ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT IN COMPLEX ENVIRONMENTAL SETTINGS S. Mahimairaja,1,* N. S. Bolan,1 D. C. A...
74 downloads
1642 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
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT IN COMPLEX ENVIRONMENTAL SETTINGS S. Mahimairaja,1,* N. S. Bolan,1 D. C. Adriano2 and B. Robinson3 1
Institute of Natural Resources, Massey University, Palmerston North, New Zealand 2 Savannah River Ecology Laboratory, The University of Georgia, Aiken, South Carolina 29802 3 HortResearch, Palmerston North, New Zealand
I. Introduction II. Origin and Sources of Arsenic Contamination A. Geogenic B . Anthropogenic C . Biogenic Redistribution III. Distribution and Speciation of Arsenic in the Environment A. Distribution in Soil B . Distribution in the Aquatic Environment C . Chemical Form and Speciation IV. Biogeochemistry of Arsenic in the Environment A. Biogeochemistry of Arsenic in the Soil B . Biogeochemistry of Arsenic in Aquatic Environments V. Bioavailability and Toxicity of Arsenic to Biota A. Toxicity to Plants and Microorganisms B . Risk to Animals and Humans VI. Risk Management of Arsenic in Contaminated Environments A. Remediation of Arsenic-Contaminated Soil B . Removal of Arsenic from Aquatic Environments C . Multiscalar-Integrated Risk Management VII. Summary and Future Research Needs Acknowledgments References
Contamination of terrestrial and aquatic ecosystems by arsenic (As) is a very sensitive environmental issue due to its adverse impact on human health. Although not anthropogenic in origin, the problem of As contamination in
*Current address: Department of Environmental Sciences, Tamil Nadu Agricultural University, Coimbatore 641003, India. 1 Advances in Agronomy, Volume 86 Copyright 2005, Elsevier Inc. All rights reserved. 0065-2113/05 $35.00
2
S. MAHIMAIRAJA ET AL. groundwaters of West Bengal (India) and Bangladesh has been considered of calamitous proportion because significant segment of the population is at high risk, with untold numbers already suffering from irreversible effects of As poisoning. Elsewhere, indiscriminate disposal of industrial and mining wastes has led to extensive contamination of lands, thereby exacerbating the potential for food chain contamination. With greater public awareness of As poisoning in animal and human nutrition, there has been a growing interest in developing regulatory guidelines and remediation technologies for mitigating As-contaminated ecosystems. Although the immediate needs revolve around the stripping of As from domestic water supplies as exemplified by the affected areas in Bangladesh and West Bengal, a remediation scheme should also be explored to be able to cope with pivotal needs to abate the contamination of soils, sediments, and water and the potential to compromise the quality of the food chain. A range of technologies, including bioremediation, has been applied with varying levels of success either to remove As from the contaminated medium or to reduce its biotoxicity. This review provides general overview of the various biogeochemical processes that regulate As bioavailability to organisms, including microbes, plants, animals and humans. In turn, the role of the source term, chemical form, and chemical species of As are discussed as an overture to As bioavailability. Having laid the fundamental mechanisms and factors regulating As bioavailability, we then assembled the various physical, chemical, and biological mitigative methods that have been demonstrated, some being practical, highlighting their special strengths and potential for more effective and economical widespread applications. Because of the complexity involved in dealing with contaminated sites, exacerbated by site characteristics, nature of hydrogeology, source term, chemical form, land use, and so on, no one remedial technology might suffice. Therefore, we have attempted to offer an “integrated” approach of employing a combination of technologies at multiscalar levels, depending on extenuating circumstance, with the aim of securing viable methods, economically and technologically. Future research needs, especially in the area of As bioavailability and remediation strategies, are identified. ß 2005, Elsevier Inc.
Arsenic is a unique carcinogen. It is the only known human carcinogen for which there is adequate evidence of carcinogenic risk by both inhalation and ingestion. While arsenic is released to the environment from natural sources such as wind-blown dirt and volcanoes, releases from anthropogenic sources far exceed those from natural sources. Oral exposure of arsenic to human beings however, is usually not the result of anthropogenic activity as it is with many carcinogens, but the result of natural contamination of well-water supplies by arsenic-rich geologic strata. Centeno et al. (2002)
I. INTRODUCTION Arsenic (As) is a toxic metalloid found in rocks, soil, water, sediments, and air. It enters into the terrestrial and aquatic ecosystems through a
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
3
combination of natural processes such as weathering reactions, biological activity, and volcanic emissions, as well as a result of anthropogenic activities. Excessive use of As-based pesticides and indiscriminate disposal of domestic (sewage) and industrial (timber, tannery, paints, electroplating, etc.) wastes, as well as mining activities, have resulted in widespread As contamination of soils and waterways. Arsenic in terrestrial and aquatic ecosystems attracts worldwide attention primarily because of its adverse impact on human health. The general population may be exposed to As from air, food, and water (Adriano, 2001; Sparks, 1995). Of the various sources of As in the environment, drinking water probably poses the greatest threat to human health (Smedley and Kinniburgh, 2002). People drinking As-contaminated water over prolonged periods often show typical arsenical lesions, which are a late manifestation of As toxicity. Arsenic has been unequivocally demonstrated to be both toxic and carcinogenic to humans and animals. Although trace levels of As have been shown to be beneficial in plant and animal nutrition (Leonard, 1991; Smith et al., 1998; USEPA, 1993), no comparable data are available for humans (Adriano, 2001), and elevated concentrations of As in the biosphere pose a significant threat to mankind. Arsenic contamination of surface and groundwaters occurs worldwide and has become a sociopolitical issue in several parts of the globe. For example, several million people are at risk from drinking As-contaminated water in West Bengal (India) (Chakraborti et al., 2002; Chatterjee et al., 1995) and Bangladesh (Smith et al., 2000). Scores of people from China (Wang, 1984), Vietnam (Berg et al., 2001), Taiwan (Lu, 1990), Chile (Smith et al., 1998), Argentina (Hopenhayn-Rich et al., 1998), and Mexico (Del Razo et al., 1990) are likely at risk as well. The problem of As contamination in groundwaters of West Bengal and Bangladesh has been considered of calamitous proportion because a significant segment of the population is at high risk, with untold number already suffering from irreversible effects of As poisoning (Chatterjee et al., 1995). “For many people in Bangladesh it can sometimes literally be a choice between death by arsenic poisoning and death by diarrhea,” says Timothy Claydon, country representative of Water Aid (http://Phys4. Harvard.Edu/Wilson/Arsenic). Elsewhere, indiscriminate disposal of industrial and mining wastes has led to extensive contamination of lands. Consequently, thousands of As-contaminated sites have been reported around the world (Eisler, 2004; ETCS, 1998; Smith et al., 1998; USEPA, 1997). The economic consequences of As contamination include loss of productivity, healthcare costs, and, most importantly, imposition of As contamination as a nontariff trade barrier, preventing export sales to some countries. With greater public awareness of As poisoning in animal and human nutrition, there has been growing interest in developing guidelines and remediation technologies for mitigating As-contaminated ecosystems. A
4
S. MAHIMAIRAJA ET AL.
range of technologies, including chemical immobilization and bioremediation, has been applied with varying levels of success either to completely remove As from the system or to reduce its biotoxicity. Phytoremediation, an emerging form of bioremediation technology that uses plants to remove or stabilize contaminants, may offer a low-cost and ecologically viable means for the mitigation of As toxicity in the environment. There have been a number of reviews on As in soil (Matschullat, 2000; Smith et al., 1998) and aquatic (Korte and Fernando, 1991; Smedley and Kinniburgh, 2002) environments. However, there has been no comprehensive review on the biogeochemistry and transformation of As in relation to its remediation. The present review, therefore, aims to integrate fundamental aspects of As transformation and recent developments on As speciation in relation to remediation strategies for the risk management of As-contaminated terrestrial and aquatic ecosystems. The review first discusses the various sources and distribution of As in soil, sediments, and water. The transformation of As in these systems is examined in relation to As speciation and bioavailability. The detrimental effects of As on plant growth, microbial functions, and animal and human health are discussed with relevant examples. Various physical, chemical, and biological techniques available for remediation of As-contaminated sites are synthesized with an aim to develop integrated practical strategies at multiscalar levels to manage As-contaminated sites. Future research needs, especially in the area of As bioavailability and long-term remediation strategies, are identified. The review encourages greater interaction among soil scientists, agronomists, aquatic biogeochemists, and environmental and resource engineers in devising risk management strategies to resolve one of the worst environmental calamities of the 21st century.
II.
ORIGIN AND SOURCES OF ARSENIC CONTAMINATION
A range of As compounds, both organic and inorganic, are introduced into the environment through geological (geogenic) and anthropogenic (human activities) sources (Fig. 1). Small amounts of As also enter the soil and water through various biological sources (biogenic) that are rich in As (Table I). Although the anthropogenic source of As contamination is increasingly becoming important, it should be pointed out that the recent episode of extensive As contamination of groundwaters in Bangladesh and West Bengal is of geological origin, transported by rivers from sedimentary rocks in the Himalayas over tens of thousands of years, rather than anthropogenic.
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
Figure 1
5
Major sources and routes of arsenic in soil and aquatic ecosystems.
A. GEOGENIC Arsenic is widely distributed in all geological materials at varying concentrations. An average concentration of 1.5 to 2.0 mg As kg1 is expected in the continental crust of the earth. The mean concentrations of As in igneous rocks range from 1.5 to 3.0 mg kg1, whereas in sedimentary rocks range from 1.7 to 400 mg kg1 (Smith et al., 1998). Arsenic ranks 52nd in crustal abundance and it is a major constituent in more than 245 minerals (O’Neill, 1995). These are mostly sulfide-containing ores of copper (Cu), nickel (Ni), lead (Pb), cobalt (Co), zinc (Zn), gold (Au), or other base metals. The most important ores of As include pyrites, realgar, and orpiment. Arsenic is introduced into soil and water during the weathering of rocks and minerals followed by subsequent leaching and runoff. Therefore, the primary source of As in soil is the parent (or rock) materials from which it is derived (Yan-Chu, 1994). Geogenic contamination of As in soils (Table II)
6
S. MAHIMAIRAJA ET AL. Table I Selected References on Sources of Arsenic in Soil and Aquatic Environments
Source Broiler litter Cattle manure (composted) Coal Cow manure Dikes and ores Earthworms Fly ash FYM from cattle Lake weeds Metallurgical ore waste Mine spoils Mine tailing Mushroom (edible) from contaminated soil Poultry manure Rice straw Sewage sludge
Concentration (mg kg1)
Reference
34.6 3.0–5.2
Jackson and Miller (2000) Raven and Loeppert (1997)
2–825 15,005 6–8.5 1242–30,800 1358 2–6300 0.8–2.6 83–1262 52,700–63,000 >20,000 62,350 7000 1420
Adriano et al. (1980) Bencko and Symon (1977) Raven and Loeppert (1997) Ongley et al. (2003) Langdon et al. (2002) Page et al. (1979) Nicholson et al. (1999) Aggett and Aspell (1980) Magalhaes et al. (2001) Porter and Peterson (1975) Kim et al. (2002) Roussel et al. (2000) Larsen et al. (1998)
50 16.8 91.8 11.9–21.0
Arai et al. (2003) Jackson and Bertsch (2001) Abedin et al. (2002) Department of Health (NZ) (1992); Ross et al. (1991) Caper et al. (1978); Raven and Loeppert (1997)
8.1–14.3
and water (Table III) has been reported in many parts of the world. One typical example is the extensive As contamination of groundwaters in Bangladesh and West Bengal in India. Based on As geochemistry, three probable mechanisms have been offered for As mobility in groundwaters of West Bengal and Bangladesh (Bose and Sharma, 2002): i. Mobilization of As due to the oxidation of As-bearing pyrite minerals. Insoluble As-bearing minerals such as arsenopyrite (FeAsS) are rapidly oxidized [Eq. (1)] when exposed to atmosphere, releasing soluble arsenite [As(III)], sulfate (SO2 4 ), and ferrous iron [Fe(II)] (Mandal et al., 1996). The dissolution of these As-containing minerals is highly dependent on the availability of oxygen and the rate of oxidation of sulfide (Loeppert, 1997). The released As(III) is partially oxidized to arsenate [As(V)] by microbially mediated reactions (Wilkie and Hering, 1998).
Table II Selected References on Arsenic Concentration in Contaminated Soils Reference
Australia Australia Australia (NSW) Australia (NSW) Austria Bangladesh Belgium Belgium Brazil China England England (southwest) Germany Ghana Ghana India (West Bengal India (West Bengal) Japan Mexico Mexico New Zealand
Tannery wastes Arsenical pesticides Mining and processing of arsenopyrite ore Cattle dip Ore vein Geological Metal alloy and metallurgical industries Arsenic factory Metallurgical plant wastes Wastewater Tin, copper, and arsenic mining Geological Storage of organoarsenic-based chemical warfare agents Mining Mining Geological (through irrigation water) Disposal from arsenical pesticides manufacturing Arsenic mine and smelter Mining activities Runoff from mining waste Timber treatment with CCA
Slovakia Thailand (southern) USA USA (Colorado) USA (Florida) USA (Louisiana) USA (southern California)
Coal-burning power station Geologcal Mine tailing Pesticide spray Industrial activities Arsenic dipping vat Crude oil storage facility
<1–435 9.8–124 9300 37–3542 700–4000 1.7–56.7 36,000 25,000 –35,000 636–748 40–120 120–52,600 110 Up to 250,000 (mean 923) 2.1–48.9 189–1025 11.5–28.0 20,100–35,500 391–459 14,700 >2.0 6100 161–790 376–10,440 80–5475 8.8–139 Up to 5000 48–3421 >1000 0.2–660 555 30–2300
Sadler et al. (1994) Bishop and Chisholm (1961) Ashley and Lottermoser (1999) McLaren et al. (1998) Geiszinger et al. (2002) Alam and Sattar (2000) Cappuyns et al. (2002) Dutre et al. (1998) Magalhaes et al. (2001) Jiang and Ho (1983) Kavanagh et al. (1997) Mitchell and Barr (1995) Pitten et al. (1999) AmonooNeizer et al. (1996) Bowell et al. (1994) Amit et al. (1999); Chatterjee and Mukherjee (1999) Roychowdhury et al. (2002) Hiroki (1993) Ongley et al. (2003) Naranjo-Pulido et al. (2002) CMPS & F (1995) Yeates et al. (1994) Armishaw et al.(1994) McLaren (1992) Keegan et al. (2002) Williams et al. (1996) Jones et al. (1997) Folkes et al. (2001) Chirenje et al. (2003) Masscheleyn et al. (1991) Wellman et al. (1999)
7
As content (mg kg1)
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
Source of contamination
Country
8
Table III Selected References on Arsenic Concentration in Contaminated Aquatic Media
Argentina Australia Australia (NSW)
Bangladesh Bangladesh Bangladesh Bangladesh Bangladesh Brazil Chile Chile Chile (north) China (inner Mongolia) England (SW) Germany (northern Bavaria) India (West Bengal)
Water source
Source of contamination
As content (g liter1)
Reference
Groundwater River sediments Water sample from a mine shaft and waste dump seepage Tube wells Tube wells Groundwater Tube wells Tube well water River sediments Natural water Drinking water Drinking water Groundwater
Geological Mining Geological
3000 32.8–42.7 13,900
Sbarato and Sanchez (2001) Taylor (1996) Ashley and Lottermoser (1999)
Geological Geological Geological Geological Geological Metallurigical plant Geological Geological Geological Geological
260–830 >50 0.7–640 1–535 0.01–0.071 347 mg kg1 950–13,080 750–800 600 1088–1354
Ali and Tarafdar (2003) Yokota et al. (2002) Frisbie et al. (2002) Watanabe et al. (2001) Alam and Sattar (2000) Magalhaes et al. (2001) Munoz et al. (2000) Smith et al. (2000) Hopenhayn-Rich et al. (1996) Guo et al. (2001)
River Deep water wells
Tin mine drainage Geological
Dissolved As(III) 240 10–150
Hunt and Howard (1994) Heinrichs and Udluft (1999)
Groundwater
Geological
0.5–135.9
Nag et al. (1996)
S. MAHIMAIRAJA ET AL.
Country
Tube well water Drinking water Tube well water Tube well water Tube well water Groundwater Groundwater Well water Tube wells Tube wells River, Waikato Lake Ohakuri Sediments from Waikato river Deep well water Well water Geothermal water Lake Groundwater from a confined sandstone aquifer Groundwater
Geological Geological Geological Geological Geological Geological Geological Geological Geological Geological Geothermal release Geothermal release Geothermal Geological Geological Geological Geological Geological
22–2000 212 82–170 85 2.7–170 200–3700 293 267–1070 >10 >50 3–121 37–60 8700–156,100 >10 671 1135 200 mol liter1 12,000
Mazumder et al. (1988) Mahata et al. (2003) Roychoudhury et al. (2002a) Roychoudhury et al. (2002b) Tokunaga et al. (2002) Mandal et al. (1996) Kondo et al. (1999) Gomez-Arroyo et al. (1997) Neku and Tandukar (2003) Shrestha et al. (2003) Robinson et al. (1995) Aggett and Aspell (1980) Robinson et al. (1995) Wai et al. (2003) Chen et al. (1995) Buyuktuncel et al. (1997) Oremland et al. (2000) Schreiber et al. (2000)
16–176
Nimick (1998)
Groundwater
Natural hydrological and geochemical Geological
USA (New England) USA (New Hampshire) Vietnam
>10
Ayotte et al. (2003)
Well water
Geological
0.003–180
Peters et al. (1999)
Tube well water
Geological
1–3050
Berg et al. (2001)
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
India (West Bengal) India (West Bengal) India (West Bengal) India (West Bengal) India (West Bengal) India (West Bengal) Japan Mexico Nepal Nepal New Zealand New Zealand New Zealand Taiwan Taiwan Turkey USA (California) USA (eastern Wisconsin) USA (Madison)
9
10
S. MAHIMAIRAJA ET AL. þ FeAsS þ 13Fe3þ þ 8H2 O ! 14Fe2þ þ SO2 4 þ 13H þ H3 AsO4 ðaqÞ
ð1Þ
ii. Dissolution of As-rich iron oxyhydroxides (FeOOH) due to onset of reducing conditions in the subsurface. Under oxidizing conditions, and in the presence of Fe, inorganic species of As are predominantly retained in the solid phase through interaction with FeOOH coatings on soil particles. The onset of reducing conditions in such environments can lead to the dissolution of FeOOH coatings. Fermentation of peat in the subsurface releases organic molecules (e.g., acetate) to drive reductive dissolution of FeOOH, resulting in the release of Fe(II), As(III), and As (V) present on such coatings [Eq. (2)] (McArthur et al., 2000; Nickson et al., 2000). 8FeOOH AsðsÞ þ CH3 COOH þ 14H2 CO3 ! 8Fe2þ þ AsðdÞ þ 16HCO 3 þ 12H2 O
ð2Þ
where As(s) is sorbed As and As(d) is dissolved As. iii. Release of As sorbed to aquifer minerals by competitive exchange with phosphate (H2 PO 4 ) ions that migrate into aquifers from the application of fertilizers to surface soil (Acharya et al., 1999). However, the second mechanism involving dissolution of FeOOH under reducing conditions is considered to be the most probable reason for excessive As accumulation in groundwater (Harvey et al., 2002; Smedley and Kinniburgh, 2002). Relatively high concentrations of naturally occurring As can appear in some areas as a result of inputs from geothermal sources or As-rich groundwaters (Smedley and Kinniburgh, 2002). For example, Robinson et al. (1995) found high As concentrations (3800 g liter1) in waste geothermal brine from the main drain at Wairakei geothermal field in New Zealand. River and lake waters receiving inputs of geothermal waters were found to contain up to 121 g As liter1. Arsenic concentration is usually higher in soil and shales than in earth crust because of its continuous accumulation during weathering and translocation in colloidal fractions. Arsenic may also be coprecipitated with Fe hydroxides and sulfides in sedimentary rocks. Therefore, Fe deposits and sedimentary Fe ores are rich in As, and the soils derived from such sedimentary rocks may contain as high as 20 to 30 mg As kg1 (Zou, 1986). Arsenic in the natural environment occurs in soil at an average concentration of about 5 to 6 mg kg1 (i.e., background level), but this varies among geological regions (Peterson et al., 1981). Volcanoes are also considered as a geogenic source of As to the environment with the total atmospheric annual
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
11
emissions from volcanoes being estimated at 31,000 mg (Smith et al., 1998; Walsh et al., 1979).
B. ANTHROPOGENIC Arsenic is also being introduced into the environment through various anthropogenic activities. These sources release As compounds that differ greatly in chemical nature (speciation) and bioavailability. Major sources of As discharged onto land originate from commercial wastes (40%), coal ash (22%), mining industry (16%), and the atmospheric fallout from the steel industry (13%) (Eisler, 2004; Nriagu and Pacyna, 1988). Arsenic trioxide (As2O3) is used extensively in the manufacturing of ceramic and glass, electronics, pigments and antifouling agents, cosmetics, fireworks, and Cubased alloys (Leonard, 1991). Arsenic is also used for wood preservation in conjunction with Cu and chromium (Cr), i.e., copper–chromium–arsenate (CCA). Some important physicochemical properties of As compounds are presented in Table IV. Industries that manufacture As-containing pesticides and herbicides release As-laden liquid and solid wastes that, upon disposal, are likely to contaminate soil and water bodies. For example, indiscriminate discharge of industrial effluents from the manufacturing of Paris Green (copper acetoarsenite, an arsenical pesticide) resulted in the contamination of soil and
Table IV Physicochemical Properties of Arsenic Compoundsa
Compounds Arsenic–As (element) Arsenic trioxide or arsenous oxide–As2O3 Arsenic oxide or arsenic pentoxide–As2O5 Arsenic sulfide or arsenic trisulfide–As2S3 Dimethylarsinic acid or cacodylic acid (CH3)2AsO(OH) Arsenate or salts of arsenic acid–HAsO4 a
Density (g cm3)
Water solubility (g liter1)
Melting point ( C)
Boiling point ( C)
5.727 3.738
Insoluble 37 at 20 C
613 312.3
– 465
4.32
1500 at 16 C
–
3.43
5104 at 18 C
315 (decomposes) 300
–
829 at 22 C
200
–
5.79
Very slightly
720 (decomposes)
–
From Lide (1992) and IARC (1980).
707
12
S. MAHIMAIRAJA ET AL.
groundwater in residential area of Calcutta, India (Chatterjee et al., 1999). Similarly, in New Zealand, timber treatment effluent is considered to be the major source of As contamination in aquatic and terrestrial environments (Bolan and Thiyagarajan, 2001). Because As is widely distributed in the sulfide ores of Pb, Zn, Au, and Cu, it is released during their mining and smelting processes. The flue gases and particulate from smelters can contaminate nearby ecosystems downwind from the operation with a range of toxic metal(loid)s, including As (Adriano, 2001). Coal combustion not only releases gaseous As into the atmosphere, but also generates fly and bottom ash containing varied amounts of As. Disposal of these materials often leads to As contamination of soil and water (Beretka and Nelson, 1994). Arsenic is present in many pesticides, herbicides, and fertilizers. The use of horticultural pesticides, lead arsenate (PbAsO4), calcium arsenate (CaAsO4), magnesium arsenate (MgAsO4), zinc arsenate (ZnAsO4), zinc arsenite [Zn(AsO2)2], and Paris Green [Cu(CH3COO)2.3Cu(AsO2)2] in orchards has contributed to soil As contamination in many parts of the world (Merry et al., 1983; Peryea and Creger, 1994). Soil contamination due to the use of organoarsenical herbicides such as monosodium methanearsonate (MSMA) and disodium methanearsonate (DSMA) was also reported (Gilmore and Wells, 1980; Smith et al., 1998). The use of sodium arsenite (NaAsO2) to control aquatic weeds has contaminated small fish ponds and lakes in several parts of United States with As (Adriano, 2001). Arsenic contamination in soil was also reported due to the arsenical pesticides used in sheep and cattle dips to control ticks, fleas, and lice (McBride et al., 1998; McLaren et al., 1998). A study of 11 dip sites in New South Wales indicated considerable surface soil (0 –10 cm) contamination with As (37–3542 mg kg1) and significant movement of As (57–2282 mg kg1) down the soil profile at 20–40 cm depth (McLaren et al., 1998). Continuous application of fertilizers that contain trace levels of As also results in As contamination of soil, thereby reaching the food chain through plant uptake (McLaughlin et al., 1996).
C. BIOGENIC REDISTRIBUTION Biological sources contribute only small amounts of As into soil and water ecosystems. However, plants and micro- and macroorganisms affect the redistribution of As through their bioaccumulation (e.g., biosorption), biotransformation (e.g., biomethylation), and transfer (e.g., volatilization). Arsenic accumulates readily in living tissues because of its strong affinity for proteins, lipids, and other cellular components (Ferguson and Gavis, 1972). Aquatic organisms are particularly known to accumulate As, resulting in considerably higher concentrations than in the water in which they live (i.e.,
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
13
biomagnification). Upon disposal or consumption they subsequently become a source of environmental contamination. Arsenic could be transferred from soil to plants and then to animals and humans, involving terrestrial and aquatic food chains. For example, poultry manure addition is considered to be one of the major sources of As input to soils. In the Delaware–Maryland– Virginia peninsula along the eastern shore of the United States, 20–50 mg of As is introduced annually to the environment through the use of As compounds (e.g., Roxarsone, ROX) in poultry feed (Christen, 2001). However, in many situations the soil–plant transfer of As is low (Smith et al., 1998) and it is important to recognize that metal(loid)s loading through manure application may overestimate their actual net accumulation in soil, as a substantial portion of the metal(loid)s in manure originate in crop uptake and are therefore being recycled within a production system (Bolan et al., 2004).
