–125 loose, polymerization, 109–115 macromolecularity, 62–63 modern understanding, 60 molecular dimensions, 61 as polymers, 61–62 self-assembling supramolecular associations chemical and spectroscopic evidence, 105–109 overall concepts, 105 preparative HPSEC, 96–99 size-exclusion chromatography, 75–95 solute–gel–eﬂuent, 100–104 size-exclusion chromatography high-pressure type, 70–74

low-pressure type, 66–70 soil quantities, 59 Hydration water, soil–radiation interactions, 197 Hydrogen peroxide, humic substance polymerization, 110 Hydrophobic domains, humic substances, 106–107 Hydroxy-bicarboxylic malic acid, humic substance effects, 94–95 Hygroscopic water, soil–radiation interactions, 197–198 Hyperaccumulator plants metal acquisition rhizosphere processes, 22–26 metal tolerance, 17–22 metal translocation from root to shoot, 17 metal uptake, 14–17 phytoextraction arsenic, 9–10 cadmium, 8, 10–16 cobalt, 8–9 copper, 8–9 deﬁnition, 6–7 lead, 9 nickel, 7–8, 13 other metals, 10, 14 selenium, 9–10 zinc, 8 zinc and cadmium, 10–13 Hyperspectral remote sensing, soils, 181–183 I Iberis intermdia, metal hyperaccumulation, 10, 14 Iron, soil–radiation interactions, 200–203 Irrigation water, China crop yield relationship, 158–159 limited irrigation, 159 overview, 137–138 L Layer minerals, remote sensing of soil, 186 Lead, chemically induced phytoextraction, 28–32 Lead hyperaccumulators, soil phytoextraction, 9 Lecythis ollaria, selenium and arsenic hyperaccumulation, 9

248

INDEX

Liriodendron tulipifera, mercury phytovolatilization, 42 Low-pressure size-exclusion chromatography, humic substances basic substances, 66–70 self-assembling molecules, 75–77 Low-quality water, China, 149–150 M Macromolecularity, humic substances, 62–63 Marketing, water-saving products in China, 166–167 Mercury, phytovolatilization, 41–43 Metal hyperaccumulators metal acquisition rhizosphere processes, 22–26 metal tolerance, 17–22 metal translocation from root to shoot, 17 metal uptake, 14–17 phytoextraction arsenic, 9–10 cadmium, 8, 10–16 cobalt, 8–9 copper, 8–9 deﬁnition, 6–7 lead, 9 nickel, 7–8, 13 other metals, 10, 14 selenium, 9–10 zinc, 8 zinc and cadmium, 10–13 Metalloids, soil removal technologies, 4–6 risks, 2–3 Metals, soil removal technologies, 4–6 risks, 2–3 Microbes, plant hyperaccumulator rhizospheres, 24 Minerals, remote sensing of soil, 185–187 Molar absorptivity, humic molecule chromatography, 88–90 N Natural hyperaccumulation, comtaminated soils, 37–38 NCP, see North China Plain Near-infrared reﬂectance analysis, soils

channel responses, 212–213 examples, 229–231 general analytical method, 227–228 method and applications, 177–178 pixel size, 215 sampling techniques, 215 signal-to-noise ratio, 213 spatial and spectral aspects, 180 spectral resolution, 212–213 Nickel hyperaccumulators mechanism, 16–17 metal tolerance, 20–21 soil phytoextraction, 7–8, 13 NIRA, see Near-infrared reﬂectance analysis Nonclay minerals, remote sensing of soil, 186–187 North China Plain groundwater, 137–138, 148 rainfall, 144–145 soil fertility, 163 water conveyance structures, 151–152 Nuclear magnetic resonance, humic substances, 90, 92, 98–99, 108, 114, 118 O Organic acids, humic substance effects, 94 Organic matter, soil–radiation interactions, 192–195 P pH plant hyperaccumulator rhizospheres, 24 self-assembling humic molecule chromatography, 84–85 Phaseolus acutifolius, radionuclide phytoextraction, 34 Phyllanthus palawanensis, metal tolerance, 21 Physical crust, remote sensing of soil, 221–222 Phytoextraction, soils chemically induced applications, 27–28 chelating agent concerns, 38–39 heavy metals, 32–33 lead, 28–32 natural hyperaccumulation comparison, 37–38 overview, 26–27 radionuclides, 33–35