III. DISTRIBUTION AND SPECIATION OF ARSENIC IN THE ENVIRONMENT A. DISTRIBUTION IN SOIL Generally, As concentrations in uncontaminated soils seldom exceed 10 mg kg1. However, anthropogenic sources of As have elevated the background concentration of As in soils (Adriano, 2001). For example, in areas near As mineral deposits, As levels in soils may reach up to 9300 mg kg1 (Ashley and Lottermoser, 1999). The distribution of As in contaminated soils around the world is presented in Table II. Depending on the nature of the geogenic and anthropogenic sources, As concentration in soils can range from <1 to 250,000 mg kg1. However, there is a large fluctuation among countries due to variation in soil parent material, for example, calcareous soils can be expected to have higher levels of As than noncalcareous soils (Aichberger and Hofer, 1989). As discussed in Section II,B, although the dominant source of As in soils is geological, additional inputs may also be derived locally from industrial sources, such as smelting and fossilfuel combustion products and agricultural sources, namely pesticides and phosphatic fertilizers. In soils, As forms a variety of inorganic and organic compounds (Vaughan, 1993). Arsenic forms solid precipitates with Fe, aluminium (Al), calcium (Ca), magnesium (Mg), and Ni. A number of studies involving solidphase speciation have shown that As is prevalent mostly in the oxalate fractions associated with amorphous and crystalline Fe and Al oxides, indicating the strong affinity of As for these soil components (Wenzel et al., 2001). The soluble As concentration in soil is largely determined by
14
S. MAHIMAIRAJA ET AL.
redox conditions, pH, biological activity, and adsorption reactions. The adsorption and mobility of As in soil are affected more strongly by the presence of H2PO 2 ion than any other anions. Arsenic is subject to both chemical and biological transformations in soils, resulting in the formation of various species.
B. DISTRIBUTION
IN THE
AQUATIC ENVIRONMENT
Arsenic in an aquatic environment is distributed in both the aqueous solution and sediments. Elevated concentrations of As in natural waters are usually associated with As-rich sedimentary rocks of marine origin, weathered volcanic rocks, fossil fuels, geothermal areas, mineral deposits, mining wastes, agricultural use, and irrigation practices (Korte and Fernando, 1991). Uncontaminated waters usually contain less than 0.001 g As liter1. In contaminated areas, however, high levels of As have been reported in water bodies (Table III). It should be noted that considerable variation in As concentration exists within the same geological area as reported by different researchers. The World Health Organization (WHO, 1981) recommends that the As concentration in drinking water not exceed 10 g liter1. However, the limit in many countries, including Bangladesh and the United States, is still 50 g As liter1. The widespread occurrence of high concentrations of As in water in many parts of the world caused the U.S. President George W. Bush to state “Arsenic is a natural substance that sometimes causes problems,” and to reverse the previous government’s decision to accept a five times lower WHO standard (i.e., 10 g liter1) (Kaiser, 2001). As discussed earlier, one of the principal causes of high As concentrations in subsurface waters is the reductive dissolution of hydrous Fe oxides and/or the release of adsorbed As (Smedley and Kinniburgh, 2002). Deuel and Swoboda (1972) proposed that the release was primarily due to reduction (and dissolution) of “ferric arsenates” instead of changes in the As speciation. The high As in groundwater can be associated with reducing conditions, resulting from the presence of dissolved organic carbon, particularly in alluvial and delta environments. The groundwater of the Bengal basin is the most notable example. While the exact mechanisms responsible for this remain uncertain, it is possible that both reductive dissolution and desorption of As from oxides and clay play an important role in elevating As concentration (Smedley and Kinniburgh, 2002). A significant proportion of As in aquatic environment is derived from the sediments, and the relative distribution of As in water and sediments depends mainly on the nature and amounts of sediments (Table III). Arsenic in river sediments is highly variable, ranging from 32.8–42.7 mg kg1 (Australia) (Taylor, 1996) to 8700–156100 mg kg1 (New Zealand)
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
15
(Robinson et al., 1995). The As-rich sediments act as a buffer in maintaining the As concentration in water bodies, thereby controlling the dynamics and bioavailability of As in the aquatic environment.
C. CHEMICAL FORM AND SPECIATION Speciation of metal(loid)s can be achieved by both analytical processes and on the basis of theoretical consideration. The analytical processes involved in the speciation of metal(loid)s in soils can be grouped into solidphase speciation and solution-phase speciation. In view of the limitations of many of the analytical procedures used in speciation, often species distribution is predicted using a number of speciation models (e.g., GEOCHEM by Mattigod and Sposito, 1979; MINTEQ2 by Allison et al., 1991) that are based on theoretical chemical (thermodynamic) concepts. Although the fundamental thermodynamic principles that drive these models are based on scientific facts, problems arise when these principles are applied to complex natural matrixes. A large number of sequential extraction schemes have been used for soils, generally attempting to identify metal(loid)s held in any of the following fractions: soluble, exchangeable, sulfide/carbonate bound, organically bound, oxides bound, and residual or lattice mineral bound. The bioavailability of metal(loid)s in soils has been examined using the physiologically based in vitro chemical fractionation schemes that include the physiologically based extraction test (PBET), potentially bioavailable sequential extraction (PBASE), and gastrointestinal (GI) test. These innovative tests predict the bioavailability of metal(loid)s in soil/sediments when ingested by animals and humans. A vast number of analytical techniques are available for solution-phase characterization and quantification of metal(loid)s. These include electroanalytical techniques, cation/anion-exchange resins and chemical adsorbents to fractionate ionic and nonionic forms, ultrafiltration, dialysis, and gel permeation techniques for molecular size fractionation, spectroscopic techniques measuring the oxidation state of elements, X-ray techniques to measure trace element distribution, and chromatographic techniques to measure the phase distribution of metal(loid)s. Arsenic speciation is determined by both biotic and abiotic variables. Arsenic speciation is important not only for understanding the biogeochemical cycling of As in different ecosystems and mechanisms of As accumulation and detoxification, but also for designing safe disposal options of As-rich biomass (Tu et al., 2003; Watt and Le, 2003). In soil, As occurs both as inorganic [As(III) and As(V)] and as organic forms. Trivalent As can exist as arsenous oxide (As2O3), arsenious acid
16
S. MAHIMAIRAJA ET AL.
2 3 (HAsO2), arsenite (H2AsO 3 , HAsO3 , AsO3 ) ions, arsenic trichloride (AsCl3), arsenic sulfide (AsS3), and arsine (AsH3). Pentavalent As commonly occurs as arsenic pentoxide (As2O5), orthoarsenic acid (H3AsO4), metaarse2 3 nic acid (HAsO3), and arsenate (H2AsO 4 , HAsO4 , AsO4 ) ions. The presence of different forms of organic As, such as monomethylarsonic acid [MMA, CH3AsO(OH)2], dimethylarsenic acid [DMA, (CH3)2AsO(OH)], trimethylarsine oxide [(CH3)3AsO], methylarsine (CH3AsH2), dimethylarsine [(CH3)2AsH] and trimethylarsine [TMA, (CH3)3As], has also been observed in contaminated soil and water (Gao and Burau, 1997). The most common forms of As in the environment are the inorganic oxyions of As(III) and As(V). Arsenite [As(III)] is more toxic and relatively mobile in contaminated soils, whereas arsenate [As(V)] is relatively less toxic. Both As(III) and As(V) compounds are highly soluble in water and may change valency states depending on the pH and redox conditions. Results of a literature search on the speciation of As in environmental and biological samples are presented in Table V. In contaminated soils, generally As(V) predominates over As(III), whereas in waters, the relative proportion of these two species varies depending on a number of factors, including As sources, redox potential, pH, and microbial activity. Masscheleyn et al. (1991) studied the influence of redox potential and pH on As speciation and solubility in a contaminated soil. They observed that alterations in the oxidation state of As, as influenced by redox potential and pH, greatly affected its solubility in soil. At oxic redox levels (500–200 mV), As solubility was low and the major part (65–98%) of the As in soil solution was present as As(V). At alkaline pH, the reduction of As(V) to As(III) released substantial proportions of As into solution. Under moderately reducing conditions (0–100 mV), As solubility was controlled by the dissolution of Fe oxyhydroxides. At an anoxic redox level of 200 mV, soluble As increased 13-fold as compared to an oxic redox level of 500 mV. The apparent slow kinetics of the As(V) to As(III) transformation and the high concentrations of manganese (Mn) present indicate that, under reducing conditions, As solubility could be controlled by the Mn3(AsO4)2 phase. In a study conducted in New Zealand, Aggett and Aspell (1976) showed that with the occasional exception of a few summer months, over 90% of the As in water of the Waikato River and dams was present as As(V). Freeman (1985) detected As(III) in the Waikato River only when cyanobacteria (Anabaena oscillaroides) reduced As(V) to As(III). While reviewing the As cycle in natural waters, Ferguson and Gavis (1972) suggested that As(III) is stable and mobile only in a narrow range of Eh and pH conditions. Conditions must be reducing enough to produce dissolved As(III) but not so reducing as to produce sulfide, which could precipitate As(III). Under conditions where sulfide is formed, realgar (AsS) and orpiment (As2S3) occur as stable solids. At low pH, HAsS2(aq) is the
Table V Selected References on Chemical Speciation of Arsenic in Various Media Speciation techniquea
Acid mine drainage
LC-ICP-MS
Coal fly ash Drinking water (Natural water)
IC-ICP-MS –
Geothermal waters Groundwater
HPLC/GFAS and HPLC/HGAAS LC-ICP-MS
Groundwater
FI-HG-AAS
Groundwater
FI-HGAAS
Groundwater close to cattle tick dip sites Human urine (Bangladesh) Mine tailings
ICP-AES
Mine tailings Mung bean seedlings
XANES and EXAFS LC-ICP-MS
Mushroom (edible)
HPLC-ICP-MS
Plant—Chinese brake (Pteris vittata L.)
HPLC - AFS
IC-FI-HG-AAS AAS
Fraction/concentration As(III) ¼ 13,000 g liter1 As(V) ¼ 3700 g liter1 As(V) >> As(III) Particulate and soluble As contributed 11.4 and 88.6% of the total As, respectively. In the case of soluble As, As(III) and As(V) were 47.3 and 52.7%, respectively Na2HAsO4 was predominant 1
As(III) ¼ 720 g liter As(V) ¼ 1080 g liter1 As(III) and As(V) were present in 1:1 ratio As(III) was present at about 50% of the total As As(V) was dominant
Bednar et al. (2002) Jackson and Miller (1998) Thirunavukkarasu et al. (2001)
Buyuktuncel et al. (1997) Bednar et al. (2002) Samanta et al. (1999) Chatterjee et al. (1995) Kimber et al. (2002) Alauddin et al. (2003) Kim et al. (2002) Foster et al. (1997) Van den Broeck et al. (1998) Larsen et al. (1998)
Tu et al. (2003)
17
As (III) was the major species Total As ¼ 62350 mg kg1 63–99% as As(V) As(V) was dominant Roots: As(III) > As(V) Leaves: As(V) >> As(III) DMA 68–74% Methylarsonic acid 0.3–2.9% Trimethylarsine oxide 0.6–2.0% Arsenic acid 0.1–6.1% 94% of As in fronds was primarily as As(III)
Reference
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
Environment
(continued )
18
Table V (continued) Environment
Speciation techniquea
IC-ICP-MS
Poultry wastes
IC-ICP-MS
Rice grain
IC-ICP-MS
Rice straw River waters
Sewage sludge
HPLC-ICP-MS HGAAS using Na-tetrahydro borate(III) reductant HG-CT-AAS
Soil (contaminated)
HPLC-ICP-MS
Soil (contaminated)
Soil (contaminated) Well waters
Extraction with 1 M phosphoric acid plus 0.1 M ascorbic acid and measurement in LC-UV-HG-ICP/MS XAFS AAS
Wetlands
XANES
Dissolved As mostly as As(V) 130 g liter1 Organoarsenic compounds (Roxarsone) was dominant with trace levels of DMA and As(V) Total As 0.11–0.34 mg kg1 Inorganic As 11–91% remaining DMA As(V) > As(III) As(V) was the principal species
Reference
Gault et al. (2003) Jackson and Bertsch (2001)
Heitkemper et al. (2001) Abedin et al. (2002) Quinaia and Rollember (2001)
At pH 5.0 inorganic-As > organic-As At pH 6.5 organic-As > inorganic-As Total As ¼ 10000 mg kg1 As(V) ¼ >90% As(V) was the major species
Carbonell-Barrachina et al. (2000)
Mg3(AsO4)2 8H2O 670 g liter1 total dissolved arsenic; As(III) was dominant: As(III)/As(V) ratio ¼ 2.6 As(III) > As(V)
Foster et al. (1997) Chen et al. (1994)
Matera et al. (2003) Garcia-Manyes et al. (2002)
La Force et al. (2000)
a HPLC, high-performance liquid chromatography; ICP, inductively coupled plasmanalysis; MS, mass spectroscopy; LC, liquid chromatography; HG, hydride generation; XAFS, X-ray absorption fine structure spectroscopy; FI, flow injection; AAS, atomic absorption spectrometry; GFAAS, graphite furnace atomic absorption spectrometry; XANES, X-ray absorption near edge structure (XANES) spectroscopy; EXAFS, extended X-ray absorption fine-structure spectroscopy; AFS, atomic florescence spectrometry; CT, cold trapping.
S. MAHIMAIRAJA ET AL.
Polluted urban watercourse
Fraction concentration
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
19
predominant species if sulfide is present, whereas AsS 2 species predominate at pH greater than 3.7. Studying a stratified lake, Seyler and Martin (1989) showed that Mn, which has a higher redox potential than Fe and As, was reduced before the complete depletion of dissolved oxygen, and any dissolved As was present predominantly in the form of As(V). As conditions became more reducing, there was a rapid and concomitant increase of Fe and As and a reversal of As speciation such that as As(III) became more dominant, As2S3 and As concentrations correspondingly decreased. In groundwater, As is predominantly present as As(III) and As(V). The major As species in freshwater are As(III) and As(V), and small amounts of MMA, DMA, and methylated As(III) have also been detected. In seawater, As speciation differs in the surface and deep zones, with As(V) and As(III) species dominating the respective zone. In addition to the aforementioned species, Watt and Le (2003) noticed that an array of uncharacterized As species also appeared to constitute a significant portion of the total As present in water. The identification of these compounds is necessary to fully understand the As biogeochemistry in water.
IV. BIOGEOCHEMISTRY OF ARSENIC IN THE ENVIRONMENT The biogeochemistry and dynamics of As and other metal(loid)s vary between soil and aquatic environments. In the case of soil environment, a substantial proportion of the metal(loid)s is associated with the solid phase and their fate is strongly influenced by physicochemical interactions (e.g., adsorption–desorption) with the solid phase. Whereas in the case of aquatic environment, depending on the sediment content, a substantial proportion of metal(loid)s remains in solution and their fate is controlled largely by biological transformation.
A. BIOGEOCHEMISTRY OF ARSENIC IN THE SOIL Smith et al. (1998) presented a comprehensive review on the biogeochemistry of As in the soil environment. Here we include a brief discussion on various biogeochemical reactions of As in soil, which is helpful in understanding its behavior and in developing remediation strategies. As already discussed, As can exist in soil in different oxidation states but mostly as inorganic species, As(V) or As(III) (Adriano, 2001; Masscheleyn et al., 1991). In addition to inorganic species, microbial methylation of As in soil results in the release of organic methylarsenic compounds, such as MMA and
20
S. MAHIMAIRAJA ET AL.
Figure 2 Arsenic dynamics in contaminated soil and aquatic ecosystems.
DMA, and ultimately arsine gas (Smith et al., 1998; Vaughan, 1993). Both inorganic and organic species of As undergo various biological and chemical transformations in soils, including adsorption, desorption, precipitation, complexation, volatilization, and methylation (Fig. 2). Some important biogeochemical reactions of As and their significance in soil and aquatic environments are given in Table VI. The most thermodynamically stable 2 species of As(III) (i.e., H3AsO3 and H2 AsO 4 ) and As(V) (i.e., HAsO4 ) occur over the normal soil pH range of 4 to 8.
1. Adsorption and Surface Complexation The adsorption and retention of As by soils determine its persistence, reactions, movement, transformation, and ecological effects (toxicity). As in the case of most other metal(loid)s and nonmetals, one of the most
Table VI Some Important Biochemical Reactions of Arsenic and their Environmental Significance Process
Oxidation
Eq. No.
2 þ AsO3 4 þ H ¼ HAsO4 ðlog Ka ¼ 11:60Þ þ AsO3 4 þ 2H ¼ H2 AsO4 ðlog Ka ¼ 18:35Þ þ AsO3 4 þ 3H ¼ H3 AsO4 ðlog Ka¼ 20:60Þ 2 þ AsO3 3 þ H ¼ HAsO3 ðlog Ka ¼ 13:41Þ þ AsO3 þ 2H ¼ H AsO 2 3 3 ðlog Ka ¼ 25:52Þ þ AsO3 3 þ 3H ¼ H3 AsO3 ðlog Ka ¼ 34:74Þ Chemical 2 2HFeðVIÞO 4 þ 3H3 AsðIIIÞO3 ! 2FeðIIIÞ þ 3HAsðVÞO4 2þ þ þ HAsO þ 2H O ! 2Fe þ H AsO þ 2H 2Feþ 2 2 3 4 3 þ H3 AsOo3 þ OH þ O2 ðgÞ ! H2 AsO 4 þ O2 þ 2H 3 þ MnO2 þ 2H þ AsO3 ! Mn2þ þ AsO3 þ H2 O 4 MnO2 þ HAsO2 þ 2Hþ ! Mn2þ þ H3 AsO4
Microbial 2þ Fe2 O3 þ 4Hþ þ AsO3 þ AsO3 þ 2H2 O 3 ! 2Fe 4
ðCH3 Þ2 AsH ! ðCH3 Þ2 AsOðOHÞ ðCH3 Þ3 As ! ðCH3 Þ2 AsOðOHÞ Reduction
Significance
Reference
3 4 5 6 7 8
As(V), a less toxic As species, can exist in solution as H3AsO4, 3 3 H2AsO As(III), 4 , HAsO4 , and AsO4 a highly toxic As species, exists at natural pH values as H3AsO3, and H2AsO 3
Wilkie and Hering (1996)
9 10 11 12 13
As(III) is more toxic and mobile and hence it is desirable to oxidize to As(V), which is less toxic and relatively immobile. Chemical oxidation of As(III) may occur via Fe, or H2O2, or MnO2(VI) and Fe(VI) and is found very effective in the removal of As from water
Kocar and Inskeep (2003); Lee et al. (2003); Oscarson et al. (1981)
14
Competition of Fe(III) as a terminal electron acceptor in microbial respiration results in the oxidation of As(III) Arsine (di- and trimethyl) compounds can be oxidized by bacteria and fungi in the methylation process In waters reduction of As(V) to As(III) is possible at low pH and pE Reduction of As(V) to As(III) is possible in the presence of Fe even at a pE value of 0.5 at pH 7, while at pH 8 such reduction is not possible unless pE is <1.5 The formation of sulfides in reducing environment facilitates the reduction of As(V) to As(III) with the latter species dominating in the porewater
Masscheleyn et al. (1991)
15 16 17
2þ þ 5H2 O ! H3 AsO3 þ 2FeðOHÞ3 ðs Þ þ 3Hþ H2 AsO 4 þ 2Fe 2þ HAsO2 þ 5H2 O ! H3 AsO3 þ 2FeðOHÞ3 ðs Þ þ 2Hþ 4 þ 2Fe
18 19
þ H2 AsO 4 þ 3H þ 2e ! H3 AsO3 þ H2 O 2H3 AsO3 þ 6Hþ þ 3S 2 ! As2 S3 þ 6H2 O 2As2 S3 þ 4e ! 4AsS þ 2S2
20 21 22
Bose and Sharma (2002) Bose and Sharma (2002)
Moore et al. (1988)
21
3 þ AsO3 4 þ 2H þ 2e ! AsO3 þ H2 OðlogK ¼ 5:293Þ
O’Neill (1995)
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
Acid–base reactions
Reactions
(continued )
22
Table VI (continued ) Process
Methylation
Reactions
Significance
ðCH3 Þ2 AsOðOHÞ ! ðCH3 Þ2 AsH
23
H3 AsO4 ! H3 AsO3 ! ðCH3 ÞAsOðOHÞ ! ðCH3 Þ AsOðOHÞ 2 2 ðCH3 Þ2 AsOðOHÞ ! ðCH3 Þ AsH 2 ðCH3 Þ2 AsOðOHÞ ! ðCH3 Þ As
24 25 26
þ FeOHþ AsO3 4 þ 3H ! FeH2 AsO4 þ H2 O þ FeOHþ AsO3 4 þ 2H ! FeHAsO4 þ H2 O 3 FeOHþ AsO3 4 ! FeOHAsO4 þ FeOHþ AsO3 3 þ 3H ! FeH2 AsO3 þ H2 O
27 28 29 30
3 AlOHþ AsO3 4 ! AlOHAsO4 2 þ þ H ! AlAsO AlOHþ AsO3 4 4 þ H2 O þ AlOHþ AsO3 4 þ 2H ! AlHAsO4 þ H2 O
31 32 33
Dimethylarsinic acid can be reduced by bacteria to dimethyl arsine Biochemical transformations are mediated by microorganisms in terrestrial and aquatic environments. Biomethylation of inorganic and organic As is considered a major detoxification process Arsenic removal from water and wastewater is governed by sorption processes. Hydrous ferric oxide (FeOH) is an important sorbent in natural and engineered aquatic systems. Adsorption of As(III) increases with decreasing As/Fe ratios. As(V) adsorption is higher at high pH Natural Boehmite (monohydrates of trivalent aluminium oxide) is found to adsorb large amounts of As(V) and thus is suitable for As removal from water and wastewaters A molar ratios (FeAs) of 4, and an optimum pH of 5 at 33 C achieved less residual As in solution Liming results in the precipitation of As as calcium arsenate, which is unstable in aqueous environment and becomes insoluble As(V) can be immobilized through coprecipitation with hydrous Fe oxide or hydrous Mn oxide
3
Adsorption
Precipitation Fe2 ðSO4 Þ3 þ 2H3 AsO4 ! 2FeAsO4 þ 3H2 SO4
34
H3 AsO4 þ CaðOHÞ2 ! CaHAsO4 :2H2 O
35
FeðOHÞ3 þ H3 AsO4 ! FeAsO4 :2H2 Oþ H2 O þ 3MnOOHþ 2HAsO2 4 þ 7H þ 3e ! Mn3 ðAsO4 Þ2 þ 6H2 O
36 37
Reference O’Neill (1995) O’Neill (1995)
Wilkie and Hering (1996)
Dousova et al. (2003)
Papassiopi et al. (1996)
Stefanakis and Kontopoulos (1988) Masscheleyn et al. (1991)
S. MAHIMAIRAJA ET AL.
Eq. No.
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
23
commonly reported, and perhaps the first reaction to occur in soils, is As adsorption onto soil particles. Numerous studies have dealt with As sorption on to specific minerals and uncontaminated soils. Ferrous oxides/ hydroxides are involved most commonly in the adsorption of As in both acidic and alkaline soils. Carbonate minerals adsorb As in calcareous soils. In acidic soils, Mn oxides and biogenic particles play a dominant role in the adsorption of As (Arai et al., 2003; Oscarson et al., 1981). Arsenic is known to have high affinity for oxide surfaces, and several biogeochemical factors are found to play a major role in adsorption. Soil particle size, organic matter, type and nature of constituent minerals, pH, redox potential, and competing ions have all been shown to influence As adsorption (Chiu and Hering, 2000; Jones et al., 2000; Smith et al., 1998). In general, adsorption of As(V) decreases with increasing pH. In contrast, adsorption of As(III) increases with increasing pH. The effect of pH on As adsorption varies considerably among soils and is dependent on the nature of mineral surface. In soils containing low oxidic minerals, increasing the pH has little effect on the amount of As(V) adsorbed, whereas in highly oxidic soils, adsorption of As(V) decreases with increasing pH (Smith et al., 1998). This decrease is attributed to two interacting factors: (i) the increasing negative surface potential on the plane of adsorption and (ii) the increasing concentration of negatively charged As(V) species present in the soil solution. Brookins (1988) observed that amorphous Al and Fe hydroxides adsorbed more As(V) than As(III). The surface charge properties of variable charge soil components are strongly influenced by pH. At acid pH these soil components contain large amounts of positive charges, and adsorption of As (V) may become important. Arsenate ions are attracted to positively charged colloidal surfaces either at broken clay lattice edges where charged Al3þ groups are exposed or on surfaces of Fe and Al oxides and hydroxide films. Many researchers have investigated As(III) and As(V) adsorption reactions and surface speciation on major soil minerals (i.e., metal oxyhydroxides and phyllosilicate minerals) using various macroscopic and spectroscopic techniques. Arsenate is strongly adsorbed at acidic pH values on amorphous Al (OH)3, a-Al2O3, ferrihydrite, and hematite (Arai et al., 2001; Raven et al., 1998; Xu et al., 1988). Several spectroscopic [e.g., extended X-ray adsorption fine structure spectroscopy (EXAFS)], macroscopic [e.g., electrophoretic mobility (EM)], and thermodynamic modeling (e.g., surface complexation model) have revealed innersphere bidentate binuclear and/or monodentate As(V) complexes on ferrihydrite, goethite, amorphous Fe and Al oxides, and the bayerite polymorph (Arai et al., 2001; Fendorf et al., 1997) and on both inner sphere and outer sphere As(III) complexes on Al oxides (Arai et al., 2001; Goldberg and Johnston, 2001). In general, As(V) sorption on amorphous Al and Fe oxides is characterized by an apparent sorption maximum
24
S. MAHIMAIRAJA ET AL.
at pH 4, whereas As(III) sorption maximum occurs in the pH range of 7 to 8.5. The type and quantity of silicate clay minerals present in soil also influence the retention of As. Soils having higher clay content retain more As than sandy soils with low clay content. The degree of As sorption onto silicate clay minerals decreases in the order of kaolinite > vermiculite > montmorillonite (Goldberg and Glaubig, 1988; Manning and Goldberg, 1997). The silicate clay minerals also generally adsorb more As(V) than As (III), and adsorption by clay minerals is affected by pH (Lin and Puls, 2000). Arsenic and P belong to the same chemical group and both have comparable dissociation constants for their acids and solubility products for their salts. Therefore, H2AsO 4 and H2PO4 ions compete for the same sorption sites in soils, although some sites are preferentially available for the sorption of either H2PO 4 or H2AsO4 ions. A number of studies have shown that among the competing anions, the H2PO 4 suppresses As(V) sorption by soil more significantly than chloride (Cl), nitrate (NO 3 ), and sulfate (SO2 ) (Matera and LeHecho, 2001; Manful et al., 1989; O’Neill, 4 1995; Thanabalasingam and Pickering, 1986). Soil organic matter content also affects the adsorption of As and thus its bioavailability as organic molecules compete with As for sorption to surface sites. Thanabalasingam and Pickering (1986) showed that the maximum adsorption of As(V) on humic acids occurred around pH 5.5, whereas adsorption of As(III) increased up to pH 8. At high pH, the solubilization of humic substances reduces As retention. While there is very little information available on the effects of organic matter on As adsorption, Grafe et al. (2001) have shown that humic acid reduces both As(V) and As(III) adsorption on geothite between pH 3 and 9. Several functional groups present on these complex organic polymers may be responsible for binding As. Further, dissolved organic carbon substances are capable of increasing the mobility and bioavailability of As in soil and water ecosystems through redox reactions and soluble complex formation. Depending on various factors affecting the adsorption of As, part of the As adsorbed onto soil constituents is desorbed and released into the soil solution. Soil pH and phosphate addition are the most important factors that control the desorption of As. For example, Woolson et al. (1973) observed that phosphate addition to an As-contaminated soil displaced about 77% of the total As in the soil. Although phosphate addition increases As solubility, Peryea (1991) reported that desorption of As was dependent on the soil type, as no increase in As concentration in soil solution from a volcanic soil (with high anion-fixing and pH-buffering capacity) was observed. This suggests that only large additions of P (>400 mg kg1) would affect the As solubility in these soils (Chen et al., 2002; Smith et al., 1998). In long-term poultry litter-amended agricultural soils, Arai et al. (2003)
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
25
observed that the extent of As desorption from the litter increased with increasing pH from 4.5 to 7, but only 15% of the total As was released at pH 7, indicating the presence of insoluble phases and/or strongly retained soluble compounds. Elkhatib et al. (1984) suggested that the sorption of As (III) is not reversible in soil. One of the important factors affecting the adsorption/desorption characteristics of As is the contact time (residence time) in soils and sediments. For example, Arai and Sparks (2002) reported that the longer the residence time (1 year), the greater the decrease in As(V) desorption at pH 4.5 and 7.8, suggesting nonsingular reactions. The surface transformation processes, such as rearrangement of surface complexes and conversion of surface complexes into aluminum arsenate-like precipitates, might be responsible for the decrease in As(V) reversibility with aging. Thus, the fate and transport of the contaminants must be predicted/modeled not only on short-term adsorption and desorption studies, but also on long-term reactions. Although the desorption process is important in relation to the bioavailability and mobility of As, only a few studies have focused on desorption of As from soil constituents. Further studies on desorption are needed to fully understand the chemistry of As in soils, which might help in developing appropriate remediation technologies.
2.
Redox Reactions
In soil and aquatic environments, redox reactions not only determine the nature of chemical species, but also the solubility and mobility of As and thus its environmental significance. Arsenic in soils is subject to both abiotic and biotic redox reactions [Eqs. (9–23) in Table VI]. The Fe(III) oxides, Mn (III) oxides, and organic compounds in soils play a major role in catalyzing the abiotic oxidation of As(III) through an electron transfer mechanism (Adriano, 2001; Oscarson et al., 1981). Similarly, abiotic redox reactions are also responsible for the release of As from arsenopyrite through oxidation by Fe(III), considered to be a predominating process inducing the release of As into the groundwater in areas where well waters are highly contaminated with As [Eq. (1)]. Under moderately reducing conditions, As(III) is often found to be the predominant species in soil solution (Marin et al., 1993; Masscheleyn et al., 1991; Onken and Hossner, 1995). Studies by Deuel and Swoboda (1972) showed that there was an increase of As(III) in soil solution over time under flooded conditions. This was attributed to the release of As(V) during reductive dissolution of Fe oxyhydroxide minerals that have a strong affinity for As(V) and the subsequent reduction of As(V) to As(III).
26
S. MAHIMAIRAJA ET AL.
Biotransformation of As, involving the oxidation of As(III) to As(V) and the reduction of As(V) to As(III) by a variety of microorganisms, may occur in contaminated soil. For example, Alcaligenes faecalis was found to oxidize As(III) to As(V) (Osborne and Ehrlich, 1976; Phillips and Taylor, 1976). Bacteria, fungi, and algae are also able to reduce As(V) to As(III) and subsequently to arsine (Frankenberger and Losi, 1995). However, the effect of microbial activity on the transformation and movement of As in soil is difficult to quantify (Smith et al., 1998).
3.
Biomethylation
Arsenic in soil is also subject to biological transformation resulting in the formation of organo-arsenicals and other compounds [Eqs. (24 –26) in Table VI]. Inorganic As can undergo microbially mediated biochemical transformation, i.e., the hydroxyl group of arsenic acid [AsO(OH)3] is replaced by the CH3 group to form MMA, DMA, and TMA (Maeda, 1994). The pathway of As(V) methylation initially involves the reduction of As(V) to As(III), with the subsequent methylation of As(III) to dimethylarsine by coenzyme S-adenosylmethionine (Frankenberger and Losi, 1995). Methylation is often enhanced by sulfate-reducing bacteria. In addition to bacteria, several fungal species also have shown their ability to reduce As. Inorganic As is incorporated by autotrophic organisms such as algae and is then transported through the food chain. Arsenic becomes progressively methylated during this transfer. Therefore, methylation of As is considered a major detoxifying processes for these microorganisms (Adriano, 2001). The methylated As species is also subject to volatilization and photochemical reactions that may eliminate As from soil. Demethylation of methylarsenicals can occur under both aerobic and anaerobic conditions. Anaerobic demethylation reactions may result in the formation of toxic and reactive AsH3 from less toxic DMA, whereas aerobic demethylation of DMA is likely to yield As(V), thereby retaining As in the system. Although AsH3 undergoes rapid chemical oxidation under oxic conditions, it can exist for long periods in an aerobic environment. Because the demethylation process often produces CO2 in addition to CH4, it is preceded by oxidative assimilatory pathways used in substrate metabolism rather than by dissimilatory lyses. Methylation, demethylation, and reduction reactions are also important in controlling the mobilization and subsequent distribution of arsenicals in soils. These transformations are promoted by microbes; however, it is still not clear if in situ biomethylation is a common phenomenon. Although the presence of organic forms of As in soil can be associated with the application of anthropogenic compounds, such as fertilizers and pesticides (O’Neill,
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
27
1995), their presence is often linked to biomethylation. However, biomethylation reactions occur readily in aquatic environment and these reactions are discussed in Section IV.B.
4.
Leaching
Due to its strong adsorption onto organic and clay colloids, As(V) is likely to persist in soils for a long time, especially in fine-textured soils with high Fe contents (Woolson, 1983). In these soils, leaching of As(V) is low and therefore As contamination of groundwater is considered unlikely (Woolson, 1983). However, under certain environmental conditions (i.e., low pH and low Eh), As would leach in the soil profile, thereby contaminating the surface and groundwaters (Hingston et al., 2001; Ruokolainen et al., 2000). Considerable amounts of solubilized As could move downward in the soil profile with leaching water, especially in coarse-textured soils. It is for this reason that abandoned wood preservative (CCA) sites may threaten groundwater quality. For example, in examining the leaching of Cu, Cr, and As from CCA solution through free-draining, coarse-textured surface and subsurface soils using undisturbed soil lysimeters, McLaren et al. (1994) observed that the cumulative amounts of As leached ranged from 4 to 30% of the total As applied. Arsenic is present as a simple salt (soluble Na2HAsO4) in CCA, which is liable for leaching losses, especially in coarse-textured soils. Whereas when As is present as an organically complexed form (e.g., in sewage sludge), it is not readily leached in soils (McLaren et al., 1994). Again the role of H2PO 4 ions in enhancing the mobility of As, especially AsO2 ions, should be noted. For example, Qafoku et al. (1999) noticed 4 that the leaching of As in a column containing mineral soil incorporated with As-rich poultry manure increased with the addition of a phosphate compound. The arsenic concentration in the leachate was approximately 10 times higher when Ca(H2PO4)2 was used to leach the soil column as compared to the CaSO4 solution. In the presence of the Ca(H2PO4)2 solution, a maximum As concentration of 800 g liter1 was found in the leachate, much higher than the WHO maximum permissible limit of 10 g liter1 for drinking water.
B. BIOGEOCHEMISTRY OF ARSENIC IN AQUATIC ENVIRONMENTS As in the case of soil systems, the environmental and ecological significance of As dynamics in aquatic ecosystem is largely determined by its biogeochemical reactions, which are discussed in this section.
28
S. MAHIMAIRAJA ET AL.
1.
Adsorption and Desorption
Arsenic is stable in four oxidation states (+5, +3, 0, 3) under the Eh conditions that occur in aquatic systems. At high Eh values (mostly exist in 2 oxygenated waters), arsenic acid species (i.e., H3AsO4, H2AsO 4 , HAsO4 , 3 and AsO4 ) are stable. At mildly reducing conditions, arsenious acid species 2 (i.e., H3AsO3, H2AsO 3 , and HAsO3 ) become stable (Korte and Fernando, 1991; Penrose, 1974; Smith, 1986). The speciation of As in aquatic environment is critical in controlling the adsorption/desorption reactions with sediments. Adsorption to sediment particles may remove As(V) from contaminated water, as well as inhibiting the precipitation of As minerals such as scorodite (FeAsO4 2H2O) that control the equilibrium aqueous concentration (Foster et al., 1997). Under the aerobic and acidic to near-neutral conditions (typical of many aquatic environments), As(V) is adsorbed very strongly by oxide minerals in sediments. The highly nonlinear nature of the adsorption isotherm for As(V) in oxide minerals ensures that the amount of As adsorbed is relatively large, even when dissolved aqueous concentrations of As are low. Such adsorption occurring in natural environments protects water bodies from widespread As toxicity problems. Adsorption of As species by sediments are as follows: As(V) > As(III) > As (II) > DMA (Smedley and Kinniburgh, 2002). In As-contaminated sediments, Clement and Faust (1981) found that a significant portion of the As was bound in organo-complex forms and indicated that adsorption–desorption equilibrium must be considered as well as the redox effects in examining the dynamics of As in aquatic environment. As pH increases, especially above pH 8.5, As desorbs from the oxide surfaces, thereby increasing the concentration of As in solution. Desorption of As from As-contaminated sediments at high pH is the most likely mechanism for the development of groundwater As problems under the oxidizing conditions (Robertson, 1989; Smedley et al., 2002). These adsorption and desorption reactions of As in the aquatic environment have not been studied in detail under varied ecological conditions and therefore require greater attention.
2. Biotransformation Arsenic undergoes a series of biological transformations in the aquatic environment, yielding a large number of compounds, especially organoarsenicals. Certain reactions, such as oxidation of As(III) to As(V), may occur both in the presence and in the absence of microorganisms, whereas other reactions, such as methylation, are not thermodynamically favorable in water and can occur only in the presence of organisms. In neutral
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
29
oxygenated waters, As(V) is the thermodynamically favored form, whereas As(III) is stable under reducing conditions (Ferguson and Gavis, 1972). Some bacteria and marine phytoplankton are capable of reducing As(V) to As(III) or oxidizing As(III) to As(V) (Andreae, 1977). Biological reduction of As(V) to As(III) reportedly occurs most easily at a pH between 6 and 6.7 (Korte and Fernando, 1991). For example, Aggett and Aspell (1980) noticed that As was usually found as As(V) in the Waikato River of New Zealand, but during the spring and summer months, As(III) was often found to predominate. The reduction of As(V) to As(III) has been attributed to biological components of the river ecosystem. This biotransformation has been reported to occur in various aquatic systems, mediated by bacteria (Johnson, 1972; Myers et al., 1973) and algae (Andreae and Klumpp, 1979; Sanders 1983; Sanders and Windom, 1980). A cyanobacteria (Anabaena oscillaroides)–bacteria assemblage was also found to reduce As(V) to As (III) (Freeman, 1985). Benthic microbes are capable of methylating As under both aerobic and anaerobic conditions to produce methylarsines and methyl-arsenic compounds with a generic formula (CH3)nAs(O)(OH)3n where n may be 1, 2, or 3. MMA and DMA are the common organoarsenicals in river water. Methylated As species could result from direct excretion by algae or microbes or from degradation of the excreted arsenicals or more complex cellular organoarsenicals. Methylation may play a significant role in the mobilization of As by releasing it from the sediments to aqueous environment. The presence of organoarsenicals in river sediments is evidence that methylation occurs in the sediments (Anderson and Bruland, 1991). The rate of methylation/demethylation reactions and the consequent mobilization of arsenicals are affected by adsorption by sediments and soils. Primary producers such as algae take up As(V) from solution and reduce this to As(III) prior to methylation of the latter to produce MMA and DMA; the methylated derivatives are then excreted. This may be considered to represent a detoxification process in respect to the organism involved. Arsenic is taken up by algae due to its chemical similarity to phosphate. Although the detoxification of As by microorganism can be achieved through methylation, the element may be of significant toxicity to phytoplankton and periphyton communities in marine environments. Both macro- and microorganisms accumulate As in their tissues. Concentrations in organisms may be considerably higher than in the water in which they live, but unlike mercury (Hg), there is little, if any, concentration upward through the food chain (i.e., bioaugmentation). The toxicity of As to aquatic organisms is similar to its effects on terrestrial life, i.e., As(V) is much less toxic than As(III) (Ferguson and Gavis, 1972). Arsenate can replace H2PO 4 uptake in phosphate-deficient waters and can then be accumulated by algae. In a study of As accumulation in the food
30
S. MAHIMAIRAJA ET AL.
chain, it has been reported that most of the As accumulated by algae was in a nonmethylated form, which was bound strongly to protein or polysaccharides in the algal cell (Maeda et al., 1990). Such transformation can be stimulated by adding nutrients. Microbial formation of volatile arsine or other volatile-reduced compounds may play a role in the discharge of As to the atmosphere. Arsenite can be reduced and methylated to DMA, which can be further methylated or reduced and may eventually volatilize (Korte and Fernando, 1991).
V.
BIOAVAILABILITY AND TOXICITY OF ARSENIC TO BIOTA
Arsenic is used as an additive in various metal alloys and in wood preservation. Its toxic properties are exploited in the formulation of arsenical herbicides and insecticides. To date, however, geogenic As is largely responsible for most human poisoning (Smith et al., 2000). Due to its environmental and human health impact, As toxicity has been researched and documented more extensively than any other metal(loid)s.
A. TOXICITY
TO
PLANTS
AND
MICROORGANISMS
Arsenic contamination of soil and water poses a serious threat to plants and animals. Plants and microorganisms are known to accumulate As in their tissues and exhibit a certain degree of tolerance. However, at high concentrations, As is toxic to nearly all forms of life. Some selected references on toxicity (risks) of As in microorganisms, higher plants, and animals are presented in Table VII. Biotoxicity is mostly determined by the nature and bioavailability of As species present in the contaminated habitat. An average toxicity threshold of 40 mg kg1 has been established for crop plants (Sheppard, 1992). At high concentrations, As in plants inhibits plant metabolic processes, such as photosynthesis through interference of the pentose–phosphate pathway, thereby inhibiting growth and often leading to death (Marques and Anderson, 1986; Tu and Ma, 2002). Arsenite penetrates the plant cuticle to a greater degree than As(V) and generally results in the loss of turgor (Adriano, 2001). Biomass production and yields of a variety of crops have been shown to reduce significantly at high concentrations of As in soils (CarbonellBarrachina et al., 1997). For example, significant yield reductions of barley (Hordeum vulgare L.) and ryegrass (Lolium perenne L.) have been reported
Table VII Potential Risks of Arsenic to Terrestrial Biota
Soil
360 50–100 70–100
Soil
Soil
0, 15, 20, 30, 50, and 100 as power station fly ash or disodium hydrogen arsenate 100
Seedling beds
1000 and 2000
Soil
0–280 kg As ha1 (fine sandy loam soil) 0–560 kg As ha1 (clay soil). NaAsO2 applied at rates up to 720 kg As ha1 0.01, 0.1, or 1.0 mM PbCl2 or Na2HAsO4 in 1% agar þ modified Arnon and Hoagland solution. 1.0–5.5 1.0–5.0
Soil Water & nutrient solutions Soilless culture Soilless culture
Effect Yield reduction in barley; plants showed symptoms of As toxicity and P deficiency Reduction in growth of vegetative and root system in tomatoes As contents in rice cultivars exceeded the WHO standard 50% yield reduction in wheat, barley, and oats. Sensitivity to As was in the order: oats > wheat > barley Decreased the height of the apple tree: 100% growth inhibition at above 100 mg kg1 Substantial growth reduction in white spruce seedlings Significant growth reduction in cotton and soyabean As toxicity persisted for four growing seasons in potatoes and peas Growth inhibition of pea seedlings at all concentrations. As resulted in more growth inhibition than Pb No phytotoxic effect on radish Organic arsenicals (MAA > DMA) more phytotoxic than inorganic As to turnip, accumulating above the threshold for As in food crops (1.0 mg kg1)
Reference Lambkin and Alloway (2003) Miteva (2002) Xie and Huang (1998) Toth and Hruskovicova (1977)
Benson (1976) Rosehart and Lee (1973) Deuel and Swoboda (1972)
Steevens et al. (1972) Paivoke (1979)
Carbonell-Barrachina et al. (1999a) Carbonell-Barrachina et al. (1999b)
31
(continued )
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
Concentrationa
Medium
32
Table VII Concentrationa
Medium 0–10
Water culture
0, 0.04, 0.4, 4.0, and 20
Green algae in culture medium
78.7 g liter1 As(III) 159.3 g liter1 As(V) 12.4 (MMA) 35.7 (DMA) >400 Up to 8000
Earthworms
0.5–50 M As
PDA (phenyldichloroarsine), As(III) and As(V) at varied concentrations mg kg1 or mg liter1 unless specified.
a
Effect
Reference
Significant yield reduction in tomato (no tissue chlorosis or necrosis was observed) Growth inhibition of mung bean above 2.2 g g1 of As in the dry mass Increasing As decreased plant dry weight in cabbage. Most As remained in the roots with only 10–25% transported to the tops, 2% entered the inner leaves Raising phosphate concentration in the medium increased As(V) toxicity to freshwater green alga Scenedesmus obliguus
Carbonell-Barrachina et al. (1997)
Caused total fatality to earthworms Tolerated by Lumbricus rubellus and Dendrodrilus rubidus tolerated Toxicity follows: PDA > As(III) > As(V) and 24 h LD50 values 189.5, 191.0, and 519.4 mol kg1, respectively
Van den Broeck et al. (1997) Hara et al. (1977)
Chen et al. (1994)
Yeates et al. (1994) Langdon et al. (1999) Li et al. (1994)
S. MAHIMAIRAJA ET AL.
Nutrient solution Growth medium
(continued )
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
33
with the application of only 50 mg As kg1 soil (Jiang and Singh, 1994). Plant uptake of As is greatly influenced by its species in soil. As has already been discussed, different species have different solubility and mobility, thereby differing in their bioavailability to plants. Marin et al. (1992) reported that the order of As availability to rice (Oryza sativa L.) is as follows: As(III) > MMA > As(V) > DMA. They observed that upon absorption, DMA is readily translocated to the plant shoot, whereas As (III), As(V), and MMA accumulate primarily in the roots. While the application of As(V) and DMA did not affect rice growth, both As(III) and MMA were found to be phytotoxic to rice. Burlo et al. (1999) noted that both MMA and DMA in tomato plants (Lycopersicon esculentum Mill.) had a greater upward translocation than As(III) and As(V). In general, the accumulation of As in the edible parts of most plants is low (O’Neill, 1995), which is attributed to a number of reasons, including (Wang et al., 2002) (i) low bioavailability of As in soil; (ii) restricted uptake by plant roots; (iii) limited translocation of As from roots to shoots; and (iv) phytotoxicity and subsequent premature plant death at relatively low As concentrations in plant tissues. Apart from chemical forms, it has been shown that the phytotoxicity of As varies with the soil conditions. For example, Reed and Sturgis (1963) reported that As inhibits rice plant growth more strongly under submerged soil conditions than under upland soil conditions, because As(V) is reduced to As(III), which is more soluble and more toxic to plants in submerged soil. Arsenic phytotoxicity is expected to be greater in sandy soils than in other soil types, as the former soils generally contain low amounts of Fe and Al oxides and silicate clays, which have been implicated in the adsorption of As from soil solution (Sheppard, 1992; Smith et al., 1998). The antagonistic and synergistic effects of various nutrient anions also determine the phytotoxicity of As to some extent. For example, Davenport and Peryea (1991) reported a reduction of As uptake by plants with the application of phosphate, which was attributed to H2PO 4 ion-induced inhibition of As(V) uptake by plant roots. In contrast, Woolson (1973) observed that a phosphate application increased As availability and As uptake by plants, which was attributed to the H2PO 4 ion-induced release of As(V) to the soil solution. Most plants do not accumulate enough As to be toxic to animals and humans. Growth reductions and crop failure are the main consequences of soil As contamination (Walsh and Keeney, 1975). Thus the major hazard for animal and human systems is derived from direct ingestion of As-contaminated soil or water (Smith et al., 1998). Arsenic contamination of soil and water has a direct impact on microbial community and structure. At high concentrations, a reduction in the soil microbial population has been reported by a number of researchers (Bisessar, 1982; Van Zwieten et al., 2003). In general, as in the case of higher plants, As(III) is more toxic to microorganisms than As(V) (Maliszewska
34
S. MAHIMAIRAJA ET AL.
et al., 1985). Hiroki (1993) has shown that As(III) is more toxic to bacteria and actinomycetes than As(V) and that fungi not only display a higher tolerance to As(III) than bacteria and actinomycetes, but also show the same tolerance to both As(V) and As(III). Arsenite also inhibits enzyme activities in soil (Tabatabai, 1977). However, many bacterial communities are found to adapt to As-contaminated environments by developing resistance and tolerance mechanisms (Smith et al., 1998). Earthworms usually have a high capacity for accumulating toxic elements; however, the extent of accumulation is dependent on the type of element and on soil properties (Ma, 1982). Earthworms are known to inhabit As-rich metalliferous soils (Langdon et al., 1999). They are likely to accumulate As present in soils through ingestion of solid-phase As and dermal contact with pore water As. Yeates et al. (1994) observed a complete elimination of earthworms in soils contaminated by As derived from timber preservatives at concentrations of 400 and 800 mg As kg1, but few earthworms at 100 mg As kg1. In contrast, Langdon et al. (1999) found populations of Lumbricus rubellus and Dendrodrilus rubidus resistant to As(V) and Cu present in mine spoil containing up to 8000 mg As kg1 and 750 mg Cu kg1. The difference in the threshold levels of As for earthworms between these two experiments may be attributed to the difference in the bioavailability of As, which is a function of speciation and substrate matrix. Earthworms generally show resistance to As toxicity; however, the mechanisms of such resistance are not fully understood (Langdon et al., 2003).
B. RISK
TO
ANIMALS
AND
HUMANS
Drinking water is the most important source of dietary intake of As by animals and humans (Fitz and Wenzel, 2002). However, food also forms a source of As exposure (Adriano, 2001). The occurrence of inorganic As in drinking water has been identified as a source of risk for human health even at relatively low concentrations. As a consequence, more stringent safer limits for As in drinking water have been proposed (Wenzel et al., 2001). Soluble As compounds are rapidly absorbed from the gastrointestinal tract (Hindmarsh and McCurdy, 1986). Several studies in humans indicate that both As(III) and As(V) are well absorbed across the gastrointestinal tract (USDHHS, 2000). Studies involving the measurement of As in fecal excretion in humans indicated that almost 95% of oral intake of As(III) is absorbed (Bettley and O’Shea, 1975). This was supported by studies in which urinary excretion in humans was found to account for 55–80% of daily intakes of As(III) or As(V) (Buchet et al., 1981; Crecelius, 1977; Mappes, 1977). It has also been reported that both MMA and DMA are also well absorbed (75–85%) across the gastrointestinal tract (Buchet et al., 1981).