249

INDEX metal hyperaccumulators arsenic, 9–10 cadmium, 8 cobalt, 8–9 copper, 8–9 deﬁnition, 6–7 lead, 9 nickel, 7–8, 13 other metals, 10, 14 selenium, 9–10 zinc, 8 zinc and cadmium, 10–13 overview, 4–6 Phytomining, contaminated soils, 4–6, 35–36 Phytostabilization, contaminated soils, 4–5 Phytovolatilization, contaminated soils, 39–43 Plants, see Hyperaccumulator plants Polycarpaea synandra, zinc and cadmium hyperaccumulation, 8 Polymerization, loose humic substances, 109–115 Polymers, humic substances as, 61–62 Polysaccharides humic substance chromatography, 74 self-assembling humic molecule chromatography, 86 Polystyrene sulfonates humic substance chromatography, 73–74 self-assembling humic molecule chromatography, 86 Precipitation, China, 136–137, 144 PSS, see Polystyrene sulfonates Pteris vittata lead hyperaccumulation, 9 selenium and arsenic hyperaccumulation, 9 Pyrene, humic substance studies, 107 Pyrolysis-gas chromatography–mass spectrometry, humic substances, 98–99 R Radiocesium, chemically induced phytoextraction, 33–34 Radionuclides, chemically induced phytoextraction, 33–35 Rainwater, utilization in China, 144–146 RDI, see Regulated deﬁcit irrigation Regulated deﬁcit irrigation, China, 159–160 Remote sensing

hyperspectral, see Hyperspectral remote sensing soil–radiation interactions, chemical processes carbonates, 190–192 chromophores, 204–205 clay minerals, 188–190 iron, 200–203 organic matter, 192–195 salinity, 203–204 water, 195–200 soil–radiation interactions, physical processes, 205–206 soils examples, 228–231 general analytical methods, 225–228 HSR sensors cost and availability, 223–225 types, 222–223 overview, 174–176 parameters biogenic crust, 220–221 physical crust, 221–222 vegetation coverage, 218–220 problems atmospheric attenuation, 207–210 channel responses, 210–213 measurement geometry, 216–218 overview, 206–207 pixel size, 215–216 sampling techniques, 215–216 signal-to-noise ratio, 213–215 spectral resolution, 210–213 spatial and spectral aspects, 180–181 spectral chromophores, 184–187 spectral measurements, 183–184 spectroscopy in laboratory, 177–178 Removal technologies, contaminated soils, 4–6 Rhizosphere processes, metal acquisition, hyperaccumulator plants, 22–26 RI detector, self-assembling humic molecule chromatography, 82–84 Roots, hyperaccumulator plants, 23–25 RS, see Remote sensing S Salicornia bigelovii, selenium phytovolatilization, 40 Salinity, soil–radiation interactions, 203–204

250

INDEX

Scirpus robustus, selenium phytovolatilization, 40 Sebertia acuminata, nickel hyperaccumulation, 7 Selenium, phytovolatilization, 39–41 Selenium hyperaccumulators, soil phytoextraction, 9–10 Senecio coronatus, metal tolerance, 20 Signal-to-noise ratio, remote sensing of soil, 213–215 Size-exclusion chromatography, humic substances high-pressure basic substances, 70–74, 96–104 preparative chromatography, 96–99 self-assembling molecules, 77–95 solute–gel–eﬂuent, 100–104 low-pressure basic substances, 66–70 self-assembling molecules, 75–77 SNR, see Signal-to-noise ratio Soil organic matter, hydrophobic humic superstructure role organic carbon sequestration, 118–123 overview, 116 stabilization, 123–125 Soil–Plant–Atmosphere Continuum, China, 146–147 Soil–radiation interactions chemical processes carbonates, 190–192 chromophores, 204–205 clay minerals, 188–190 iron, 200–203 organic matter, 192–195 salinity, 203–204 water, 195–200 physical processes, 205–206 Soils China fertility, 162–163 moisture conservation, 161–162 contaminated, see Contaminated soils humic substance quantities, 59 hydrophobic humic superstructure role organic carbon sequestration, 118–123 overview, 115–118 stabilization, 123–125 remote sensing atmospheric attenuation, 207–210 biogenic crust, 220–221