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
35
Once absorbed, simultaneous partial oxidation of As(III) to As(V) and partial reduction of As(V) to As(III) occur, yielding a mixture of As(III) and As(V) in the blood. The As(III) may undergo enzymatic methylation primarily in the liver to form MMA and DMA, but the rate and relative proportion of methylation production vary among animal species. Most As is promptly excreted in the urine as a mixture of As(III), As(V), MMA, and DMA, and relatively smaller amounts are excreted in the feces. Some As may remain bound to tissues, depending on the rate and extent of methylation. Monomethylarsonic acid may be methylated to DMA, but neither MMA nor DMA is demethylated to yield As(III) or As(V). Arsenic may accumulate in skin, bone, and muscle and its half-life in humans is between 2 and 40 days (USDHHS, 2000). Teratogenic effects of As in chicks, golden hamsters, and mice have been reported. Arsenic does not appear to be mutagenic in bacterial and mammalian assays, although it can induce chromosomal breakage, chromosomal aberration, and chromatid exchange. Studies have shown that As may be an essential element at trace concentrations for several animals such as goats, rats, and poultry, but there is no evidence that it is essential for humans (USEPA, 1988). The acute toxicity of As compounds in humans is a function of their rate of removal from the body. Arsine is considered to be the most toxic form, followed by As(III), As(V) and organic As compounds (MMA and DMA). Lethal doses in humans range from 1.5 mg kg1 (diarsenic trioxide) to 500 mg kg1 of body weight (DMA). Acute As intoxication associated with the ingestion of contaminated well water has been reported in many countries (Table VIII). The single most characteristic effect of long-term exposure to As is a pattern of skin changes, including hyperkeratosis (a darkening of the skin and appearance of small “corns” or “warts” on the palms, soles, and torso; Fig. 3). A small number of the “corns” may ultimately develop into skin cancer (USDHHS, 2000). Early symptoms of As poisoning in humans include abdominal pain, vomiting, diarrhea, muscular pain, and weakness, with flushing of the skin (Armstrong et al., 1984; Cullen et al., 1995; Moore et al., 1994). These symptoms are often followed by numbness and tingling of the extremities, muscular cramping, and the appearance of an erythematous rash. Further symptoms may appear within a month, including burning paraesthesias of the extremities, hyper/hypopigmentation (mottled or multicolor skin), Mee’s lines on fingernails, and progressive deterioration in motor and sensory responses (Fennell and Stacy, 1981; Murphy et al., 1981). Acute oral As poisoning at doses of 8 mg As kg1 and above have been reported to affect the respiratory system (Civantos et al., 1995). A number of studies in humans have shown that As ingestion may lead to serious effects on the cardiovascular system (Cullen et al., 1995). Anemia and leukopenia
36
S. MAHIMAIRAJA ET AL. Table VIII Selected References on Effect of Arsenic on Human Health
Effect and/or symptoms Neoplasia and induce DNA damage and inhibit DNA hypermethylation Malanosis, melanokeratosis (malignancy) in adults Hyper pigmentation, keratosis, weakness, anemia, burning sensation of eyes, solid swelling of legs, liver fibrosis, chronic lung disease, gangrene of toes, neuropathy Chromosomal aberrations and chromatid exchanges Skin cancer
Bladder cancer Lung cancer Peripheral vascular, cardiovascular, cerebrovascular diseases Diabetes Adverse reproductive outcome Neuropathy
Countries
Reference
France USA
Burnichon et al. (2003) Goering et al. (1999)
Bangladesh and India
Saha (2003)
Bangladesh Bangladesh Bangladesh and India Bangladesh Bangladesh
Karim (2000) Mazumder (2003) Rahman et al. (2001) Kadono et al. (2002) Karim (2000)
India
Mahata et al. (2003)
Bangladesh India USA USA USA Bangladesh India USA USA USA USA
Mazumder (2003) Mukherjee et al. (2003) Brown and Ross (2002) Hamadeh et al. (2002) Hall (2002) Kadono et al. (2002) Das et al. (1996) Brown and Ross (2002) Brown and Ross (2002) Hall (2002) Brown and Ross (2002)
USA USA Bangladesh and India India India
Brown and Ross (2002) Brown and Ross (2002) Mazumder (2003) Mukherjee et al. (2003) Mukherjee et al. (2003)
India India
Mukherjee et al. (2003) Chattopadhyay et al. (2002) Yih et al. (2002) Hughes (2002) Hughes (2002)
Paresthesias and pains in the distal parts of extremities Dysfunction of sensory nerve Apoptosis and necrosis in developing brain cells Inducement of oxidative stress, activating stress gene expression Altered DNA methylation and cell proliferation Bone marrow depression Hypertension Gastrointestinal disturbances
USA India USA
Hepatocellular carcinoma
China
Taiwan USA USA
Hall (2002) Rahman et al. (1999) Cullen et al. (1995); Hall (2002) Liu et al. (2001) (continued )
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
37
Table VIII (continued ) Effect and/or symptoms Hepatic fibrosis Blackfoot disease
Acute intake results: vomiting, diarrhea, low blood pressure, and high heart beat Teratogenesis in unborn children
Countries
Reference
India Taiwan Bangladesh, China, India, Taiwan, and USA USA
Santra et al. (2000) Wang et al. (1997) Wai et al. (2003)
Bangladesh
Karim (2000)
Cullen et al. (1995)
Figure 3 Skin lesions (hyperkeratosis) at various stages due to arsenic poisoning.
were also found to be the common effects of As poisoning in humans resulting from prolonged oral exposure at doses of 0.05 mg As kg1 day1 or more (Armstrong et al., 1984; Mazumder et al., 1988; Saha et al., 2003).
38
S. MAHIMAIRAJA ET AL.
Studies have also revealed hepatic effects of As poisoning (USDHHS, 2000), as indicated by swollen and tender liver with elevated levels of hepatic enzymes in blood (Armstrong et al., 1984).
VI. RISK MANAGEMENT OF ARSENIC IN CONTAMINATED ENVIRONMENTS Risk management of contaminated sites includes source reduction, site remediation, and environmental protection. Selection of optimal risk management strategies requires consideration of core objectives such as technical practicability, feasibility, and cost effectiveness of the strategy and wider environmental, social, and economic impacts. Arriving at an optimal risk management solution for a specific contaminated site involves three main phases of the decision-making process. These include problem identification, development of problem solving alternatives (i.e., remediation technologies), and management of the site. The next section discusses the various remediation technologies considered suitable for managing As-contaminated soil and aquatic environments.
A. REMEDIATION
OF
ARSENIC-CONTAMINATED SOIL
Remediation of As-contaminated soil involves physical, chemical, and biological approaches that may achieve either the partial/complete removal of As from soil or the reduction of its bioavailability in order to minimize toxicity (Fig. 4). A large variety of methods have been developed to remediate metal(loid)s-contaminated sites. These methods can also be applicable for the remediation of As-contaminated soils. The selection and adoption of these technologies depend on the extent and nature of As contamination, type of soil, characteristics of the contaminated site, cost of operation, availability of materials, and relevant regulations.
1.
Physical Remediation
Major physical in situ treatment technologies to remediate metal(loid)contaminated sites include capping, soil mixing, soil washing, and solidification. The simplest technique for reducing the toxic concentration of As in soils is mixing the contaminated soil with uncontaminated soil. This results in the dilution of As to acceptable levels. This can be achieved by importing clean soil and mixing it with As-contaminated soil or redistributing clean
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
39
Figure 4 Viable remediation technologies for arsenic-contaminated soil/sediment and aquatic ecosystems.
materials already available in the contaminated site. Another dilution technique, especially in cultivated soils, relies on deep ploughing, during which the vertical mixing of the contaminated surface soil with less contaminated subsoil reduces the surface contamination, thereby minimizing the potential for As uptake by plants and ingestion of As by grazing animals. However, in this method the total concentration of As in soil will remain the same. Soil washing or extraction has also been used widely for the remediation of metal(loid)-contaminated soils in Europe (Tuin and Tels, 1991) and this method may be applicable for As-contaminated soils to some extent. Tokunaga and Hakuta (2002) evaluated an acid-washing process to extract the bulk of As(V) from a highly contaminated (2830 mg As kg1 soil) Kuroboku soil (Andosol) so as to minimize the risk of As to human health and the environment. The contaminated soil was washed with different concentrations of hydrogen fluoride, phosphoric acid, sulfuric acid, hydrogen chloride, nitric acid, perchloric acid, hydrogen bromide, acetic acid,
40
S. MAHIMAIRAJA ET AL.
hydrogen peroxide, 3:1 hydrogen chloride–nitric acid, or 2:1 nitric acid– perchloric acid. Phosphoric acid proved to be most promising as an extractant, attaining 99.9% As extraction at 9.4% acid concentration. Sulfuric acid also attained a high percentage extraction. The acid-washed soil was further stabilized by the addition of lanthanum (La), cerium (Ce), and Fe(III) salts or their oxides/hydroxides, which form an insoluble complex with dissolved As. Both salts and oxides of La and Ce were effective in immobilizing As in the soil attaining less than 0.01 mg liter1 As in the leachate. The success of soil washing largely depends on speciation of As present in the contaminated soils, as it is based on the desorption or dissolution of As from the soil inorganic and organic matrix during washing with acids and chelating agents. Although soil washing is suitable for off-site treatment of soil, it can also be used for on-site remediation using mobile equipment. However, the high cost of chelating agents and choice of extractant may restrict their usage to only small-scale operations. Arsenic-contaminated soil may be bound into a solid mass by using materials such as cement, gypsum, or asphalt. However, there are issues associated with the long-term stability of the solidified material. Capping the contaminated sites with clean soil is used to isolate contaminated sites as it is less expensive than other remedial options (Kookana and Naidu, 2000). Such covers should obviously prevent upward migration of contaminants through the capillary movement of soil water. The depth of such cover or “cap” required for contaminated sites should be assessed carefully. Using a simulated experiment, Kookana and Naidu (2000) demonstrated that when the water table is deeper than 2 m from the surface of cap, the upward migration of As through the cap is likely to be less than 0.5 m in 5 years. Where the water table is shallow enough to supply water to the surface (i.e., 1.5 to 2 m in most soils), dissolved As could take <10 years to reach the surface. They have also indicated that when the cap is of a different soil type than the underlying contaminated soil, a coarse-textured cap is very effective in reducing the capillary rise and therefore the cap should always be designed to include a coarser layer to break the capillary continuity.
2.
Chemical Remediation
Remediation, based on chemical reactions, is becoming increasingly popular largely because of a high rate of success. A number of methods have been developed mainly involving adsorption, immobilization, precipitation, and complexation reactions [Eqs. (27–37) in Table VI]. However, such methods are often expensive for the remediation of large areas. Two approaches are often used in the chemical remediation of metal(loid)contaminated soils: (i) immobilization of metal(loid)s using inorganic and
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
41
organic soil amendments in order to reduce their bioavailability and (ii) mobilization of metal(loid)s and their subsequent removal through plant uptake (phytoremediation) or soil washing. This section discusses the immobilization techniques used for the remediation of As-contaminated soil. The second approach is discussed in Section VI.A.3. Chemical immobilization is achieved mainly through adsorption/precipitation of As in contaminated sites through the addition of soil amendments. The mobilization of metal(loid)s in soils for plant uptake and leaching to groundwater can be minimized by reducing their bioavailability through chemical and biological immobilization (Bolan et al., 2004). There has been interest in the immobilization of metal(loid)s using a range of inorganic compounds such as lime, P fertilizers (e.g., phosphate rocks) and alkaline waste materials, and organic compounds such as biosolids (Basta et al., 2001; Knox et al., 2000). Depending on the source, the application of P compounds can cause direct adsorption of As onto these materials, promote As complex formation, or induce desorption of As through competition. This method is considered more economical and less disruptive than the conventional remediation option of soil removal (Bolan et al., 2003). Immobilization of As may be achieved by (i) changing the physical properties of the soil so that As is more tightly bound and therefore becomes less bioavailable; (ii) chemically immobilizing As either by sorption onto a mineral surface or by precipitation as a discrete insoluble compound; and/ or (iii) mixing the contaminated soil with uncontaminated soil, thereby increasing the number of As-binding sites (Naidu et al., 2003). A number of organic and inorganic amendments are known to immobilize a range of metal(loid)s including As by chemical adsorption. These include ion-exchange resin, ferrous sulfate, silica gel, gypsum, clay minerals such as bentonite, kaolin, and zeolite, green sand, and liming materials. These materials are naturally occurring and nontoxic with a large specific surface area and a significant amount of surface charge. The use of naturally occurring clay minerals such as zeolite as adsorbents is a novel method for the remediation of metal(loid)-contaminated soils (Minato et al., 2000). The advantages of zeolite application are its high efficiency for retention of metal(loid)s in soils, low cost, and easy application. Naidu et al. (2000) examined the potential for using strongly weathered oxidic soils as reactive barriers and found a strong affinity for As as it retains almost 5000 mg As kg1. Boisson et al. (1999) assessed the effectiveness of soil additives in reducing contaminant mobility. Their results indicated that the lowest amount of As was extracted when the soil was amended with beringite, steel shots, and their combination. Although the addition of hydroxyapatite decreased the mobility of metals such as Cd and Pb, it increased the mobility of As mainly due to H2PO4–AsO4 competition for the sorption sites. Therefore, the use
42
S. MAHIMAIRAJA ET AL.
of hydroxyapatite at multimetal(loid)-contaminated sites requires careful attention. Liming is increasingly being used as an important soil management practice in reducing the toxicity of certain metal(loid)s in soils. In addition to the traditional agricultural lime, a large number of studies have examined the potential value of other liming materials as immobilizing agents in reducing the bioavailability of a range of metal(loid)s in soils (Bolan et al., 2003). However, the effect of liming soils on As mobility has been rather inconsistent. Lime addition to As-contaminated soil induces the formation of CaH(AsO4)2 [Eq. (35) in Table VI], thereby reducing the soluble As in the soil solution for plant uptake and leaching. However, the solubility product of this compound is greater than that for Fe and Al arsenates, which are readily formed in most soils. For this reason, liming is not practiced widely to overcome As toxicity in soils (Jones et al., 1997), although liming has been reported to increase the immobilization of As (Bothe and Brown, 1999) and to decrease the plant uptake of As (Jiang and Singh, 1994; Tyler and Olsson, 2001). Naidu et al. (2003) evaluated the potential value of the chemical immobilization technique in the remediation of an As-contaminated site under field conditions in Australia. The site was a former railway depot that had previously been shown to be extensively contaminated with As. The As levels in the soil exceeded both ecological (20 mg kg1) and health investigation levels (100 mg kg1) and was appreciably water soluble, indicating that large amounts of As were potentially mobile at this site. The historical source of the contamination appears to be the ubiquitous use of As-based herbicides. Exposure pathway analyses showed that the highly mobile As posed a risk to both the groundwater and the residents living in the area. The contaminated site was identified for industrial development with Australian industrial guidelines for As set at 500 mg kg1 soil. Options for managing contaminated soil included in situ cleanup, excavation, and transport to landfill sites or application of risk-based land management strategy. Both in situ cleanup and excavation and transport to landfill were found to be prohibitively expensive and ranged from $500,000 to $1,000,000. A risk reduction strategy was adopted with the aim to reduce the mobility of As through chemical immobilization. Ferrous salt was used to generate in situ mineral phases to immobilize As [Eq. (38)]. This reaction requires oxygen to be available to the soil and also generates considerable amounts of acid, which may be counterproductive to As immobilization in poorly buffered soils. The increased acidity could be neutralized by the amendment with lime [Eq. (39)]. The redox conditions of the soil also influence the speciation of As, and an example of two possible redox couples is given later [Eqs. (40) and (41)]. Following initial detailed laboratory studies, a mixture of Fe/Mn/gypsum was used as the stabilizing chemical. As shown in
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
43
Fig. 5, application of the mixed chemical led to a significant decline in mobile As. Subsequent studies involving aging of the treated soil showed complete elimination of risk posed by As. The total cost using this strategy was < $100,000, thus providing significant savings to the client. þ 4FeSO4 þ O2ðgÞ þ 6H2 O ! 4FeOOHðsÞ þ 4SO2 4 þ 8H
1 2FeSO4 þ O2ðgÞ þ 2CaCO3 þ 5H2 O ! 2
ð38Þ
2FeOOHðsÞ þ2CaSO4 2H2 O þ 2CO2ðgÞ ð39Þ
þ
AsO3 3
! 2Fe2þ þ AsO3 4 þ 2H2 O
þ
AsO3 3
! Mn2þ þ AsO3 4 þ H2 O Eo ¼ 0:67 V
Fe2 O3ðsÞ þ 4H þ MnO2ðsÞ þ 2H þ
Eo ¼ 0:21 V
ð40Þ ð41Þ
Results of a field experiment conducted by Xie and Huang (1998) on an As-polluted soil (Typentiaqualf ) in China have shown that the application of Fe (as FeCl3 at 25 mg Fe kg1 soil) or Mn (as MnO2 at 25 mg Mn kg1 soil) markedly lowered the total water- soluble As [As(III) þ As(V)] (24–26%) and As(III) (17–82%) in the soil and made the rice plants grow better than the control treatment, resulting in a higher rice grain yield and lesser As content in rice husk. This was attributed to the oxidation of As(III) to As(V) by MnO2 and the subsequent strong adsorption of As(V) by Fe and Mn oxides.
Figure 5 Variation of water-extractable As (1:5) for a subsurface-contaminated soil with soil treatment and incubation temperature. Treatments were (0) control soil; (A) Fe; (B) Fe þ lime; (C) Fe þ Mn; and (D) Fe þ Mn þ Al (Naidu et al., 2003).
44
S. MAHIMAIRAJA ET AL.
3.
Biological Remediation
a. Bioremediation. Bioremediation of soils contaminated with organic compounds such as pesticides and hydrocarbons is widely accepted in which native or introduced microorganisms and/or biological materials, such as compost, animal manures, and plant residues, are used to detoxify or transform contaminants. There has been increasing interest in the application of this technology for the remediation of metal(loid)-contaminated soils, especially for those metal(loid)s that undergo biological transformation. Although it has several limitations, this technology holds continuing interest because of its cost effectiveness. The unique aspect in bioremediation is that it relies mainly on natural processes and does not necessarily require the addition of chemical amendments other than microbial cultures and biological wastes. Because As undergoes biological transformation in soil, appropriate microorganisms may be used for the remediation of Ascontaminated soils. Existing and developing in situ bioremediation technologies may be grouped into the following two broad categories (NRC, 1997). i. Intrinsic bioremediation is where the essential materials required to sustain microbial activity exist in sufficient concentrations that naturally occurring microbial communities are able to degrade the target contaminants without the need for human intervention. This technique is better suited for remediation of soils with low levels of As over an extensive area. ii. Engineered bioremediation relies on various approaches to accelerate in situ microbial degradation rates. This is accomplished by optimizing the environmental conditions by adding nutrients and/or an electron donor/acceptor, thus promoting the proliferation and activity of existing microbial consortia. It is favored for highly contaminated localized sites. Three approaches could be used in the bioremediation of As-contaminated soils: (i) As could be immobilized into microbial cells through biosorption (bioaccumulation), (ii) toxic As(III) could be oxidized to less toxic As(V), and (iii) As compounds could be removed from the soil by volatilization. i. Bioaccumulation: Microorganisms exhibit a strong ability to accumulate (bioaccumulation) As from a substrate containing very low concentrations of this element. Bioaccumulation is activated by two processes, namely biosorption of As by microbial biomass and its byproducts and physiological uptake of As by microorganisms through metabolically active and passive processes. Factors such as soil pH, moisture and aeration, temperature, concentration and speciation of As, soil amendments, and rhizosphere are known to influence the process of bioaccumulation of As in microbial cells. While a number of bacterial and
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
45
fungal species have been known to bioaccumulate As, some algal species (Fucus gardneri and Chlorella vulgaris) are also known to accumulate As (Granchinho et al., 2001; Maeda et al., 1985). This technique has often been used successfully to remove metal(loid) ions from the aquatic environment and is therefore discussed further in Section VI.B.2. ii. Microbial redox reactions: Heterotrophic bacteria have been found to oxidize toxic As(III) in soils and sediments to less toxic As(V) and thus could play an important role in the remediation of contaminated environment (Wakao et al., 1988). Because As(V) is strongly adsorbed onto inorganic soil components, microbial oxidation could result in the immobilization of As. Strains of Bacillus and Pseudomonas spp. (Frankenberger and Losi, 1995) and Alcaligenes faecalis (Phillips and Taylor, 1976) and Alcaligenes spp. (Osborne and Ehrlich, 1976) were found capable of oxidizing As(III) to As(V). A dissimilatory metal(loid) reduction has the potential to be a helpful mechanism for both intrinsic and engineered bioremediation of contaminated environments. Arsenic can be reduced to Aso, which is subsequently precipitated as a result of microbial sulfate reduction. Desulfototomaculum auripigmentum, which reduces both As(V) to As(III) and SO2 4 to H2S leads to As2S3 precipitation (Newman et al., 1997). Because arsenite is more soluble than As(V), the latter can be reduced to As(III) using bacteria in soil and subsequently leached. iii. Methylation of As: A variety of microbes could transform inorganic As into its metallic hydride or methylated forms. Due to their low boiling point and/or high vapor pressure, these compounds are susceptible for volatilization and could easily be lost to the atmosphere (Braman and Foreback, 1973). Methylation is considered a major biological transformation through which As is volatilized and lost. As discussed earlier, biomethylation of As in soils and aquatic systems is well documented, as it is important in controlling the mobilization and subsequent distribution of arsenicals in the environment (Frankenberger and Losi, 1995; Gao and Burae, 1997; McBride and Wolfe, 1971; Tamaki and Frankenberger, 1992). Methanogenic bacteria, commonly present in sewage sludge, freshwater sediments, and composts, are capable of methylating inorganic As to volatile DMA. Arsenate, As(III), and MAA can serve as substrates in DMA formation. Inorganic As methylation is coupled to the CH4 biosynthetic pathway and may be a widely occurring mechanism for As removal and detoxification (Frankenberger and Losi, 1995). In addition to bacteria, certain soil fungi also are able to volatilize As as methylarsine compounds, which are derived from inorganic and organic As species.
46
S. MAHIMAIRAJA ET AL.
Woolson (1977) demonstrated the release of alkylarsines in a number of soils. Dimethylarsine and trimethylarsine are produced when soils were amended with inorganic and methylated arsenic herbicides. The organisms responsible for volatilization of As originate from diverse environments, suggesting that a number of species have the capacity to produce alkylarsines (Frankenberger and Losi, 1995; Woolson, 1977). Some examples of the organisms involved in the biomethylation of As are given in Table IX. In most cases, these organisms were tested in laboratory conditions; however, their performance should be assessed under field conditions in contaminated sites. b. Phytoremediation. Phytoremediation is considered a subset of bioremediation that employs plants and their associated root-bound microbial community to remove, contain, degrade, or render environmental contaminants harmless (Raskin et al., 1997; Robinson et al., 2003b). This terminology applies to all plant-influenced biological, chemical, and physical processes that aid in the remediation of contaminated medium (Cunningham and Lee, 1995). It involves soil–plant systems in which metal(loid)s-accumulating plants are grown in contaminated sites. It is considered an economically feasible and environmentally viable technology for remediating metal(loid)-contaminated systems. The effectiveness of this technology is, however, variable and highly site dependent. In phytoremediation, plants are exploited as a biopump that use the energy of the sun to remove water and contaminants from the soil to the aboveground portion and return some of the products of photosynthesis back into the root zone in the form of root exudates involved in the (im) mobilization of contaminants. Transpiration is the driving force for phytoremediation. By removing water from the medium, plants help reduce erosion, runoff, and leaching, thereby limiting the movement of contaminants off-site. Some contaminants are taken up in the transpiration stream, where they may be metabolized, and may be eventually volatilized. By removing excess water from the soil profile, plant roots may also create an aerobic environment where metal(loid) mobility is reduced and biological activity is enhanced. Plants stimulate microbiological activity in the root zone by providing a carbon source from root exudates and decaying root materials (Robinson et al., 2003b). Phytoremediation technologies have been grouped into various categories that include phytostabilization, rhizofiltration, and phytoextraction (Cunningham et al., 1995). In phytostabilization, transpiration and root growth are used to immobilize contaminants, including As by reducing leaching, controlling erosion, creating an aerobic environment in the root zone, and adding organic matter to the substrate that binds As. It involves the establishment of metal(loid)-tolerant vegetation on the contaminated site
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT Table IX Microorganisms Proven Capable of Biomethylating Arsenic Compounds in Soil and Aquatic Environments Organisms
Mechanism
Scopulariopsis koningii Methylate As to Fomitopsis pinicola trimethylarsenic(V) Penicillium gladioli species, precursors to volatile trimethylarsine Fusarium oxysporum Accumulates As(V) meloni and converts to dimethylarsine Fucus gardneri Methylates As(V) to dimethylarsine Closterium aciculare Methylates As(V) to methylarsenic(III) species S. brevicaule Transforms As(V) to (CH3)3As species Chlorella vulgaris Biosorption and accumulation of As and converting into compound of (CH3)2AsO(OH) Polyphysa peniculus Methylates As(V) to dimethylarsine Penicillium sp. At pH 5 to 6 methylates CH3AsO(OH)2 and (CH3)2AsO(OH) to (CH3)3As. Aeromonas sp. Methylates (CH3)2AsO(OH) to (CH3)3AsO 3 Alcaligenes sp. Methylates AsO 2 or AsO4 Pseudomonas sp. into AsH3 under aerobic condition Flavobacterium sp. Methylates (CH3)2AsO(OH) to (CH3)3AsO Candida humicola Methylates As(V) into a volatile As species Methanobacteriaum Methylates As(V), As(III) and CH3AsO(OH)2 to (CH3)2AsH under anaerobic condition C. humicola Methylates CH3AsO(OH)2 and (CH3)2AsO(OH) acid to (CH3)3As. Gliocladium roseum 3 [C. humicola uses AsO Penicillium sp. 2 and AsO4 as substrates to produce (CH3)3As]
Reference Lehr et al. (2003)
Granchinho et al. (2002)
Granchinho et al. (2001) Hasegawa et al. (2001) Andrewes et al. (2000) Kaise et al. (1997)
Cullen et al. (1994) Huysmans and Frankenberger (1991) Baker et al. (1983) Cheng and Focht (1979) Chau and Wong (1978) Cullen et al. (1984) McBride and Wolfe (1971)
Cox and Alexander (1973)
47
48
S. MAHIMAIRAJA ET AL.
that is left in perpetuity. The stabilization of As in the root zone could be achieved through the addition of organic matter as well as soil amendments. In rhizofiltration, the roots can be used to adsorb or absorb metal (loid)s, which are subsequently removed by harvesting the whole plant. In this case, metal(loid) tolerance and translocation of the metal(loid)s to aerial parts are largely irrelevant. In phytoextraction, plants can be grown on contaminated soil and the aerial parts [and the metal(loid)s they contain] harvested. In this case, plants need to be tolerant only if the soil metal(loid) content is very high, but they need to accumulate very high concentrations in their aerial parts. Phytoextraction involves repeated cropping of plants until the metal(loid) concentration in the soil has reached the acceptable (targeted) level. Certain plants, termed “hyperaccumulators” (Brooks et al., 1977), accumulate an inordinate concentration of metal(loid)s in their aboveground biomass. These plants may even accumulate metal(loid)s that are nonessential and often toxic to plants. The minimum concentration of As required for a plant to be classified as a hyperaccumulator of As was set at 1000 mg kg1 (0.1%) on a dry weight basis (Ma et al., 2001). The hyperaccumulation of metal(loid)s involves uptake of the soluble metal(loid) species by the root system, translocation to the aerial parts, and storage in a nontoxic form in the aerial portions. Chaney et al. (1997) suggested that this process necessarily requires tolerance to high concentrations of metal (loid)s. Using a combination of techniques, including X-ray absorption spectroscopy, Pickering et al. (2000) studied the biological mechanisms involved in the accumulation of As in Indian mustard (Brassica juncea) and established the biochemical fate of As taken up by this plant. Arsenic was taken up by roots as oxyanions [As(V) and As(III)], possibly via the H2PO 4 transport mechanism, and a small fraction was exported to the shoot via xylem. Once in the shoot, the As is stored as an As-III-tris-thiolate complex. The majority of the As remains in the roots as an As-III-tris-thiolate complex, which is indistinguishable from that found in the shoots and from As-III-trisglutathione. The thiolate donors are thus probably either glutathione or phytochelatins. Addition of the dithiol arsenic chelator dimercaptosuccinate to the hydroponic culture medium caused a fivefold increase in the As level in the leaves, although the total As accumulation was increased only marginally. This indicates that the addition of dimercaptosuccinate to Ascontaminated soils is likely to facilitate As bioaccumulation in plant shoots, a prerequisite for efficient phytoremediation strategy. The high cost of this compound, however, would be an economic concern unless the plants would be able to synthesize it. At present there are about 400 species of known terrestrial plants that hyperaccumulate one or more of several metal(loid)s (Robinson et al.,
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
49
1995). However, until recently no As-hyperaccumulating plants were reported. Ma et al. (2001) discovered an As-hyperaccumulating plant, ladder brake (Pteris vittata L.), a terrestrial fern, which accumulates large amounts (23,000 mg kg1-dry weight basis) of As from soils. The unique property of As hyperaccumulation by the Chinese brake fern is of great significance in the phytoremediation of As-contaminated soils. Therefore, the potential of this fern for phytoremediation of As-contaminated soil was assessed by Tu et al. (2002) in a glasshouse experiment using soils from an abandoned wood preservation site. Results have shown that the Chinese brake accumulated huge amounts of As from soil and that its As concentration increased with the growth period. The As concentration in the fronds was 6000 mg kg1 dry mass after 8 weeks of transplanting and increased to 7230 mg kg1 after 20 weeks. The As concentration increased as fronds aged, with old fronds accumulating as much as 13,800 mg As kg1. Another silver fern [Pityrogramma calomelanos (L.) Link] has also been reported to hyperaccumulate As up to 8350 mg kg1 dry mass from soil containing 135 mg kg1 (Francesconi et al., 2002). It occurs in tropical and subtropical regions of the world and is widely distributed in Thailand where it favors open, high rainfall areas. Some of the studies involving phytoremediation of As in the soil are presented in Table X. Arsenic uptake by plants is associated with the H2PO 4 uptake mechanism, where presumably As(V) is taken up as a H2PO analogue (Pickering 4 et al., 2000). Therefore, there is a growing interest in using P fertilizer to enhance As uptake by plants. Tu and Ma (2003) suggested that phosphate application may be an important strategy for the efficient use of Chinese brake (Pteris vittata L.) to phytoremediate As-contaminated soils. The addition of P fertilizer to As-contaminated soil was found to increase As solubility and mobility and thus increase plant uptake of soil As (Creger and Peryea, 1994). Some selected references on the mobilization of As by phosphate compounds are reported in Table XI. In an hydroponic experiment, Wang et al. (2002) investigated the interactions of As(V) and H2PO 4 on the uptake and distribution of As and P, and As speciation in P. vittata. They found that the plants accumulated As in the fronds up to 27,000 mg kg 1 dry weight, and the frond As to root As concentration ratio varied between 1.3 and 6.7. Increasing the phosphate supply decreased the As uptake markedly, with the effect being greater on root As concentration than on shoot concentration. They concluded that As(V) is taken up by P. vittata via the H2PO 4 transporters, reduced to As (III), and sequestered in the fronds primarily as As(III). In a fly ashamended soil, Qafoku et al. (1999) observed that H2PO 4 displaced both As(III) and As(V), thereby increasing the mobility of As in soils. Thus, the H2PO 4 -induced plant uptake of As could be employed in the phytoremediation of As-contaminated sites.
50
S. MAHIMAIRAJA ET AL. Table X Selected References on Phytoremediation of Arsenic-Contaminated Soil
Plant used Chinese brake ferns (Pteris vittata)
White lupin (Lupinus albus)
Arabidopsis thaliana
Ladder brake (P. vittata L.) Silver fern (Pityrogramma calomelanos) Herb (Mimosa pudica)
Shrub (Melastoma malabrathriccum) Rice (Oryza sativa)
Result and remark As concentration in shoot as high as 20 times the soil As concentration under field condition. Increasing soil pH improved As uptake by plant Fern transfers As rapidly from soil to aboveground biomass with only minimal As in roots Hyperaccumulation of As enhanced by P addition Fronds accumulated as much as 13,800 mg As kg1 (90% As transported to the fronds) As(V) uptake was high. Roots accumulated As under P deficiency. Potentially a good candidate due to rapid growth and adaptability to varying edaphic status Plants accumulated large amounts of As showing some tolerance Removal of 26% of soil As within 20 weeks after transplanting Accumulating in fronds up to 8350 mg kg1 Tolerated high soil As (5200 mg As kg1), accumulating in leaves 77 mg As kg1 Tolerated high soil As (5200 mg As kg1), accumulating in leaves up to 43 mg As kg1 Plants grown on As-treated soil had higher As uptake than plants grown on untreated soil; at concentrations >1500 mg As kg1 plants died
Reference Salido et al. (2003)
Zhang et al. (2002)
Chen et al. (2002) Tu et al. (2002)
Esteban et al. (2003)
Dhankher et al. (2002)
Tu and Ma (2002)
Francesconi et al. (2002)
Visoottiviseth et al. (2002)
Visoottiviseth et al. (2002)
Onken and Hossner (1995)
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
51
Table XI Selected References on the Mobilization of Arsenic by Phosphate Compounds Phosphate compound
Method of investigation
Proposed mechanism
Ca(H2PO4)2
Transport and leaching
Desorption
NaH2PO4
Competitive adsorption
NaH2PO4
Chemical fractionation; transport and leaching studies Chemical fractionation
NaH2PO4
Phytoavailability bioassay
NaH2PO4
Phytoavailability bioassay
NH4H2PO4 Ca(H2PO4)2
Adsorption and desorption
NH4H2PO4
Transport and leaching
Hydroxyapatite
Chemical fractionation
Competitive adsorption
Competitive adsorption Competitive adsorption
Competitive adsorption Competitive adsorption
Reference Qafoku et al. (1999) Creger and Peryea (1994) Reynolds et al. (1999) Woolson et al. (1973) Livesey and Huang (1981) Peryea (1991); Peryea and Kammereck (1997) Davenport and Peryea (1991) Boisson et al. (1999)
Davenport and Peryea (1991) observed that high rates of monoammonium phosphate (MAP) or monocalcium phosphate (MCP) fertilizers significantly increased the amount of As leached from the soil. Mixing high rates of MAP or MCP fertilizers with orchard soil, Peryea (1991) reported that As release from lead–arsenate-contaminated soil was positively related to the level of P input but was not significantly influenced by the P source. Arsenic solubility was regulated by specific H2PO 4 –AsO4 exchange, whereas H2PO solubility was controlled by the equilibria of metastable P miner4 als. Results indicate that the use of P fertilizers on such soils has the potential to greatly enhance the downward movement of As (Peryea and Kammereck, 1997). Thus the increased mobilization of As resulting from phosphate input can result in its increased leaching to groundwater, especially in the absence of active plant growth. Hence attempts to use plants to remove As from soils need to take the multiple effects of phosphate into consideration. Phytoremediation has several advantages over other remediation and metal(loid) extraction technologies. The cost involved in phytoremediation is much lower than other technologies, such as soil removal, capping, and ex situ cleansing. Other advantages include the ultimate fertility of the cleaned site, the high public appeal of “green” technology, and the possibility
52
S. MAHIMAIRAJA ET AL.
of producing secondary products that offset the cost of the operation or even produce a small profit. However, some of the basic plant physiological processes, such as low biomass production and shallow root growth, nonetheless limit the scope of phytoremediation. Only surface contamination can be removed or degraded and the cleanup is restricted to areas that are amenable to plant growth. Most importantly, it may take a long time for site remediation to be effective. Phytoremediation can only be used if it meets environmental regulation during the operation as well as its end point.
B. REMOVAL OF ARSENIC
FROM
AQUATIC ENVIRONMENTS
As discussed earlier, because most cases of As toxicity in humans have resulted from the consumption of As-contaminated water, there have been intensive research efforts in developing technologies aimed at stripping As from water. A plethora of methods suitable for the removal of As from water at both household and community levels are currently available. These methods are primarily based on (i) removal of solid-phase As through coagulation, sedimentation, or filtration; (ii) removal of solution-phase As through ion exchange, osmosis, or electrodialysis; (iii) oxidation of As(III) to As(V) and its subsequent removal through adsorption and/or precipitation; (iv) biosorption using microorganisms; and (v) rhizofiltration using aquatic plants. Some of the methods that have been tested for the removal of As from water are presented in Table XII.
1.
Physicochemial Methods
Filtration, adsorption, and chemical precipitation are the most common physicochemical methods used for stripping As from water. While the particulate As in water can be removed by simple filtration, the aqueous As can be removed through adsorption or precipitation followed by filtration. a. Filtration. Most of the domestic drinking water treatment systems for As removal involve filtration. For example, the “Pitcher filter” involving porous ceramics (Neku and Tandukar, 2003) and sand filters (Yokota et al., 2001) have been found to be effective in stripping As from water. Seidel et al. (2001) noticed that the porous nanofiltration anion-exchange membrane removed about 90% of As(V) present in water at a concentration of 316 g liter1. Although this technology could achieve a high degree of As removal, it involves a high initial investment and high operation and maintenance costs.
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
53
Table XII Selected References on Methods of Arsenic Stripping from Water Method 3-Gagri (Pitcher) filter Aeration and sand filtration
Pond sand filter system Negatively charged porous nanofiltration (NF) membrane Using rare earth oxides
Iron oxide-coated sand (IOCS)
Coprecipitation with Fe
Porous NF membrane
Iron oxide-coated sand and ferrihydrite (IOCS and FH)
Iron-sulfide minerals (pyrite and pyrrhotite) Kimberlite tailing (mineral waste from diamond mining) Mesoporous anions traps (metal-chelated ligands immobilized on anionbinding silica material) Aquifer materials (composed of quartz, feldspar, calcite, chlorite, illite, and magnetite/hematite)
Remark
Reference
Removed 76–95% of As. Suitable for household use Removed 62–92% of As containing 240–320 g liter1 Removed >99 % of 5 mg As liter1 60–90% removal of As(V) from water containing 10–316 g liter1 Adsorbed As(V) rapidly and effectively; >90% of adsorption occurred within the first 10 min, adsorbed As (V) could be desorbed by washing with pH 12 solution Very effective in removing As (III) and As(V) from drinking water containing 200 to 1700 g liter1; about 94% removal efficiency Bench scale test showed 88% of As(III) in water removed by settlement over 24 h As removal by 60–90 % from drinking water containing As from 10 to 316 g liter1 90% removal of As from natural water containing 325 g liter1; adsorption of IOCS and FH estimated at 18.3 and 285 g g1, respectively Fe-sulfides are very effective in removing As [both As(III) and As(V)] from water Removed As at a rate of 270 g g1; more efficient at near neutral pH. 90–94% removal in 12 h Most As removed from water containing >120 mg liter1; adsorption at 120 mg g1
Neku and Tandukar (2003)
Removed As(III) from water through adsorption
Berg et al. (2001)
Yokota et al. (2001) Seidel et al. (2001)
Raichur and Panvekar (2002)
Yuan et al. (2002)
Mamtaz and Bache (2000)
Vrijenhoek and Waypa (2000) Thirunavkukkarasu et al. (2001)
Han and Fyfe (2000)
Dikshit et al. (2000)
Fryxell et al. (1999)
Carrillo and Drever (1998)
54
S. MAHIMAIRAJA ET AL.
k
k
k
k
b. Adsorption. A number of compounds, including activated alumina, Fe-coated sand, and ion-exchange resins are used to adsorb As. In most geologic environments, Fe2O3 carries a positive surface charge that preferentially adsorbs As. Similarly, Al(OH)3 and silicate clays also adsorb large amounts of As. Yoshida et al. (1976) investigated the removal of As from water using “brown gel,” which is a silica gel containing 6% of Fe(OH)3, and observed that the maximum adsorption (17 g As kg1) of both As(III) and As(V) occurred at pH 6. Rothbaum and Buisson (1977) found that synthetic Fe-floc [Fe(OH)3], prepared by treating FeSO4 with NaOCl at pH 3.5–5.0, removed a large percentage of As from geothermal discharge water through coprecipitation. Similarly, Yuan et al. (2002) examined the potential value of several Fetreated natural materials such as Fe-treated activated carbon, Fe-treated gel beads, and Fe oxide-coated sand in removing As from drinking water under both laboratory and field conditions. The Fe oxide-coated sand consistently achieved a high degree (>94%) of As(III) and As(V) removal. When the pH was increased from 5 to 9, As(V) adsorption decreased slightly, but As(III) adsorption remained relatively stable. Kimberlite tailings (Dikshit et al., 2000) and iron-sulfide minerals such as pyrite and pyrrhotite (Han and Fyfe, 2000) were also found to be very effective adsorbents in stripping both As(III) and As(V) from water. Hlavay and Polyak (1997) developed and tested novel adsorbents for As stripping. Porous support materials were granulated using Al2O3 and/or TiO2 and then Fe(OH)3 was freshly precipitated onto the surface of these particles. The resulting Fe(OH)3-impregnated porous adsorbent was dried at room temperature and packed into an ion-exchange column. These columns were found to remove >85% of As in water. The As(III) ions can primarily be adsorbed by chemical reaction on the surface of Fe(OH)3. The neutral functional group of { FeOH} reacts with H2AsO 3 ions, and surface compounds of { FeAsO3H2}, { FeAsO3H}, and { FeAsO2} can be formed. Das et al. (1995) demonstrated the practical application of the adsorption technique in stripping As by developing a simple household device to remove As from groundwater used for drinking and cooking purposes. The system consists of a filter, tablet, and two earthen or plastic jars. The tablet contains Fe(III) salt, an oxidizing agent, and activated charcoal. The filter is made of mainly purified fly ash with binder. When the tablet is added to water (one tablet for every 20 liters), the As(III) ions are catalytically oxidized to As(V) ions in the presence of Fe(III), which are subsequently adsorbed onto activated charcoal and hydrous ferric oxide (Fe2O3.2–3H2O). In addition to As(V), As(III) ions are also strongly adsorbed by Fe(III) oxides. The water is allowed to settle for about an hour and is then filtered. This stripping system has been installed in several locations in Bangladesh and
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
55
West Bengal, and analytical results have shown that generally 93–100% of the total As in water (with an initial concentration of 149–463 g liter1) is removed. Khan et al. (2000) evaluated the efficiency of a simple three-pitcher filter system consisting of ceramic filters (locally known as 3-kalshi) in stripping As from groundwater. In the 3-kalshi assembly, the first kalshi (pot) contains Fe chips and coarse sand, the second contains wood charcoal and fine sand, and the third is the collector for the filtered water. Depending on the size of the filtering units, this system has been shown to be capable of reducing the As concentration in water from an initial level of 1100 g liter1 to below the detection limit of 2 g liter1 with a corresponding decrease in dissolved Fe concentration (from 6000 to 200 g liter1). Similarly, Kim et al. (2004) have shown that mesoporous alumina with a wide surface area (307 m2 g1), high pore volume (0.39 m3 g1), uniform pore size (3.5 nm), and interlinked pore system is efficient in stripping As from domestic water. The mesoporous alumina is insoluble and stable within the range of pH 3–7. The maximum As adsorption was seven times higher [121 mg As(V) g1 and 47 mg As(III) g1] than that of conventional activated alumina, and the kinetics of adsorption are also rapid with complete adsorption in less than 5 h as compared to conventional alumina (about 2 days to reach half of the initial concentration). Fryxell et al. (1999) used metal-chelated ligands immobilized on mesoporous silica as a novel anion-binding material to remove As from water. Nearly complete removal of As(V) has been achieved from solutions containing more than 100 mg As (V) liter1. c. Precipitation. Arsenate can be removed by precipitation/coprecipitation using Fe and Al compounds [Eqs. (27–33) in Table VI]. Gulledge and O’Connor (1973) achieved a complete removal of As(V) from water using Fe2(SO4)3 at a pH range of 5 to 7.5 [Eq. (34)]. Hydrolyzing metal salts such as FeCl3 and alum [Al2(SO4)3] have been shown to be effective in stripping As by coagulation. Hering et al. (1997) achieved >90% removal of As(V) from water containing an initial concentration of 100 g As liter1. Shen (1973) removed As from drinking water by dosing with chlorine (Cl2) and FeCl3. Oxidation of As(III) to As(V) by Cl2 and the subsequent removal by precipitation were considered the mechanisms involved in this process. Treating drinking water with Fenton’s reagent (ferrous ammonium sulfate and H2O2) followed by passing through elemental Fe, Krishna et al. (2001) achieved As removal below the USEPA maximum permissible limit of 50 g liter1 from an initial concentration of 2000 g liter1 of As(III). This method is simple and cost effective for use at community levels. Using a bench scale test, Mamtaz and Bache (2000) demonstrated that up to 88% of the As(III) in water could be removed by coprecipitation with naturally
56
S. MAHIMAIRAJA ET AL.
occurring Fe found in groundwater. One of the advantages in chemical precipitation method is that this can be used at both household and community levels. The materials are readily available and generally inexpensive. However, a problem of disposal of toxic sludge exists and it also requires trained operators.
2. Biological Methods a. Phytoremediation using Aquatic Plants. Phytoremediation of Ascontaminated waters may be readily achieved by the use of aquatic plants because unlike soil, most of the As in water is available for plant uptake. In the case of soils, the plant must first solubilize the metal(loid)s in the rhizosphere and then should have the ability to transport it to the aerial tissue (Brooks and Robinson, 1998). The use of freshwater vascular plants for the removal of metal(loid)s from water has been long established. There are two approaches in using these plants for the remediation of polluted water: The first involves monospecific pond cultures of free-floating plants such as water hyacinth. The plants accumulate the metal(loid)s until a steady state of equilibrium is achieved. They are then harvested by removal from the pond. The second approach involves growing rooted emergent species in trickling bed filters. Rhizosphere microbes usually facilitate the removal of metal(loid)s in these systems. Rhizofiltration usually involves the hydroponic culture of plants in a stationary or moving aqueous environment wherein the plant roots absorb metal(loid)s from the water (Brooks and Robinson, 1998). Ideal plants for rhizofiltration should have extensive root systems and be able to remove metal(loid)s over an extended period. Some of the aquatic plants capable of accumulating large amounts of As are presented in Table XIII. Robinson et al. (2004) undertook a field survey in which a number of terrestrial and aquatic plant samples were taken at several sites within the Taupo volcanic zone (TVZ) in New Zealand. The TVZ covers an area of 600,000 ha in the central North Island of New Zealand and the area is rich in geothermal activity. There have been previous reports of elevated As concentrations in some waterways and associated lands in the TVZ (Liddle, 1982). The known sources of As pollution in the TVZ include (i) As arising from naturally occurring geothermal activity; (ii) geothermal bores that release As-rich water into the aquatic biosphere; (iii) runoff of As-based pesticides; (iv) As from timber treatment sites such as the pulp and paper mill at Kinleith; and (v) As added to lakes to control weeds (e.g., NaAsO2 added to Lake Rotorua). The mean As concentrations in all the plants tested from the TVZ are given in Fig. 6. Data clearly display the difference of As accumulation
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
57
Table XIII Selected Aquatic Plants for Potentially Stripping Arsenic from Water
Name of plant Agrostis capillaris Ceratophyllum demersum C. demersum C. demersum Egeria densa Lagarosiphon major Rorippa naturtium (subsp. Aquaticum) Cynodan dactylon Spergularia grandis Paspalum tuberosum Fern (Pteris vittata) Fern (P. vittata) Silver fern (Pityrogramma calomelanos) Fern (Pteris cretica) Fern (P. longifolia) Fern (P. umbrosa) Watercress (Lepidium sativum) Myriophyllum propinquum Elodea canadensis Agrostis sp a
Level of As accumulation (mg kg1)a
Reference
3470 650
Porter and Peterson (1975) Reay (1972)
265–1121 44–1160 94–1120 11–1200 >400
Liddle (1982) Robinson et al. (1995) Robinson et al. (1995) Robinson et al. (1995) Robinson et al. (1995)
1600 1175 1130 22,630 8960–27,000 8350
Jonnalagadda and Nenzou (1997) Bech et al. (1997) Bech et al. (1997) Ma et al. (2001) Wang et al. (2002) Visoottiviseth et al. (2002)
6200–7600
Zhao et al. (2002)
12–1766
Robinson et al. (2003a)
974–3900 1628–1857 800
Machetti (2003) Machetti (2003) Machetti (2003)
Dry weight basis.
between aquatic and terrestrial plants. Aquatic plants, grouped on the left-hand side of Fig. 6, had As concentrations up to 4000 mg kg1 on a dry matter basis. In contrast, terrestrial plants, on the right-hand side of Fig. 6, showed much lower As concentrations. All the aquatic plants tested accumulated As at concentrations greater than 5 mg kg1 on a dry matter basis, and none of the terrestrial plants tested had As concentrations surpassing 11 mg kg1. Most of the terrestrial plants tested were below the detection limit for As (0.5 mg kg1) even when growing in soil containing up to 89 mg As kg1. The difference in metal(loid) accumulation between aquatic and terrestrial plants was noticed by Outridge and Noller (1991) in their review of hyperaccumulation of elements by aquatic plants. Although they did not provide an explanation of this phenomenon, various reasons could be
58
S. MAHIMAIRAJA ET AL.
Figure 6 Mean arsenic concentration in plants collected from the Taupo volcanic zone (TVZ) (Robinson et al., 2004).
attributed for the difference in As accumulation between aquatic and terrestrial plants. For instance, in terrestrial systems, the solubilization of As in the rhizosphere is necessary to allow the plant roots to take up and transport this element to the aerial parts of the plant. This is not the case when the plant grows in an aqueous medium, where the metal(loid) is already present in a bioavailable form (Brooks and Robinson, 1998). b. Microbial Removal of Arsenic. Biosorption and biomethylation are the two important processes by which metal(loid)s, including As, are removed from water using microorganisms. The biosorptive process generally lacks specificity in metal(loid) binding and is sensitive to ambient environmental conditions, such as pH, solution composition, and the presence of chelators. Genetically engineered microorganisms (e.g., Escherichia coli) that express a metal(loid)-binding protein (i.e., metallothionein) and a metal(loid)-specific transport system have been found to be successful in their selectivity for accumulation of a specific metal (loid) in the presence of a high concentration of other metal(loid)s and chelating agents in solution (Chen and Wilson, 1997). These organisms also have potential application to remove specific metal(loid)s from contaminated soil and sediments.
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
59
Biosorption is one of the promising technologies involved in removing As from water and wastewater. Several chemically modified sorbents have been examined for their efficiency in removing metalloids. Loukidou et al. (2003) examined the potential of Penicillum chrysogenum, a waste by-product from antibiotic production, for the removal of As(V) from wastewaters. They reported that the pretreatment of biomass with common surfactants (as hexadecyl-trimethylammonium bromide and dodecylamine) and a cationic polyelectrolyte was found to remove a significant amount of As(V) from waters. At pH 3, the removal capacities of modified biomass ranged from 33.3 to 56.1 mg As g1 biomass. Methylation is the most reliable biological process through which As can be removed from aquatic medium. Certain fungi, yeasts, and bacteria are known to methylate As to gaseous derivatives of arsine. Commercial application of biotransformation of metal(loid)s in relation to the remediation of metal(loid)-contaminated water was documented by Bender et al. (1995). They examined the removal and transformation of metal(loid)s using microbial mats, which were constructed by combining cyanobacteria with a sediment inoculum from a contaminated site. When water containing high concentrations of metal(loid)s was passed through the microbial mat, there was a rapid removal of the metal(loid)s from the water. The mat was found to be tolerant of high concentrations of toxic metal(loid)s such as Cd, Pb, Cr, Se, and As (up to 350 mg liter1). Management of toxic metal(loid)s by the mat was attributed to the deposition of metal(loid) compounds outside the cell surfaces, as well as chemical modification of the aqueous environment surrounding the mat. Large quantities of metal(loid)-binding polysaccharides were produced by the cyanobacterial component of the mat. Photosynthetic oxygen production at the surface and heterotrophic consumption in the deeper regions resulted in steep gradients of redox condition in the mat. Additionally, sulfur-reducing bacteria colonized the lower strata, removing and utilizing the metal(loid) sulfide. Thus, depending on the biochemical characteristics of the microzone of the mat, the sequestered metal(loid)s could be oxidized, reduced, and precipitated as sulfides or oxides.
C.
MULTISCALAR-INTEGRATED RISK MANAGEMENT
A number of challenging issues need to be taken into consideration when devising strategies to manage As contamination of the environment. These include the following. i. Complexity of As contamination—the severity and long-term persistence of As contamination are influenced by factors such as medium characteristics, site hydrogeology, land and water use, source term, chemical form and speciation, and target organism.