channel responses, 210–213 examples, 228–231 general analytical methods, 225–228 hyperspectral sensing, 181–183 measurement geometry, 216–218 overview, 174–176, 206–207 physical crust, 221–222 pixel size, 215–216 sampling techniques, 215–216 sensor cost and availability, 223–225 sensor types, 222–223 signal-to-noise ratio, 213–215 soil crust, 220–222 soil surface, 220–222 spatial and spectral aspects, 180–181 spectral chromophores, 184–187 spectral measurements, 183–184 spectral resolution, 210–213 vegetation coverage, 218–220 spectroscopy in laboratory, 176–178 SOM, see Soil organic matter SPAC, see Soil–Plant–Atmosphere Continuum Streptanthus polygaloides nickel hyperaccumulation, 13 root to shoot metal translocation, 17 Surface water, China, 147–149 T Thlaspi alpestre, see Thlaspi caerulescens Thlaspi arvense, root exudates, 25 Thlaspi caerulescens metal hyperaccumulation metal tolerance, 17–22 metal translocation from root to shoot, 17 metal uptake, 14–17 natural vs. chemically enhanced phytoextraction, 37 overview, 6 nickel hyperaccumulation, 13 rhizosphere pH, 24 root exudates, 24–25 rooting pattern, 23 zinc and cadmium hyperaccumulation, 8, 10–13 Thlaspi calaminare, see Thlaspi caerulescens Thlaspi goesingense metal tolerance, 20–21 root exudates, 25

251

INDEX Thlaspi ochroleucum, zinc and cadmium hyperaccumulation, 8 Thlaspi rotundifolium spp. cepaeifolium, lead hyperaccumulation, 9 TSK column, self-assembling humic molecule chromatography, 78–81 Typha latifolia, selenium phytovolatilization, 40 U Ultraviolet detector, self-assembling humic molecule chromatography, 84–85 Uranium, chemically induced phytoextraction, 34–35 V Vegetation, remote sensing of soil, 218–220 Viewing angle, remote sensing of soils, 217–218 Viola calaminaria, as metal hyperaccumulator, 6 W Walsura monophylla, metal tolerance, 21 Water, soil–radiation interactions, 195–200 Water conveyance structures, China, 151–152 Water-matched agricultural structure, China, 154–155 Water-matched planting, China, 155–156 Water resources, China biological water-saving technology, 156–158 ﬁve-water interaction systems, 146–147 groundwater, 147–149 irrigation water, 158–159 low-quality water use, 149–150 NCP, groundwater, 137–138 prices and fees, 164–165 rainwater utilization, 144–146 resource management, 138 SPAC interface processes, 146–147 surface water, 147–149 Water-saving agronomy, China

agricultural structure, 154–155 biological water-saving technology, 156–158 expert systems, 166 irrigation schemes, 158–161 planting, 155–156 soil fertility, 162–163 soil moisture conservation, 161–162 system speciﬁcations, 139–140 Water-saving breeding, China, 157–158 Water-saving engineering, China conveyance structures, 151–152 irrigation techniques, 152–154 Water-saving irrigation, China, 152–154, 158–161 Water-saving management, China administrative systems, 163–164 expert systems, 166 laws and regulations, 165 prices and fees, 164–165 products and marketing, 166–167 technical training and services, 165–166 Water-use efﬁciency, China canopy level, 142 community level, 142 crop yield–irrigation water relationship, 158 ﬁeld level, 142 molecular level, 141 overview, 138, 140–141 regional level, 143 single-leaf level, 141–142 soil fertility, 162–163 water-saving breeding, 157–158 WUE, see Water-use efﬁciency Z Zea mays, lead hyperaccumulation, 9 Zinc hyperaccumulators mechanism, 14–16 metal tolerance, 20–22 soil phytoextraction, 8, 10–13