60
S. MAHIMAIRAJA ET AL.
ii. Presence of multichemical species—As undergoes several biogeochemical transformation processes, resulting in the release of an array of chemical species that differ in their biogeochemical reactions, bioavailability, and biotoxicity. iii. Extent and magnitude of As contamination of groundwater resource— for example, in Bangladesh, As in groundwater is derived from geological weathering of parent rock materials from the Indo-Gangetic alluvial plains spread over an area of millions of hectares iv. Multipurpose end use of contaminated resources—water is used for drinking, cooking, and other household purposes and for irrigation; similarly, soil is used for agricultural production and recreational activities. It is therefore important to formulate and/or devise integrated risk management strategies involving source avoidance, source reduction, and remediation. Source avoidance, which refers to avoiding the most contaminated source of the groundwater relative to certain geological strata, can be practiced to minimize the risk resulting from As contamination of soil and water resources. For example, in Bangladesh, shallow dug wells are increasingly becoming popular as an alternative to pump water from deeper strata. In some cases, the relatively contaminant-free strata are below 250-m deep zones. However, sanitation of these shallow wells is paramount to avoid gastroenteritis and other pathogenic-borne diseases. Another strategy is source reduction, which refers to removing or stopping the source of contamination. Source reduction can be achieved easily when the contamination source is of anthropogenic origin, such as those in landfills or similar point sources. As discussed earlier, in most regions, As contamination of groundwater is largely of geogenic origin, and source reduction may not be a feasible option to manage As contamination. Remediation of contaminated soil and water resources requires both short-term and long-term solutions to the As problem. Therefore, the remediation strategies should be aimed at multiscalar levels, i.e., household level to community and regional levels, representing the various levels of complexicity. Depending on the efficiency and cost effectiveness of the system, a combination of technologies may be required at certain levels. The potential technologies for remediation of As-contaminated soil and water resources at different scales in relation to the end use of the resources are depicted in Fig. 7. For example, at the least complex household level, remediation strategies involving only a simple filter (sorptive) system can be used to remove As (i.e., As stripping) from water used for drinking and cooking purposes, whereas at a more complex community level, more sophisticated precipitation technologies should be used to strip As from the community water supply so that cost can be shared and the system can be managed
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
61
Figure 7 Multiscalar risk management for arsenic-contaminated soil and aquatic ecosystems.
efficiently. More sophisticated stripping methods, which may require a series of a filtering–sorptive (precipitation) setup, are necessary in order to cope with the enormous volume of groundwater that needs to be treated before distribution to the community. Even at the community scale, the situation becomes even more complex when dealing with impacted soils, especially those geared for food production. In this case, land use is a very important factor to address. For example, in parks, applying soil amendments such as those high in Fe2O3 may suffice to mitigate As risk. In contrast, technologies might be paired in a situation when the food chain might be compromised, as typified by rangeland, rice paddy, and so on. A viable approach in this circumstance is to apply phytoremediation during the initial period (1 to 2 years) to strip the “bioavailable” fraction, subsequently followed by soil amendments before committing to the intended land use. It is very important to observe that as the level of contamination becomes more complex, a monitoring scheme should be in place. Hence, a successful remediation scheme for an As-contaminated environment should aim for an integrated approach involving the possible combination of physical, chemical, and/or biological mechanisms. It is essential that the integration of remediation technologies should enhance efficiency, both technologically and economically, resulting in a
62
S. MAHIMAIRAJA ET AL.
Figure 8 Conceptual integrated approach for remediation of arsenic-contaminated soil and aquatic ecosystems, focusing on phytoremediation.
reduction in the time required for achieving targeted levels of As. For example, phytoremediation is a promising new technology, which is relatively inexpensive and has been proven effective in the large–scale remediation of both soil and water resources. Further, it would also add “green” value (aesthetic) to the environment. Integrating physical, chemical, and/or bioremedial measures with phytoremediation as depicted in Fig. 8 could enhance a higher uptake of As by plants, can more effectively minimize biotoxicity through microbial and chemical immobilization, and can potentially eliminate As through the inducement of biomethylation and subsequent volatilization from the system.
VII. SUMMARY AND FUTURE RESEARCH NEEDS Arsenic is an extremely toxic and carcinogenic metalloid contaminant that adversely affects the environment and human health. Widespread As contamination of terrestrial and aquatic environments from both geogenic
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
63
and anthropogenic sources has been reported in many countries. Although not anthropogenic, drinking of As-contaminated water has already affected millions of people, particularly in developing countries with the biggest known As calamity occurring in Bangladesh and West Bengal in India. Arsenic in soil and water exists in a different valence state, but predominantly as toxic As(III) and less toxic As(V). The biogeochemistry of As in soil and water is complex and is mostly determined by its chemical speciation resulting from chemical and biological transformations. The chemistry of soil and water (i.e., pH and Eh) and predominantly microbial assemblages play a major role in As dynamics. Although bioaccumulation of As in plants and organisms has been reported, its biochemical transformations within the plant and other biota are still largely unknown. Risk management of As-contaminated soil and aquatic ecosystems is an important issue and a great challenge; its success is necessary to promote sustainable environmental health and also to minimize the adverse impact on humans. A number of physical, chemical, and biological technologies involving simple filtration, precipitation, biosorption, and rhizofiltration have been developed to remediate As-contaminated soil and water. Conventional physical and chemical remedial measures usually are quite expensive but may prove highly effective. However, most of these technologies have been tested only at the laboratory and pilot scale levels. Large-scale application of such technologies requires trained personnel for the operation of equipment to treat soils and waters. However, phytoremediation, which is relatively inexpensive, has been proven effective in the remediation of metal (loid)-contaminated sites. Certain As-hyperaccumulating plants offer a wide scope for the phytoremediation of As-contaminated soil and water. Nonedible crops, such as ornamental and fuel crops, may be suitable for phytoremediation through which the entry of As into the food chain could largely be avoided. Bioremediation, using biological wastes and/or microbial strains, offers another avenue for remediation. However, as in the case of physical and chemical technologies, most of the research involving bioremediation has been demonstrated in the laboratory only. As such, its feasibility should be tested under diverse field conditions. Remediation of As-contaminated soils and As stripping from potable and irrigation waters require a multiscalar approach. This involves an “end-use” specific (i.e., drinking vs irrigation and agricultural vs recreational sites) integrated approach, involving a combination of physical, chemical, and biological technologies for the successful and effective management of Ascontaminated environments. Future research is, therefore, needed for the following: • Biogeochemical mechanisms governing As dynamics in different media using advanced spectroscopic-based techniques.
64
S. MAHIMAIRAJA ET AL.
• Elucidation of soil and water environmental factors (e.g., pH and Eh) that govern chemical and biological transformations of As. • Examination of solid-phase and solution-phase speciation of As in soil and water. • Identification of biochemical mechanisms involved in the accumulation of As in specific tissues or organs in plants, animals, and humans. This includes the interactive effects of As(V) and H2PO 4 on hyperaccumulators such as Chinese brake and water cress. • Evaluation of As phytotoxicity under field conditions. • Rhizosphere processes underpinning effective phytoremediation technologies. • Mycorrhizal role in the bioremediation of As regarding biomethylation, biooxidation, and immobilization of As. • Developing genetically engineered microorganisms and genetically modified plants to detoxify As in contaminated soil and water. • In situ immobilization techniques in contaminated soils/sediments using inexpensive industrial by-products high in metallic oxides; effect of aging on the release of As from the immobilized media. • Biomonitors of As as a tool in the risk assessment of As-contaminated sites. • Highly effective and expensive stripping methods for the removal of As in domestic water supplies destined for irrigation and human consumption.
ACKNOWLEDGMENTS The senior author thanks Massey University Research Foundation for the award of the Postdoctoral Fellowship. The U.S. Department of Energy Contract Number DE-FC-09-96SR18546 with the University of Georgia’s Savannah River Ecology Laboratory supported Drs. Bolan and Adriano’s writing/editing time.
REFERENCES Abedin, M. J., Cresser, M. S., Meharg, A., Feldmann, J., and Cotter-Howells, J. (2002). Arsenic accumulation and metabolism in rice (Oryza sativa L.). Environ. Sci. Technol. 36, 962–968. Acharyya, S. K., Chakraborty, P., Lahiri, S., Raymahashay, B. C., Guha, S., and Bhowmik, A. (1999). Arsenic poisoning in the Ganges delta. Nature 401, 545. Adriano, D. C. (2001). “Trace Elements in Terrestrial Environments: Biogeochemistry, Bioavailability and Risks of Metals”, 2nd edn, Springer, New York.
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
65
Adriano, D. C., Page, A. L., Elseewi, A. A., Chang, A. C., and Straughan, I. (1980). Utilization and disposal of fly ash and other residues in terrestrial ecosystems: A review. J. Environ. Qual. 9, 333–344. Aggett, J., and Aspell, A. C. (1976). The determination of arsenic(III) and total arsenic by atomic absorption spectroscopy. Analyst 101, 341–347. Aggett, J., and Aspell, A. (1980). Arsenic from geothermal sources in the Waikato catchment. N. Z. J. Sci. 23, 77–82. Aichberger, K., and Hofer, G. F. (1989). Contents of arsenic, mercury and selenium in agricultural soils of upper Austria. Bodenkultur 40, 1–11. Alam, M. B., and Sattar, M. A. (2000). Assessment of arsenic contamination in soils and waters in some areas of Bangladesh. Water Sci. Technol. 42, 185–192. Alauddin, M., Alauddin, S. T., Bhattacharjee, M., Sultana, S., Chowdhury, D., Bibi, H., and Rabbani, G. H. (2003). Speciation of arsenic metabolite intermediates in human urine by ion-exchange chromatography and flow injection hydride generation atomic absorption spectrometry. J. Environ. Sci. Health A. 38, 115–128. Ali, M., and Tarafdar, S. A. (2003). Arsenic in drinking water and scalp hair by EDXRF: A major recent health hazard in Bangladesh. J. Radioanal. Nuc. Chem. 256, 297–305. Allison, J. D., Brown, D. S., and Novo-Gardac, K. J. (1991). “MINTEQA2/PRODEFA2, a Geochemical Assessment Model for Environmental System (EPA/600/3-91/021)”. US Environmental Protection Agency, Athens, GA. Amit, C., Ajaylal, M., Chatterjee, A., and Mukherjee, A. (1999). Hydrogeological investigation of ground water arsenic contamination in south Calcutta. Sci. Total Environ. 225, 249–262. AmonooNeizer, E. H., Nyamah, D., and Bakiamoh, S. B. (1996). Mercury and arsenic pollution in soil and biological samples around the mining town of Obuasi, Ghana Wat. Air Soil Pollut. 91, 363–373. Anderson, L. C. D., and Bruland, K. W. (1991). Biogeochemistry of arsenic in natural waters; the importance of methylated species. Environ. Sci. Technol. 25, 420–427. Andreae, M. O. (1977). Determination of arsenic species in natural waters. Anal. Chem. 49, 820–825. Andreae, M. O., and Klumpp, D. (1979). Biosynthesis and release of organoarsenic compounds by marine algae. Environ. Sci. Technol. 13, 738–741. Andrewes, P., Cullen, W. R., and Polishchuk, E. (2000). Arsenic and antimony biomethylation by Scopulariopsis brevicaulis: Interaction of arsenic and antimony compounds. Environ. Sci. Technol. 34, 2249–2253. Arai, Y., Elzinga, E. J., and Sparks, D. L. (2001). X-ray absorption spectroscopic investigation of arsenite and arsenate adsorption at the aluminum oxide-water interface. J. Colloid Interface Sci. 235, 80–88. Arai, Y., Lanzirotti, A., Sutton, S., Davis, J. A., and Sparks, D. L. (2003). Arsenic speciation and reactivity in poultry litter. Environ. Sci. Technol. 37, 4083–4090. Arai, Y., and Sparks, D. L. (2002). Residence time effects on arsenate surface speciation at the aluminium oxide-water interface. Soil Sci. 167, 303–314. Armishaw, R. F., Fricker, A. G., and Fenton, G. A. (1994). Soil and groundwater studies at some CCA timber treatment sites. Water Wastes NZ. 79, 44–48. Armstrong, C. W., Stroube, R. B., and Rubio, T. (1984). Outbreak of fatal arsenic poisoning caused by contaminated drinking water. Arch. Environ. Health 39, 276–279. Ashley, P. M., and Lottermoser, B. G. (1999). Arsenic contamination at the Mole River Mine, northern New South Wales. Aust. J. Earth Sci. 46, 861–874. Ayotte, J. D., Montgomery, D. L., Flanagan, S. M., and Robinson, K. W. (2003). Arsenic in groundwater in eastern New England: Occurrence, controls, and human health implications. Environ. Sci. Technol. 37, 2075–2083.
66
S. MAHIMAIRAJA ET AL.
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. Basta, N. T., Gradwohl, R., Snethen, K. L., and Schroder, J. L. (2001). Chemical immobilization of lead, zinc, and cadmium in smelter-contaminated soils using biosolids and rock phosphate. J. Environ. Qual. 30, 1222–1230. Bech, J., Poschnerieder, C., Llugany, M., Barcelo, J., Tume, P., Tobias, F. J., Barranzuela, J. L., and Vasquez, E. R. (1997). Arsenic and heavy metal contamination of soil and vegetation around a copper mine in northern Peru. Sci. Total Environ. 203, 83–91. Bednar, A. J., Garbarino, J. R., Ranville, J. F., and Wildeman, T. R. (2002). Preserving the distribution of inorganic arsenic species in groundwater and acid mine drainage samples. Environ. Sci. Technol. 36, 2213–2218. Bencko, V., and Symon, K. (1977). Health aspects of burning coal with a high arsenic content. Environ. Res. 113, 378–385. Bender, J., Lee, R. F., and Phillips, P. (1995). Uptake and transformation of metals and metalloids by microbial mats and their use in bioremediation. J. Indust. Microbiol. 14, 113–118. Benson, N. R. (1976). Retardation of apple tree growth by soil arsenic residues. J. Am. Soc. Hort. Sci. 101, 251–253. Beretka, J., and Nelson, P. (1994). The current state of utilization of fly ash in Australia. In “Ash: A Valuable Resource”, Vol. 1, pp. 51–63. South African Coal Ash Association. Berg, M., Tran, H. C., Nguyen, T. C., Pham, H. V., Schertenleib, R., and Giger, W. (2001). Arsenic contamination of groundwater and drinking water in Vietnam: A human health threat. Environ. Sci. Technol. 35, 2621–2626. Bettley, F. R., and O’Shea, J. A. (1975). The absorption of arsenic and its relation to carcinoma. Br. J. Dermatol. 92, 563–568. Bisessar, S. (1982). Effects of heavy metals on microorganisms in soils near a secondary lead smelter. Wat. Air Soil Pollut. 17, 305–308. Bishop, R. F., and Chisholm, D. (1961). Arsenic accumulation in Annapolis Valley orchards. Can. J. Soil Sci. 42, 77–80. Boisson, J., Mench, M., Vangronsveld, J., Ruttens, A., Kopponen, P., and DeKoe, T. (1999). Immobilization of trace metals and arsenic by different soil additives: Evaluation by means of chemical extractions. Commun. Soil Sci. Plant Anal. 30, 365–387. Bolan, N. S., Adriano, D. C., and Mahimairaja, S. (2004). Distribution and bioavailability of trace elements in livestock and poultry manure by-products. CRC Crit. Rev. Environ. Sci. Tech. 34, 291–338. Bolan, N. S., Adriano, D. C., and Naidu, R. (2003). Role of phosphorus in (im)mobilization and bioavailability of heavy metals in the soil-plant system. Rev. Environ. Contam. Toxicol. 177, 1–44. Bolan, N. S., and Thiyagarajan, S. (2001). Retention and plant availability of chromium in soils as affected by lime and organic amendments. Aust. J. Soil Res. 39, 1091–1103. Bose, P., and Sharma, A. (2002). Role of iron in controlling speciation and mobilization of arsenic in subsurface environment. Wat. Res. 36, 4916–4926. Bothe, J. V., and Brown, P. W. (1999). Arsenic immobilization by calcium arsenate formation. Environ. Sci. Tech. 33, 3806–3811. Bowell, R. J., Morley, N. H., and Din, V. K. (1994). Arsenic speciation in soil porewaters from the Ashanti mine, Ghana. Appl. Geochem. 9, 15–22. Braman, R. S., and Foreback, C. C. (1973). Methylated forms of arsenic in the environment. Science 182, 1247–1249. Brookins, D. G. (1988). In “Eh-pH Diagrams for Geochemistry”, p. 175. Springer-Verlag, New York.
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
67
Brooks, R. R., Lee, J., Reeves, R. D., and Jaffre, T. (1977). Detection of nickeliferous rocks by herbarium species of indicator plants. J. Geochem. Explor. 7, 49–57. Brooks, R. R., and Robinson, B. H. (1998). Aquatic phytoremediation by accumulator plants. In “Plants that Hyperacccumulate Heavy Metals: Their Role in Phytoremediation, Microbiology, Archaeology, Mineral Exploration and Phytomining” (R. R. Brooks, Ed.), pp. 203–226. CAB International, Wallingford. Brown, K. G., and Ross, G. L. (2002). Arsenic, drinking water, and health: A position paper of the American Council on Science and Health. Regul. Toxicol. Pharmacol. 36, 162–174. Buchet, J. P., Lauwerys, R., and Roels, H. (1981). Comparison of the urinary excretion of arsenic metabolites after a single oral dose of sodium arsenite, monomethylarsonate or dimethylarsinate in man. Int. Arch. Occup Environ. Health 48, 71–79. Burlo, F., Guijarro, I., Barrachina, A. A. C., and Vlaero, D. (1999). Arsenic species: Effects on and accumulation by tomato plants. J. Agric. Food Chem. 47, 1247–1253. Burnichon, V., Jean, S., Bellon, L., Maraninchi, M., Bideau, C., Orsiere, T., Margotat, A., Gerolami, V., Botta, A., and Berge-Lefranc, J. L. (2003). Patterns of gene expressions induced by arsenic trioxide in cultured human fibroblasts. Toxicol. Lett. 143, 155–162. Buyuktuncel, E., Bektas, S., Salih, B., Evirgen, M. M., and Genc, O. (1997). Arsenic speciation in geothermal waters by HPLC/GFAAS and HPLC/HGAAS methods. Fresenius Environ. Bull. 6, 494–501. Caper, S. G., Tanner, J. T., Friedman, M. H., and Boyer, K. W. (1978). Multielement analysis of animal feed, animal wastes, and sewage sludge. Environ. Sci. Technol. 12, 785–790. Cappuyns, V., Van Herreweghe, S., Swennen, R., Ottenburgs, R., and Deckers, J. (2002). Arsenic pollution at the industrial site of Reppel-Bocholt (north Belgium). Sci. Total Environ. 295, 217–240. Carbonell-Barrachina, A. A., Burlo, F., Burgos-Hernandez, A., Lopez, E., and Mataix, J. (1997). The influence of arsenite concentration on arsenic accumulation in tomato and bean plants. Scientia Hort. 71, 167–176. Carbonell-Barrachina, A. A., Burlo, F., Lopez, E., and Martinez-Sanchez, F. (1999a). Arsenic toxicity and accumulation in radish as affected by arsenic chemical speciation. J. Environ. Sci. Health B 34, 661–679. Carbonell-Barrachina, A. A., Burlo, F., Valero, D., Lopez, E., Martinez-Romero, D., and Martinez-Sanchez, F. (1999b). Arsenic toxicity and accumulation in turnip as affected by arsenic chemical speciation. J. Agric. Food Chem. 47, 2288–2294. Carbonell-Barrachina, A. A., Jugsujinda, A., Burlo, F., Delaune, R. D., and Patrick, W. H. (2000). Arsenic chemistry in municipal sewage sludge as affected by redox potential and pH. Water. Res. 34, 216–224. Carrillo, A., and Drever, J. I. (1998). Adsorption of arsenic by natural aquifer material in the San Antonio El Triunfo mining area, Baja California, Mexico. Environ. Geol. 35, 251–257. Centeno, J. A., Mullick, F. G., Martinez, L., Page, N. P., Gibb, H., Longfellow, D., Thompson, C., and Ladich, E. R. (2002). Pathology related to chronic arsenic exposure. Environ. Health Prespect. 110, 883–886. Chakraborti, D., Rahman, M. M., Paul, K., Chowdhury, U. K., Sengupta, M. K., Lodh, D., Chanda, C. R., Saha, K. C., and Mukherjee, S. C. (2002). Arsenic calamity in the Indian subcontinent: What lessons have been learned? Talanta 58, 3–22. Chaney, R. L., Malik, M., Li, Y. M., Brown, S. L., Brewer, E. P., Angle, J. S., and Baker, A. J. M. (1997). Phytoremediation of soil metals. Curr. Opin. Biotechnol. 8, 279–284. Chatterjee, A., Das, D., Mandal, B. K., Chowdhury, T. R., Samanta, G., and Chakraborti, D. (1995). Arsenic in ground-water in 6 districts of West-Bengal, India - the biggest arsenic calamity in the world. 1. Arsenic species in drinking-water and urine of the affected people. Analyst 120, 643–650.
68
S. MAHIMAIRAJA ET AL.
Chatterjee, A., and Mukherjee, A. (1999). Hydrogeological investigation of ground water arsenic contamination in South Calcutta. Sci. Total Environ. 225, 249–262. Chattopadhyay, S., Bhaumik, S., Purkayastha, M., Basu, S., Chaudhuri, A. N., and Das Gupta, S. (2002). Apoptosis and necrosis in developing brain cells due to arsenic toxicity and protection with antioxidants. Toxicol. Lett. 136, 65–76. Chau, Y. K., and Wong, P. T. S. (1978). Occurrence of biological methylation of elements in the environment. Am. Chem. Soc. Symp. Ser. 82, 39–53. Chen, F. H., Chen, W. Q., and Dai, S. G. (1994). Toxicities of 4 arsenic species to Scenedesmus obliguus and influence of phosphate on inorganic arsenic toxicities. Toxicol. Environ. Chem. 41, 1–7. Chen, S. L., Dzeng, S. R., Yang, M. H., Chiu, K. H., Shieh, G. M., and Wai, C. M. (1994). Arsenic species in groundwaters of the blackfoot disease area, Taiwan. Environ. Sci. Technol. 28, 877–881. Chen, S. L., and Wilson, D. B. (1997). Genetic engineering of bacteria and their potential for Hg2þ bioremediation. Biodegradation 8, 97–103. Chen, S. L., Yeh, S. J., Yang, M. H., and Lin, T. H. (1995). Trace-element concentration and arsenic speciation in the well water of a Taiwan area with endemic blackfoot disease. Biol. Trace Elem. Res. 48, 263–274. Chen, T. B., Fan, Z. L., Lei, M., Huang, Z. C., and Wei, C. Y. (2002). Effect of phosphorus on arsenic accumulation in As-hyperaccumulator Pteris vittata L. and its implication. Chin. Sci. Bull. 47, 1876–1879. Cheng, C. N., and Focht, D. D. (1979). Production of arsine and methylarsines in soil and in culture. Appl. Environ. Microbiol. 38, 494–498. Chirenje, T., Ma, L. Q., Szulczewski, M., Littell, R., Portier, K. M., and Zillioux, E. (2003). Arsenic distribution in Florida urban soils: Comparison between Gainesville and Miami. J. Environ. Qual. 32, 109–119. Chiu, V. Q., and Hering, J. G. (2000). Arsenic adsorption and oxidation at manganite surfaces. 1. Method for simultaneous determination of adsorbed and dissolved arsenic species. Environ. Sci. Technol. 34, 2029–2034. Christen, K. (2001). The arsenic threat worsens. Environ. Sci. Technol. 35, 286–291. Civantos, D. P., Rodriguez, A. L., and Aguado-Borruey, J. M. (1995). Fuminant malignant arrythmia and multiorgan failure in acute arsenic poisoning. Chest 108, 1774–1775. Clement, W. H., and Faust, S. D. (1981). The release of arsenic from contaminated sediments and muds. J. Environ. Sci. Health A16(1), 87–91. CMPS and F (1995). Audit of timber treatment sites. Draft Environment Site Assessment Report prepared for the Canterbury Regional Council, New Zealand. Cox, D. P., and Alexander, M. (1973). Production of trimethylarsine gas from various arsenic compounds by three sewage fungi. Bull. Environ. Contamin. Toxicol. 9, 84–88. Crecelius, E. A. (1977). Changes in the chemical speciation of arsenic following ingestion by man. Environ. Health Perspect. 19, 147–150. Creger, T. L., and Peryea, F. J. (1994). Phosphate fertilizer enhances arsenic uptake by apricot liners grown in lead-arsenate-enriched soil. Hort. Sci. 29, 88–92. Cullen, N. M., Wolf, L. R., and Stclair, D. (1995). Pediatric arsenic ingestion. Am. J. Emergency Med. 13, 432–435. Cullen, W. R., Li, H., Pergantis, S. A., Eigendorf, G. K., and Harrison, L. G. (1994). The methylation of arsenate by a marine alga Polyphysa peniculus in the presence of L-methionine-methyl-D(3). Chemosphere 28, 1009–1019. Cullen, W. R., McBride, B. C., Pickett, A. W., and Regalinski, J. (1984). The wood preservative chromated copper arsenate is a substrate for trimethylarsine biosynthesis. Appl. Environ. Microbiol. 47, 443–444.
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
69
Cunningham, S. D., Berti, W. R., and Huang, J. M. (1995). Phytoremediation of contaminated soils. TIBTECH 13, 393–397. Cunningham, S. D., and Lee, C. R. (1995). Phytoremediation: Plant-based remediation of contaminated soils and sediments. In “Bioremediation: Science and Applications” (H. D. Skipper and R. F. Turco, Eds.), pp. 145–156. SSSA Special Publication No 43. SSSA, ASA and CSSA, Madison, WI. Das, D., Samanta, G., Mandal, B. K., Chowdhury, T. R., and Chanda, C. R. (1995). A simple household device to remove arsenic from groundwater and two years performance report of arsenic removal plant for treating groundwater with community participation. School of Environmental Studies. Jadavpur University, Calcutta, India. Das, D., Samanta, G., Mandal, B. K., Chowdhury, T. R., Chanda, C. R., Chowdhury, P. P., Basu, G. K., and Chakraborti, D. (1996). Arsenic in groundwater in six districts of West Bengal, India. Environ. Geochem. Health 18, 5–15. Davenport, J. R., and Peryea, F. J. (1991). Phosphate fertilizers influence leaching of lead and arsenic in soil contaminated with lead arsenate. Wat. Air Soil Pollut. 57–58, 101–110. Del Razo, L. M., Rellano, M. A., and Cebrian, M. E. (1990). The oxidation states of arsenic in well-water from a chronic arsenicism area of northern Mexico. Environ. Pollut. 64, 143–153. Department of Health, New Zealand (1992). Public health guidelines for the safe use of sewage effluent and sewage sludge on land. Department of Health, Wellington, New Zealand. Deuel, L. E., and Swoboda, A. R. (1972). Arsenic toxicity to cotton and soybeans. J. Environ. Qual. 1, 317–320. Dhankher, O. P., Li, Y. J., Rosen, B. P., Shi, J., Salt, D., Senecoff, J. F., Sashti, N. A., and Meagher, R. B. (2002). Engineering tolerance and hyperaccumulation of arsenic in plants by combining arsenate reductase and gamma-glutamylcysteine synthetase expression. Nature Biotechnol. 20, 1140–1145. Dikshit, A. K., Pallamreddy, K., Reddy, L. V. P., and Saha, J. C. (2000). Arsenic in groundwater and its sorption by kimberlite tailings. J. Environ. Sci. Health A35, 65–85. Dousova, B., Machovic, V., Kolousek, D., Kovanda, F., and Dornicak, V. (2003). Sorption of As(V) species from aqueous systems. Wat. Air Soil Pollut. 149, 251–267. Dutre, V., Kestens, C., Schaep, J., and Vandecasteele, C. (1998). Study of the remediation of a site contaminated with arsenic. Sci. Total Environ. 220, 185–194. Eisler, R. (2004). Arsenic hazards to humans, plants, and animals from gold mining. Rev. Environ. Contam. Toxicol. 180, 133–165. Elkhatib, E. A., Bennett, O. L., and Wright, R. J. (1984). Arsenite sorption and desorption in soils. Soil Sci. Soc. Am. J. 48, 1025–1030. Esteban, E., Carpena, R. O., and Meharg, A. A. (2003). High-affinity phosphate/arsenate transport in white lupin (Lupinus albus) is relatively insensitive to phosphate status. New Phytol. 158, 165–173. ETCS European Topic Centre Soil (1998). Topic report: Contaminated sites. European Environment Agency. Fendorf, S., Eick, M. J., Grossl, P., and Sparks, D. L. (1997). Arsenate and chromate retention mechanisms on goethite. 1. Surface structure. Environ. Sci. Tech. 31, 315–320. Fennell, J. S., and Stacy, W. K. (1981). Electrocardiographic changes in acute arsenic poisoning. Irish J. Medical Sci. 150, 338–339. Ferguson, J. F., and Gavis, J. (1972). A review of the arsenic cycle in natural waters. Wat. Res. 6, 1259–1274. Fitz, W. J., and Wenzel, W. W. (2002). Arsenic transformations in the soil–rhizosphere–plant system: Fundamentals and potential application to phytoremediation. J. Biotechnol. 99, 259–278.
70
S. MAHIMAIRAJA ET AL.
Folkes, D. J., Kuehster, T. E., and Litle, R. A. (2001). Contribution of pesticide use to urban background concentrations of arsenic in Denver, Colorado, USA. Environ. Forens. 2, 127–139. Foster, A. L., Brown, G. E., Parks, G. A., Tingle, T. N., Voigt, D. E., and Brantley, S. L. (1997). XAFS determination of As(V) associated with Fe(III) oxyhydroxides in weathered mine tailings and contaminated soil from California, USA. J. De Physique IV 7, 815–816. Francesconi, K., Visoottiviseth, P., Sridokchan, W., and Goessler, W. (2002). Arsenic species in an arsenic hyperaccumulating fern, Pityrogramma calomelanos: A potential phytoremediator of arsenic-contaminated soils. Sci. Total Environ. 284, 27–35. Frankenberger, W. T., and Losi, M. E. (1995). Application of bioremediation in the clean up of heavy metals and metalloids. In “Bioremediation: Science and Applications” (H. D. Skipper and R. F. Turco, Eds.), pp. 173–210. SSSA Special Publication No 43. SSSA, ASA and CSSA, Madison, WI. Freeman, M. C. (1985). The reduction of arsenate to arsenite by an Anabaena bacteria assemblage isolated from the Waikato River. N. Z. J. Marine Freshwat. Res. 19, 277–282. Frisbie, S. H., Ortega, R., Maynard, D. M., and Sarkar, B. (2002). The concentrations of arsenic and other toxic elements in Bangladesh’s drinking water. Environ. Health Persp. 110, 1147–1153. Fryxell, G. E., Liu, J., Hauser, T. A., Nie, Z. M., Ferris, K. F., Mattigod, S., Gong, M. L., and Hallen, R. T. (1999). Design and synthesis of selective mesoporous anion traps. Chem. Mater. 11, 2148–2154. Gao, S., and Burau, R. G. (1997). Environmental factors affecting rates of arsine evolution from and mineralization of arsenicals in soil. J. Environ. Qual. 26, 753–763. Garcia-Manyes, S., Jimenez, G., Padro, A., Rubio, R., and Rauret, G. (2002). Arsenic speciation in contaminated soils. Talanta 58, 97–109. Gault, A. J., Polya, D. A., and Lythgoe, P. R. (2003). Seasonal variation of total dissolved arsenic and arsenic speciation in a polluted surface waterway. Environ. Geochem. Health 25, 77–85. Geiszinger, A., Goessler, W., and Kosmus, W. (2002). Organoarsenic compounds in plants and soil on top of an ore vein. Appl. Organometallic Chem. 16, 245–249. Gilmore, J. T., and Wells, B. R. (1980). Residual effects of MSMA on sterility in rice cultivars. Agron. J. 72, 1066–1067. Goering, P. L., Aposhian, H. V., Mass, M. J., Cebrian, M., Beck, B. D., and Waalkes, M. P. (1999). The enigma of arsenic carcinogenesis: Role of metabolism. Toxicol. Sci. 49, 5–14. Goldberg, S., and Glaubig, R. A. (1988). Anion sorption on a calcareous, montmorillonitic soilarsenic. Soil Sci. Soc. Am. J. 2, 1297–1300. Goldberg, S., and Johnston, C. T. (2001). Mechanisms of arsenic adsorption on amorphous oxides evaluated using macroscopic measurements, vibrational spectroscopy, and surface complexation modeling. J. Coll. Interface Sci. 234, 204–216. Gomez–Arroyo, S., Armienta, M. A., Cortes–Eslava, J., and Villalobos-Pietrini, R. (1997). Sister chromatid exchanges in Vicia faba induced by arsenic-contaminated drinking water from Zimapan, Hidalgo, Mexico. Mutat. Res. Gen. Toxicol. Environ. Mutagenesis 394, 1–3. Grafe, M., Eick, M. J., and Grossl, P. R. (2001). Adsorption of arsenate (V) and arsenite (III) on geothite in the presence and absence of dissolved organic carbon. Soil Sci. Soc. Am. J. 65, 1680–1687. Granchinho, S. C. R., Franz, C. M., Polishchuk, E., Cullen, W. R., and Reimer, K. J. (2002). Transformation of arsenic(V) by the fungus Fusarium oxysporum melonis isolated from the alga Fucus gardneri. Appl. Organometallic Chem. 16, 721–726. Granchinho, S. C. R., Polishchuk, E., Cullen, W. R., and Reimer, K. J. (2001). Biomethylation and bioaccumulation of arsenic(V) by marine alga Fucus gardneri. Appl. Organometallic Chem. 15, 553–560.
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
71
Gulledge, J. H., and O’Connor, J. T. (1973). Removal of arsenic (V) from water by adsorption on aluminium and ferric hydroxides. Am. Water Works Assoc. J. 65, 548–552. Guo, X. J., Fujino, Y., Kaneko, S., Wu, K. G., Xia, Y. J., and Yoshimura, T. (2001). Arsenic contamination of groundwater and prevalence of arsenical dermatosis in the Hetao plain area, Inner Mongolia, China. Mol. Cell. Biochem. 222, 137–140. Hall, A. H. (2002). Chronic arsenic poisoning. Toxicol. Lett. 128, 69–72. Hamadeh, H. K., Trouba, K. J., Afshari, C. A., and Germolec, D. (2002). Coordination of altered DNA repair and damage pathways in arsenite-exposed keratinocytes. Toxicol. Sci. 69, 306–316. Han, J. T., and Fyfe, W. S. (2000). Arsenic removal from water by iron-sulphide minerals. Chin. Sci. Bull. 45, 1430–1434. Hara, T., Sonoda, Y., and Iwai, I. (1977). Growth response of cabbage plants to arsenic and antimony under water culture conditions. Soil Sci. Plant Nutr. 23, 253–255. Harvey, C. F., Swartz, C. H., Badruzzaman, A. B. M., Keon-Blute, N., Yu, W., Ali, M. A., Jay, J., Beckie, R., Niedan, V., Brabander, D., Oates, P. M., Ashfaque, K. N., Islam, S., Hemond, H. F., and Ahmed, M. F. (2002). Arsenic mobility and groundwater extraction in Bangladesh. Science 298, 1602–1606. Hasegawa, H., Sohrin, Y., Seki, K., Sato, M., Norisuye, K., Naito, K., and Matsui, M. (2001). Biosynthesis and release of methylarsenic compounds during the growth of freshwater algae. Chemosphere 43, 265–272. Heinrichs, G., and Udluft, P. (1999). Natural arsenic in Triassic rocks: A source of drinkingwater contamination in Bavaria, Germany. Hydrogeol. J. 7, 468–476. Heitkemper, D. T., Vela, N. P., Stewart, K. R., and Westphal, C. S. (2001). Determination of total and speciated arsenic in rice by ion chromatography and inductively coupled plasma mass spectrometry. J. Anal. Atomic Spectromet. 16, 299–306. Hering, J. G., Chen, P.-Y., and Wilkie, J. A. (1997). Arsenic removal from drinking-water by coagulation: The role of adsorption and effects of source water composition. In “Arsenic: Exposure and Health Effects” (C. O. Abernathy, R. L. Calderon, and W. R. Chappell, Eds.), pp. 369–381. Chapman & Hall, New York. Hindmarsh, J. T., and McCurdy, R. F. (1986). Clinical and environmental aspects of arsenic toxicity. CRC Crit. Rev. Clin. Lab. Sci. 23, 315–347. Hingston, J. A., Collins, C. D., Murphy, R. J., and Lester, J. N. (2001). Leaching of chromated copper arsenate wood preservatives: A review. Environ. Pollut. 111, 53–66. Hiroki, M. (1993). Effect of arsenic pollution on soil microbial–population. Soil Sci. Plant Nutr. 39, 227–235. Hlavay, J., and Polyak, K. (1997). Removal of arsenic ions from drinking-water by novel type adsorbents. In “Arsenic : Exposure and Health Effects” (C. O. Abernathy, R. L. Calderon, and W. R. Chappell, Eds.), pp. 382–392. Chapman & Hall, New York. Hopenhayn-Rich, C., Biggs, M. L., Kalman, D. A., Moore, L. E., and Smith, A. H. (1996). Arsenic methylation patterns before and after changing from high to lower concentrations of arsenic in drinking water. Environ. Health Perspect. 104, 1200–1207. Hopenhayn-Rich, C., Biggs, M. L., and Smith, A. H. (1998). Lung and kidney cancer mortality associated with arsenic in drinking water in Cordoba, Argentina. Int. J. Epidemiol. 27, 561–569. http://phys4.harvard.edu/wilson/arsenic Hughes, M. F. (2002). Arsenic toxicity and potential mechanisms of action. Toxicol. Lett. 133, 1–16. Hunt, L. E., and Howard, A. G. (1994). Arsenic speciation and distribution in the Carnon estuary following the acute discharge of contaminated water from a disused mine. Marine Pollut. Bull. 28, 33–38.
72
S. MAHIMAIRAJA ET AL.
Huysmans, K. D., and Frankenberger, W. T., Jr. (1991). Evolution of trimethylarsine by a Penicillium sp. Isolated from agricultural evaporation pond water. Sci. Total Environ. 105, 13–28. IARC. (1980). International Agency for Research on Cancer. Some metals and metallic compounds. Lyon, IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, Vol. 23. Ishinishi, N. (1986). Arsenic. In “Handbook on the Toxicology of Metals” (L. Friberg, Ed.), Vol. II, 2nd edn, pp. 43–73. Elsevier, Amsterdam. Jackson, B. P., and Bertsch, P. M. (2001). Determination of arsenic speciation in poultry wastes by IC–ICP–MS. Environ. Sci. Technol. 35, 4868–4873. Jackson, B. P., and Miller, W. P. (1998). Arsenic and selenium speciation in coal fly ash extracts by ion chromatography-inductively coupled plasma mass spectrometry. J. Anal. At. Spectrom. 13, 1107–1112. Jackson, B. P., and Miller, W. P. (2000). Soil solution chemistry of a fly ash-, poultry litter-, and sewage sludge-amended soil. J. Environ. Qual. 29, 430–436. Jiang, Q. Q., and Singh, B. R. (1994). Effect of different forms and sources of arsenic on crop yield and arsenic concentration. Wat. Air Soil Pollut. 74, 321–343. Jiang, Y. Q., and Ho, X. P. (1983). Pollution of soils by As-containing waste water of chemical factories in Xingping. Environ. Sci (Ch) 4, 48–51. Johnson, D. L. (1972). Bacterial reduction of arsenate in sea water. Nature 240, 44–45. Jones, C. A., Inskeep, W. P., and Neuman, D. R. (1997). Arsenic transport in contaminated mine tailings following liming. J. Environ. Qual. 26, 433–439. Jones, C. A., Langner, H. W., Anderson, K., McDermott, T. R., and Inskeep, W. P. (2000). Rates of microbially mediated arsenate reduction and solubilization. Soil Sci. Soc. Am. J. 64, 600–608. Jonnalagadda, S. B., and Nenzou, G. (1997). Studies on arsenic rich mine dumps. II. The heavy element uptake by vegetation. J. Environ Sci. Health A 32, 455–464. Kadono, T., Inaoka, T., Murayama, N., Ushijima, K., Nagano, M., Nakamura, S., Watanabe, C., Tamaki, K., and Ohtsuka, R. (2002). Skin manifestations of arsenicosis in two villages in Bangladesh. Int. J. Dermatol. 41, 841–846. Kaise, T., Ogura, M., Nozaki, T., Saitoh, K., Sakurai, T., Matsubara, C., Watanabe, C., and Hanaoka, K. (1997). Biomethylation of arsenic in an arsenic-rich freshwater environment. Appl. Organometallic Chem. 11, 297–304. Kaiser, J. (2001). Science only one part of arsenic standard. Science 291, 2533–2534. Karim, M. (2000). Arsenic in groundwater and health problems in Bangladesh. Wat. Res 34, 304–310. Kavanagh, P. J., Farago, M. E., Thornton, I., and Braman, R. S. (1997). Bioavailability of arsenic in soil and mine wastes of the Tamar Valley, SW England. Chem. Spec. Bioavail. 9, 77–81. Keegan, T., Hong, B., Thornton, I., Farago, M., Jakubis, P., Jakubis, M., Pesch, B., Ranft, U., and Nieuwenhuijsen, M. J. (2002). Assessment of environmental arsenic levels in Prievidza district. J. Exposure Anal. Environ. Epidem. 12, 179–185. Khan, A. H., Rasul, S. B., Munir, A. K. M., Habibuddowla, M., Alauddin, M., Newaz, S. S., and Hussam, A. (2000). Appraisal of a simple arsenic removal method for groundwater of Bangladesh. J. Environ. Sci. Health A35, 1021–1041. Kim, M. J., Ahn, K. H., and Jung, Y. J. (2002). Distribution of inorganic arsenic species in mine tailings of abandoned mines from Korea. Chemosphere 49, 307–312. Kim, Y., Kim, C., Choi, I., Rengaraj, S., and Yi, J. (2004). Arsenic removal using mesoporous alumina prepared via a templating method. Environ. Sci. Technol. 38, 924–931. Kimber, S. W. L., Sizemore, D. J., and Slavich, P. E. G. (2002). Is there evidence of arsenic movement at cattle tick dip sites? Aust. J. Soil Res. 40, 1103–1114.
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
73
Knox, A. S., Seaman, J. C., Mench, M. J., and Vangronsveld, J. (2000). Remediation of metal and radionuclides–contaminated soils by in situ stabilization techniques. In “Environmental Restoration of Metals-Contaminated Soils” (I. K. Iskandar, Ed.), pp. 21–60. Lewis, New York. Kocar, B. D., and Inskeep, W. P. (2003). Photochemical oxidation of As(III) in ferrioxalate solutions. Environ. Sci. Technol. 37, 1581–1588. Kondo, H., Ishiguro, Y., Ohno, K., Nagase, M., Toba, M., and Takagi, M. (1999). Naturally occurring arsenic in the groundwaters in the southern region of Fukuoka Prefecture, Japan. Wat. Res. 33, 1967–1972. Kookana, R. S., and Naidu, R. (2000). Capping contaminated sites with clean cover: Assessing the upward migration of contaminants and the thickness of cap. In “Towards Better Management of Wastes and Contaminated Sites in the Australasia–Pacific Region”, Workshop Proceeding, pp. 39–40. Adelaide, South Australia. Korte, N. E., and Fernando, Q. (1991). A review of arsenic (III) in groundwater. Crit. Rev. Environ. Control. 21, 1–39. Kratochvil, D., and Volesky, B. (1998). Advances in the biosorption of heavy metals. Trends Biotechnol. 16, 291–300. Krishna, M. V. B., Chandrasekaran, K., Karunasagar, D., and Arunachalam, J. (2001). A combined treatment approach using Fenton’s reagent and zero valent iron for the removal of arsenic from drinking water. J. Hazard. Mater. 84, 229–240. La Force, M. J., Hansel, C. M., and Fendorf, S. (2000). Arsenic speciation, seasonal transformations, and co-distribution with iron in a mine waste–influenced palustrine emergent wetland. Environ. Sci. Technol. 34, 3937–3943. Lambkin, D. C., and Alloway, B. J. (2003). Arsenate-induced phosphate release from soils and its effect on plant phosphorus. Wat. Air Soil Pollut. 144, 41–56. Langdon, C. J., Meharg, A. A., Feldmann, J., Balgar, T., Charnock, J., Farquhar, M., Piearce, T. G., Semple, K. T., and Cotter-Howells, J. (2002). Arsenic speciation in arsenate-resistant populations of the earthworm, Lumbricus rubellus. J. Environ. Monitoring 4, 603–608. Langdon, C. J., Piearce, T. G., Black, S., and Semple, K. T. (1999). Resistance to arsenictoxicity in a population of the earthworm Lumbricus rubellus. Soil Biol. Biochem. 31, 1963–1967. Langdon, C. J., Piearce, T. G., Meharg, A. A., and Semple, K. T. (2003). Interactions between earthworms and arsenic in the soil environment: A review. Environ. Pollut. 124, 361–373. Larsen, E. H., Hansen, M., and Gossler, W. (1998). Speciation and health risk considerations of arsenic in the edible mushroom Laccaria amethystina collected from contaminated and uncontaminated locations. Appl. Organometallic Chem. 12, 285–291. Lee, Y., Um, I. H., and Yoon, J. (2003). Arsenic(III) oxidation by iron(VI) (ferrate) and subsequent removal of arsenic(V) by iron(III) coagulation. Environ. Sci. Technol. 37, 5750–5756. Lehr, C. R., Polishchuk, E., Delisle, M. C., Franz, C., and Cullen, W. R. (2003). Arsenic methylation by micro-organisms isolated from sheepskin bedding materials. Hum. Exp. Toxicol. 22, 325–334. Leonard, A. (1991). Arsenic. In “Metals and Their Compounds in the Environments: Occurrence, Analysis, and Biological Relevance (E. Merian, Ed.), 2nd edn, pp. 751–773. Weinheim, VCH. Li, W., Chien, P. K., and Furst, A. (1994). Evaluation of 3 antidotes on arsenic toxicity in the common earthworm (Lumbricus–terrestris). J. Appl. Toxicol. 14, 181–183. Liddle, J. R. (1982). “Arsenic and Other Elements of Geothermal Origin in the Taupo Volcanic Zone”. Massey University, New Zealand. Ph.D. Thesis. Lide, D. R. (1992). “CRC Handbook of Chemistry and Physics”, 73rd edn, CRC Press, Boca Raton, FL.
74
S. MAHIMAIRAJA ET AL.
Lin, Z., and Puls, R. W. (2000). Adsorption, desorption and oxidation of arsenic affected by clay minerals and aging process. Environ. Geol. 39, 753–759. Liu, T., Liu, J., LeCluyse, E. L., Zhou, Y. S., Cheng, M. L., and Waalkes, M. P. (2001). Application of cDNA microarray to the study of arsenic-induced liver diseases in the population of Guizhou, China. Toxicol. Sci. 59, 185–192. Livesey, N. T., and Huang, P. M. (1981). Adsorption of arsenate by soils and its relations to selected chemical properties and anions. Soil Sci. 131, 88–94. Loeppert, R. H. (1997). Arsenate, arsenite retention and release in oxide and sulfide dominated systems. Technical Report No. 176, Texas Water Resources Institute, College Station, TX. Loukidou, M. X., Matis, K. A., Zouboulis, A. I., and Kyriakidou, M. L. (2003). Removal of As (V) from wastewaters by chemically modified fungal biomass. Wat. Res. 37, 4544–4552. Lu, F. J. (1990). Blackfoot disease: Arsenic or humic acid? Lancet 336, 115–119. Ma, L. Q., Komar, K. M., Tu, C., Zhang, W. H., Cai, Y., and Kennelley, E. D. (2001). A fern that hyperaccumulates arsenic: A hardy, versatile, fast-growing plant helps to remove arsenic from contaminated soils. Nature 409, 579. Ma, W. C. (1982). The influence of soil properties and worm–related factors on the concentration of heavy metals in earthworms. Pedobiologia 24, 109–119. Maeda, S. (1994). In “Arsenic in the Environment, Part 1: Cycling and Characterization” (J. O. Nriagu, Ed.). Wiley, New York. Maeda, S., Inoue, R., Kozono, T., Tokuda, T., Ohki, A., and Takeshita, T. (1990). Arsenic metabolism in a freshwater food chain. Chemosphere 20, 101–110. Maeda, S., Nakashima, S., Takeshita, T., and Higashi, S. (1985). Bioaccumulation of arsenic by freshwater algae and the application to the removal of inorganic arsenic from an aqueous phase. 2. By Chlorella vulgaris isolated from arsenic-polluted environment. Separation Sci. Technol. 20, 153–161. Magalhaes, V. F., Carvalho, C. E. V., and Pfeiffer, W. C. (2001). Arsenic contamination and dispersion in the Engenho Inlet, Sepetiba Bay, SE, Brazil. Wat. Air Soil Pollut. 129, 83–90. Mahata, J., Basu, A., Ghoshal, S., Sarkar, J. N., Roy, A. K., Poddar, G., Nandy, A. K., Banerjee, A., Ray, K., Natarajan, A. T., Nilsson, R., and Giri, A. K. (2003). Chromosomal aberrations and sister chromatid exchanges in individuals exposed to arsenic through drinking water in West Bengal, India. Mutat. Res. Gen. Toxicol. Environ. Mutagen. 534, 133–143. Maliszewska, W., Dec, S., Wierzbicka, H., and Wozniakowska, A. (1985). The influence of various heavy metal compounds on the development and activity of soil microorganisms. Environ. Pollut. Ser. A 37, 195–215. Mamtaz, R., and Bache, D. H. (2000). Low-cost separation of arsenic from water: With special reference to Bangladesh. J. Chartered Inst. Water Environ. Mgt. 14, 260–269. Mandal, B. K., Chowdhury, T. R., Samanta, G., Basu, G. K., Choudhury, P. P., Chanda, C. R., Lodh, D., Karan, N. K., Dhar, R. K., Tamili, D. K., Das, D., Saha, K. C., and Chakraborti, D. (1996). Arsenic in groundwater in seven districts of West Bengal, India: The biggest arsenic calamity in the world. Curr. Sci. 70, 976–986. Manful, G. A., Verloo, M., and De Spiegeleer, F. (1989). Arsenate sorption by soils in relation to pH and selected anions. Pedologie 55–68. Manning, B. A., and Goldberg, S. (1997). Adsorption and stability of arsenic (III) at the clay mineral–water interface. Environ. Sci. Technol. 31, 2005–2011. Mappes, R. (1977). Experiments on excretion of arsenic in urine. Int. Arch. Occup. Environ. Health 40, 267–272. Marin, A. R., Masscheleyn, P. H., and Patrick, W. H., Jr. (1992). The influence of chemical form and concentration of arsenic on rice growth and tissue arsenic concentration. Plant Soil 139, 175–183.
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
75
Marin, A. R., Masscheleyn, P. H., and Patrick, W. H., Jr. (1993). Soil redox – pH stability of arsenic species and its influence on arsenic uptake by rice. Plant Soil 152, 245–253. Marques, I. A., and Anderson, L. E. (1986). Effects of arsenite, sulfite, and sulfate on photosynthetic carbon metabolism in isolated pea (Pisum–sativum–L, Cv Little Marvel) chloroplasts. Plant Physiol. 82, 488–493. Masscheleyn, P. H., Delaune, R. D., and Patrick, W. H., Jr. (1991). Effect of redox potential and pH on arsenic speciation and solubility in a contaminated soil. Environ. Sci. Technol. 25, 1414–1418. Matera, V., and Le Hecho, I. (2001). Arsenic behaviour in contaminated soils: Mobility and speciation. In “Heavy Metals Release in Soils” (H. M. Selim and D. L. Sparks, Eds.), pp. 207–235. Lewis Publishers, Washington, DC. Matera, V., Le Hecho, I., Laboudigue, A., Thomas, P., Tellier, S., and Astruc, M. (2003). A methodological approach for the identification of arsenic bearing phases in polluted soils. Environ. Pollut. 126, 511–564. Matschullat, J. (2000). Arsenic in the geosphere: A review. Sci. Total Environ. 249, 297–312. Mattigod, S. V., and Sposito, G. (1979). GEOCHEM: A computer program for the calculation of chemical equilibria in soils solutions and other water systems. Department of Soils and Environmental Sciences, University of California, Riverside, CA. Mazumder, D. N. G. (2003). Chronic arsenic toxicity: Clinical features, epidemiology, and treatment: Experience in West Bengal. J. Environ. Sci. Health. A 38, 141–163. Mazumder, D. N. G., Chakraborty, A. K., Ghose, A., Gupta, J. D., Chakraborty, D. P., Dey, S. B., and Chattopadhyay, N. (1988). Chronic arsenic toxicity from drinking tubewell water in rural West Bengal. Bull. World Health Organ. 66, 499–506. McArthur, J. M., Ravenscroft, P., Safiullah, S., and Thirlwall, M. F. (2000). Arsenic in groundwater: Testing pollution mechanisms for sedimentary aquifers in Bangladesh. Water Resour. Res. 37, 109–117. McBride, B. C., and Wolfe, R. S. (1971). Biosynthesis of dimethylarsine by Methanobacterium. Biochemistry 10, 4312–4317. McBride, G. L., Holland, P., and Wilson, K. (1998). Investigation of contaminated sheep dipping sites in the Waikato. In “Proceedings of Waste Management Institute New Zealand Conference”, pp. 129–137. Waste Management Institute, New Zealand Inc., Rotorua. McLaughlin, M. J., Tiller, K. G., Naidu, R., and Stevens, D. P. (1996). The behaviour and environmental impact of contaminants in fertilizers. Aust. J. Soil Res. 34, 1–54. McLaren, P. (1992). Soil contamination at CCA timber treatment plants: An initial appraisal in the Bay of Plenty: Technical Report No 27: Bay of Plenty Regional Council. McLaren, R. G., Carey, P. L., Cameron, K. C., Adams, J. A., and Sedcole, J. R. (1994). Effect of soil properties and contact period on the leaching of copper, chromium, and arsenic through undisturbed soils. In “15th World Cong. Soil Sci.”, Vol. 3a, pp. 156–169. ISSS and the Mexican Soc. Soil Sci., Acapulco, Mexico. McLaren, R. G., Naidu, R., Smith, J., and Tiller, K. G. (1998). Fractionation and distribution of arsenic in soils contaminated by cattle dip. J. Environ. Qual. 27, 348–354. Merry, R. H., Tiller, K. G., and Alston, A. M. (1983). Accumulation of copper, lead, and arsenic in some Australian orchard soils. Aust. J. Soil Res. 21, 549–561. Minato, H., Morimoto, T., and Wada, N. (2000). Mineral techniques for the remediation of contaminated water and geology. In “Proceedings of the Workshop on Towards Better Management of Wastes and Contaminated Sites in the Australasia–Pacific Region”, pp. 128–129. Adelaide, South Australia. Mitchell, P., and Barr, D. (1995). The nature and significance of public exposure to arsenic: A review of its relevance to southwest England. Environ. Geochem. Health 17, 57–82. Miteva, E. (2002). Accumulation and effect of arsenic on tomatoes. Commun. Soil Sci. Plant Anal. 33, 1917–1926.
76
S. MAHIMAIRAJA ET AL.
Moore, J. N., Ficklin, W. H., and Johns, C. (1988). Partitioning of arsenic and metals in reducing sulfidic sediments. Environ. Sci. Technol. 22, 432–437. Moore, J. N., O’Callaghan, C. A., and Berylne, G. (1994). Acute arsenic poisoning: Absence of polyneuropathy after treatment with 2,3-dimercaptopropanesulphonate (DMPS). J. Neurol. Neurosurg. Psychiatry 57, 1133–1135. Mukherjee, S. C., Rahman, M. M., Chowdhury, U. K., Sengupta, M. K., Lodh, D., Chanda, C. R., Saha, K. C., and Chakraborti, D. (2003). Neuropathy in arsenic toxicity from groundwater arsenic contamination in West Bengal, India. J. Environ. Sci. Health. A 38, 165–183. Munoz, O., Velez, D., Montoro, R., Arroyo, A., and Zamorano, M. (2000). Determination of inorganic arsenic [As (III) plus As(V)] in water samples by microwave assisted distillation and hydride generation atomic absorption spectrometry. J. Anal. Atomic Spectr. 15, 711–714. Murphy, M. J., Lyon, L. W., and Taylor, J. W. (1981). Subacute arsenic neuropathy: Clinical and electrophysiological observations. J. Neurol., Neurosurg. Psychiatry 44, 896–900. Myers, D. J., Heimbrook, M. E., Osteryoung, J., and Morrison, S. M. (1973). Arsenic oxidation state in the presence of micro-organism by differential pulse polarography. Environ. Lett. 5, 53–61. Nag, J. K., Balaram, V., Rubio, R., Alberti, J., and Das, A. K. (1996). Inorganic arsenic species in groundwater: A case study from Purbasthali (Burdwan), India. J. Trace Elements Med. Biol. 10, 20–24. Naidu, R., Bolan, N. S., and Owens, G. (2003). Risk based land management: A cost effective tool for contaminated land management. In “Environmental Management using Soil–Plant Systems (L. D. Currie, R. B. Stewart, and C. W. N. Anderson, Eds.), pp. 5–19. Occasional Report No. 16. Fertilizer and Lime Research Centre, Massey University, Palmerston North. Naidu, R., Smith, E., Smith, J., and Kookana, R. S. (2000). Is there potential for using strongly weathered oxidic soils as reactive landfill barriers? In “Proceedings of Workshop on Towards Better Management of Wastes and Contaminated Sites in the Australasia–Pacific Region”, pp. 130–133. Adelaide, South Australia. Naranjo-Pulido, A., Romero-Schmidt, H., Mendez-Rodriguez, L., Acosta-Vargas, B., and Ortega-Rubio, A. (2002). Soil arsenic contamination in the Cape Region, BCS, Mexico. J. Environ. Biol. 23, 347–352. Neku, A., and Tandukar, N. (2003). An overview of arsenic contamination in groundwater of Nepal and its removal at household level. J. De Physique 107, 941–944. Newman, D. K., Kennedy, E. K., Coates, J. D., Ahmann, D., Ellis, D. J., Lovley, D. R., and Morel, F. M. M. (1997). Dissimilatory arsenate and sulfate reduction in Desulfotomaculum auripigmentum sp. Arch. Microbiol. 168, 380–388. Nicholson, F. A., Chambers, B. J., Williams, J. R., and Unwin, R. J. (1999). Heavy metal contents of livestock feeds and animal manures in England and Wales. Biores. Technol. 23, 23–31. Nickson, R. T., McArthur, J. M., Ravenscroft, P., Burgess, W. G., and Ahmed, K. M. (2000). Mechanisms of arsenic release to groundwater, Bangladesh and West Bengal. Appl. Geochem. 15, 403–413. Nimick, D. A. (1998). Arsenic hydrogeochemistry in an irrigated river valley: A reevaluation. Ground Water 36, 743–753. NRC (1997). National Research Council. Arsenic.National Academy Press, Washington, DC. Nriagu, J. O., and Pacyna, J. M. (1988). Quantitative assessment of worldwide contamination of air, water, and soils by trace metals. Nature 333, 134–139. O’Neill, P. (1995). Arsenic. In “Heavy Metals in Soil” (B. J. Alloway, Ed.). Blackie Academic & Professional, Glasgow.
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
77
Ongley, L. K., Armienta, A., and Mango, H. (2003). Concentrations of heavy metals in soil, Zimapan, Mexico. J. De Physique IV, 107, 983–986. Onken, B. M., and Hossner, L. R. (1995). Plant uptake and determination of arsenic species in soil solution under flooded conditions. J. Environ. Qual. 24, 373–381. Oremland, R. S., Dowdle, P. R., Hoeft, S., Sharp, J. O., Schaefer, J. K., Miller, L. G., Blum, J. S., Smith, R. L., Bloom, N. S., and Wallschlaeger, D. (2000). Bacterial dissimilatory reduction of arsenate and sulfate in meromictic Mono Lake, California. Geochim. Cosmochim. Acta 64, 3073–3084. Osborne, F. H., and Ehrlich, H. L. (1976). Oxidation of arsenite by a soil isolate of alcaligenes. J. Appl. Bacteriol. 41, 295–305. Oscarson, D. W., Huang, P. M., Defosse, C., and Herbillon, (1981). Oxidative power of Mn(IV) and Fe(III) oxides with respect to As(III) in terrestrial and aquatic environments. Nature 291, 50–51. Outridge, P. M., and Noller, B. N. (1991). Accumulation of toxic trace elements by fresh-water vascular plants. Rev. Environ. Contam. Toxicol. 121, 1–63. Page, A. L., Elseewi, A. A., and Straughan, I. R. (1979). Physical and chemical properties of fly ash from coal fire power plants with reference to environmental impacts. Residue Rev. 71, 83–120. Paivoke, A. (1979). Effects of lead and arsenate on the growth and acid-phosphatase activity of pea-seedlings. Ann. Bot. Fenn. 16, 18–27. Papassiopi, N., Vircikova, E., Nenov, V., Kontopoulos, A., and Molnar, L. (1996). Removal and fixation of arsenic in the form of ferric arsenates: Three parallel experimental studies. Hydrometallurgy 41, 243–253. Penrose, W. R. (1974). Arsenic in the marine and aquatic environments: Analysis, occurrence and significance. CRC Crit. Rev. Environ. Control 4, 465–482. Peryea, F. J. (1991). Phosphate-induced release of arsenic from soils contaminated with lead arsenate. Soil Sci. Soc. Am. J. 55, 1301–1306. Peryea, F. J., and Creger, T. L. (1994). Vertical distribution of lead and arsenic in soils contaminated with lead arsenate pesticide residues. Wat. Air Soil Pollut. 78, 297–306. Peryea, F. J., and Kammereck, R. (1997). Phosphate-enhanced movement of arsenic out of lead arsenate-contaminated topsoil and through uncontaminated subsoil. Wat. Air Soil Polutl. 93, 243–254. Peters, S. C., Blum, J. D., Klaue, B., and Karagas, M. R. (1999). Arsenic occurrence in New Hampshire drinking water. Environ. Sci Technol. 33, 1328–1333. Peterson, P. J., Benson, L. M., and Zieve, R. (1981). Metalloids. Sec¼3-Arsenic. In “Effect of Heavy Metal Pollution on Plants” (N. W. Lepp, Ed.), pp. 299–322. Applied Science Publishers, London. Phillips, S. E., and Taylor, M. L. (1976). Oxidation of arsenite to arsenate by Alcaligenes faecalis. Appl. Environ. Microbiol. 32, 392–399. Pickering, I. J., Prince, R. C., George, M. J., Smith, R. D., George, G. N., and Salt, D. E. (2000). Reduction and coordination of arsenic in Indian mustard. Plant Physiol. 122, 1171–1177. Pitten, F. A., Muller, G., Konig, P., Schmidt, D., Thurow, K., and Kramer, A. (1999). Risk assessment of a former military base contaminated with organoarsenic-based warfare agents: Uptake of arsenic by terrestrial plants. Sci. Total Environ. 226, 237–245. Porter, E. K., and Peterson, P. J. (1975). Arsenic accumulation by plants on mine waste (United Kingdom). Sci. Total Environ. 4, 365–371. Qafoku, N. P., Kulier, U., Summer, M. E., and Radcliffe, D. E. (1999). Arsenate displacement from fly ash in amended soils. Wat. Air Soil Pollut. 114, 185–198. Quinaia, S. P., and Rollember, M. D. E. (2001). Selective reduction of arsenic species by hydride generation atomic absorption spectrometry. Part 2: Sample storage and arsenic determination in natural waters. J. Braz. Chem. Soc. 112, 37–41.
78
S. MAHIMAIRAJA ET AL.
Rahman, M., Tondel, M., Ahmad, S. A., Chowdhury, I. A., Faruquee, M. H., and Axelson, O. (1999). Hypertension and arsenic exposure in Bangladesh. Hypertension 33, 74–78. Rahman, M. M., Chowdhury, U. K., Mukherjee, S. C., Mondal, B. K., Paul, K., Lodh, D., Biswas, B. K., Chanda, C. R., Basu, G. K., Saha, K. C., Roy, S., Das, R., Palit, S. K., Quamruzzaman, Q., and Chakraborti, D. (2001). Chronic arsenic toxicity in Bangladesh and West Bengal, India: A review and commentary. J. Toxicol. Clin. Toxicol. 39, 683–700. Raichur, A. M., and Panvekar, V. (2002). Removal of As(V) by adsorption onto mixed rare earth oxides. Separation Sci. Technol. 37, 1095–1108. Raskin, I., Smith, R. D., and Salt, D. E. (1997). Phytoremediation of metals: Using plants to remove pollutants from the environment. Curr. Opinion Biotech. 8, 221–226. Raven, K. P., Jain, A., and Loeppert, R. H. (1998). Arsenite and arsenate adsorption on ferrihydrite: Kinetics, equilibrium, and adsorption envelopes. Environ. Sci. Technol. 32, 344–349. Raven, K. P., and Loeppert, R. H. (1997). Trace element composition of fertilizers and soil amendments. J. Environ. Qual. 26, 551–557. Reay, P. F. (1972). The accumulation of arsenic from arsenic-rich natural waters by aquatic plants. J. Appl. Ecol. 9, 557–565. Reed, J. F., and Sturgis, M. B. (1963). Toxicity from arsenic compounds to rice on flooded soils. J. Am. Soc. Agron. 28, 432–436. Reynolds, J. G., Naylor, D. V., and Fendorf, S. E. (1999). Arsenic sorption in phosphateamended soils during flooding and subsequent aeration. Soil Sci. Soc. Am. J. 63, 1149–1156. Robertson, F. N. (1989). Arsenic in groundwater under oxidizing conditions, southwest United States. Environ. Geochem. Health 11, 171–185. Robinson, B., Marchetti, M., Moni, C., Schroeter, L., van den Dijssel, C., Milne, G., Bolan, N. S., and Mahimairaja, S. (2004). Arsenic accumulation by aquatic and terrestrial plants in the Taupo Volcanic Zone (TVZ) of New Zealand. In “From Hidden Horror to Human Health: Managing Arsenic in the Environment” (R. Naidu, E. Smith, L. Smith, and P. Bhattacharya, Eds.). Academic Press (In press). Robinson, B. H., Brooks, R. R., Outred, H. A., and Kirkman, J. H. (1995). The distribution and fate of arsenic in the Waikato River System, North Island, New Zealand. Chem. Spec. Bioavail. 7, 89–96. Robinson, B. H., Duwig, C., Bolan, N. S., Kannathasan, M., and Saravanan, A. (2003a). Uptake of arsenic by New Zealand watercress (Lepidium sativum). Sci. Total Environ. 301, 67–73. Robinson, B. H., Green, S., and Mills, T. (2003b). Assessment of phytoremediation as best management practice for degraded environments. In “Environmental Management using Soil–Plant Systems” (L. D Currie, R. B Stewart, and C. W. N Anderson, Eds.), pp. 39–49. Fertilizer and Lime Research Centre, Massey University, Palmerston North, New Zealand. Occasional Report No. 16. Rosehart, R. G., and Lee, J. Y. (1973). The effect of arsenic trioxide on the growth of white spruce seedlings. Wat. Air Soil Pollut. 2, 439–443. Ross, A. D., Lawrie, R. A., Whatmuff, M. S., Keneally, J. P., and Awad, A. S. (1991). Guidelines for the use of Sewage Sludge on Agricultural Land. NSW Agriculture, Sydney, Australia. Rothbaum, H. P., and Buisson, D. H. (1977). Removal and potential recovery of arsenic from geothermal discharge waters after adsorption on iron-floc. Department of Scientific and Industrial Research, Report No. C.D. 2252, New Zealand. Roussel, C., Bril, H., and Fernandez, A. (2000). Arsenic speciation: Involvement in evaluation of environmental impact caused by mine wastes. J. Environ. Qual. 29, 182–188.
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
79
Roychowdhury, T., Uchino, T., Tokunaga, H., and Ando, M. (2002). Arsenic and other heavy metals in soils from an arsenic-affected area of West Bengal, India. Chemosphere 49, 605–618. Ruokolainen, M., Pantsar-Kallio, M., Haapa, A., and Kairesalo, T. (2000). Leaching, runoff and speciation of arsenic in a laboratory mesocosm. Sci. Total Environ. 258, 139–147. Sadler, R., Olszowy, H., Shaw, G., Biltoft, R., and Connell, D. (1994). Soil and water contamination by arsenic from tannery wastes. Wat. Air Soil Pollut. 78, 189–198. Saha, K. C. (2003). Review of arsenicosis in West Bengal, India: A clinical perspective. Crit. Rev. Environ. Sci. Technol. 33, 127–163. Salido, A. L., Hasty, K. L., Lim, J. M., and Butcher, D. J. (2003). Phytoremediation of arsenic and lead in contaminated soil using Chinese Brake Ferns (Pteris vittata) and Indian mustard (Brassica juncea). Int. J. Phytoremediat. 5, 89–103. Samanta, G., Chowdhury, T. R., Mandal, B. K., Biswas, B. K., Chowdhury, U. K., Basu, G. K., Chanda, C. R., Lodh, D., and Chakraborti, D. (1999). Flow injection hydride generation atomic absorption spectrometry for determination of arsenic in water and biological samples from arsenic-affected districts of West Bengal, India, and Bangladesh. Microchemical J. 62, 174–191. Sanders, J. G. (1983). Role of marine phytoplankton in determining the chemical speciation and biogeochemical cycling of arsenic. Can. J. Fish. Aquat. Sci. 40, 192–196. Sanders, J. G., and Windom, H. L. (1980). The uptake and reduction of arsenic species by marine algae. Estuar. Coast. Mar. Sci. 10, 555–567. Santra, A., Maiti, A., Das, S., Lahiri, S., Charkaborty, S. K., and Mazumder, D. N. G. (2000). Hepatic damage caused by chronic arsenic toxicity in experimental animals. J. Toxicol. Clin. Toxicol. 38, 395–405. Sbarato, V. M., and Sanchez, H. J. (2001). Analysis of arsenic pollution in groundwater aquifers by X-ray fluorescence. Appl. Rad. Isotopes 54, 737–740. Schreiber, M. E., Simo, J. A., and Freiberg, P. G. (2000). Stratigraphic and geochemical controls on naturally occurring arsenic in groundwater, eastern Wisconsin, USA. Hydrogeol. J. 8, 161–176. Seidel, A., Waypa, J. J., and Elimelech, M. (2001). Role of charge (Donnan) exclusion in removal of arsenic from water by a negatively charged porous nanofiltration membrane. Environ. Eng. Sci. 18, 105–113. Seyler, P., and Martin, J. M. (1989). Biogeochemical processes affecting arsenic species distribution in a permanently stratified lake. Environ. Sci. Technol. 23, 1258–1263. Shen, Y. S. (1973). Study of arsenic removal from drinking water. Am. Wat. Works Assoc. J. 65, 543–548. Sheppard, S. C. (1992). Summary of phytotoxic levels of soil arsenic. Wat. Air Soil Pollut. 64, 539–550. Shrestha, R. R., Shrestha, M. P., Upadhyay, N. P., Pradhan, R., Khadka, R., Maskey, A., Maharjan, M., Tuladhar, S., Dahal, B. M., and Shrestha, K. (2003). Groundwater arsenic contamination, its health impact and mitigation program in Nepal. J. Environ. Sci. Health A 38, 185–200. Smedley, P. L., and Kinniburgh, D. G. (2002). A review of the sources, behaviour and distribution of arsenic in natural waters. Appl. Geochem. 17, 517–568. Smedley, P. L., Nicolli, H. B., Macdonald, D. M. J., Barros, A. J., and Tullio, J. O. (2002). Hydrogeochemistry of arsenic and other inorganic constituents in groundwaters from La Pampa, Argentina. Appl. Geochem. 17, 259–284. Smith, A. H., Goycolea, M., Haque, R., and Biggs, M. L. (1998). Marked increase in bladder and lung cancer mortality in a region of northern Chile due to arsenic in drinking water. Am. J. Epidemiol. 147, 660–669.
80
S. MAHIMAIRAJA ET AL.
Smith, A. H., Lingas, O. E., and Rahman, M. (2000). Contamination of drinking water by arsenic in Bangladesh: A public health emergency. Bull. World Health Organ. 78, 1093–1103. Smith, D. G. (1986). Heavy metals in the New Zealand aquatic environment: A review. National Water and Soil Conservation Authority, Wellington, New Zealand. Water and Soil Miscellaneous Publication No. 100. Smith, E., Naidu, R., and Alston, A. M. (1998). Arsenic in the soil environment. A review. Adv. Agron. 64, 149–195. Sparks, D. L. (1995). “Environmental Soil Chemistry”. Academic Press, San Diego, CA. Steevens, D. R., Walsh, L. M., and Keeney, D. R. (1972). Arsenic phytotoxicity on a Plainfield sand as affected by ferric sulfate or aluminium sulfate. J. Environ. Qual. 1, 301–303. Stefanakis, M., and Kontopoulos, A. (1988). Production of environmentally acceptable arsenites-arsenates from solid arsenic trioxide. In “Arsenic Metallurgy, Fundamentals and Applications” (Reddy, Hendrix, and Queneau, Eds.). Metallurgy Society Inc. Tabatabai, M. A. (1977). Effects of trace elements on urease activity in soils. Soil Biol. Biochem. 9, 9–13. Tamaki, S., and Frankenberger, W. T., Jr. (1992). Environmental biochemistry of arsenic. Rev. Environ. Contam. Toxicol. 124, 79–110. Taylor, G. F. (1996). Exploration, mining and mineral processing. In “Contaminants and the Soil Environment in the Australasia–Pacific Region” (R. Naidu, R. S. Kookana, D. P. Oliver, S. Rogers, and M. J. McLaughlin, Eds.), pp. 213–266. Kluwer Academic Publishers, London. Thanabalasingam, P., and Pickering, W. F. (1986). Arsenic sorption by humic acids. Environ. Pollut. 12, 233–246. Thirunavukkarasu, O. S., Viraraghavan, T., and Subramanian, K. S. (2001). Removal of arsenic in drinking water by iron oxide-coated sand and ferrihydrite – Batch studies. Wat. Qual. Res. J. Can. 36, 55–70. Tokunaga, S., and Hakuta, T. (2002). Acid washing and stabilization of an artificial arseniccontaminated soil. Chemosphere 46, 31–38. Tokunaga, H., Roychowdhury, T., Chandraskaran, N., Uchino, T., and Ando, M. (2002). Urinary arsenic species in an arsenic-affected area of West Bengal, India. Appl. Organometal. Chem. 16, 406–414. Toth, J., and Hruskovicova, A. (1977). Effect of increasing rates of arsenic on the yield and distribution of arsenic in cereals. Pol’nohospodarstvo 23, 301–318. Tu, C., and Ma, L. Q. (2002). Effects of arsenic concentrations and forms on arsenic uptake by the hyperaccumulator ladder brake. J. Environ. Qual. 31, 641–647. Tu, C., and Ma, L. Q. (2003). Effects of arsenate and phosphate on their accumulation by an arsenic-hyperaccumulator Pteris vittata L. Plant Soil 249, 373–382. Tu, C., Ma, L. Q., and Bondada, B. (2002). Arsenic accumulation in the hyperaccumulator Chinese brake and its utilization potential for phytoremediation. J. Environ. Qual. 31, 1671–1675. Tu, C., Ma, L. Q., Zhang, W. H., Cai, Y., and Harris, W. G. (2003). Arsenic species and leachability in the fronds of the hyperaccumulator Chinese brake (Pteris vittata L.). Environ. Pollut. 124, 223–230. Tuin, B. J. W., and Tels, M. (1991). Continuous treatment of heavy metal contaminated clay soils by extraction in stirred tanks and counter current column. Environ. Technol. 12, 178–190. Tyler, G., and Olsson, T. (2001). Plant uptake of major and minor mineral elements as influenced by soil acidity and liming. Plant Soil 230, 307–321. USDHHS, United States Department of Health and Human Services (2000). Toxicological Profile for Arsenic. Agency for Toxic Substances and Disease Registry, Division of Toxicology/Toxicology Information Branch, Atlanta, GA.
ARSENIC CONTAMINATION AND ITS RISK MANAGEMENT
81
USEPA, United States Environmental Protection Agency. (1988). Special Report on Ingested Inorganic Arsenic. Skin Cancer; Nutritional Essentiality. EPA /625 /3- 87 / 013, Washington, DC. USEPA, United States Environmental Protection Agency. (1993). Integrated Risk Information System on Arsenic. Office Health Environ Assess, Cincinnati, OH. USEPA, United States Environmental Protection Agency. (1997). Recent development for in-situ treatment of metal contaminated soils. Office of Solid Waste and Emergency Response, EPA-542-R-97-004, p8. Van den Broeck, K., Vandecasteele, C., and Geuns, J. M. C. (1997). Determination of arsenic by inductively coupled plasma mass spectrometry in Mung bean seedlings for use as a bioindicator of arsenic contamination. J. Anal. Atomic Spect. 12, 987–991. Van den Broeck, K., Vandecasteele, C., and Geuns, J. M. C. (1998). Speciation by liquid chromatography-inductively coupled plasma-mass spectrometry of arsenic in mung bean seedlings used as a bioindicator for the arsenic contamination. Anal. Chim. Acta 361, 101–111. Van Zwieten, L., Ayres, M. R., and Morris, S. G. (2003). Influence of arsenic co-contamination on DDT breakdown and microbial activity. Environ. Pollut. 124, 331–339. Vaughan, G. T. (1993). The environmental chemistry and fate of arsenical pesticides in cattle tick dip sites and banana plantations. CSIRO, Division of Coal and Energy Technology Centre for Advanced Chemistry. Investigation Report CET / LHIRI 148, Sydney, Australia. Visoottiviseth, P., Francesconi, K., and Sridokchan, W. (2002). The potential of Thai indigenous plant species for the phytoremediation of arsenic contaminated land. Environ. Pollut. 118, 453–461. Vrijenhoek, E. M., and Waypa, J. J. (2000). Arsenic removal from drinking water by a “loose” nanofiltration membrane. Desalination 130, 265–277. Wai, C. M., Wang, J. S., and Yang, M. H. (2003). Arsenic contamination of groundwater, black foot disease, and other related health problems. Biogeochem. Environ. Imp. Trace Ele. ACS Symp. Ser. 835, 210–231. Wakao, N., Koyatsu, H., Komai, Y., Shimokawara, H., Sakurai, Y., and Shiota, H. (1988). Microbial oxidation of arsenite and occurrence of arsenite-oxidizing bacteria in acid-mine water from a sulfur-pyrite mine. Geomicrobiol. J. 6, 11–24. Walsh, L. M., and Keeney, D. R. (1975). Behaviour and phytotoxicity of inorganic arseninicals in soils. In “Arsenical Pesticides” (E. A. Woolson, Ed.). pp. 35–53. Am. Chem. Soc. Washington, DC. Walsh, P. R., Duce, R. A., and Fasching, J. L. (1979). Consideration of the enrichment, sources, and flux of arsenic in the troposphere. J. Geophys. Res. 84, 1719–1723. Wang, C. T., Shih, T. P., Huang, H. K., Tsai, Y. Y., Wang, R. T., and Huang, C. W. (1997). Arsenic and iron in the finger and foot muscle from blackfoot disease patients. Anal. Sci. 13, 493–496. Wang, G. (1984). Arsenic poisoning from drinking water in Xinjiang. Chin. J. Prevent. Med. 18, 105–107. Wang, J. R., Zhao, F. J., Meharg, A. A., Raab, A., Feldmann, J., and McGrath, S. P. (2002). Mechanisms of arsenic hyperaccumulation in Pteris vittata: Uptake kinetics, interactions with phosphate, and arsenic speciation. Plant Physiol. 130, 1552–1561. Watanabe, C., Inaoka, T., Kadono, T., Nagano, M., Nakamura, S., Ushijima, K., Murayama, N., Miyazaki, K., and Ohtsuka, R. (2001). Males in rural Bangladeshi communities are more susceptible to chronic arsenic poisoning than females: Analyses based on urinary arsenic. Environ. Health Perspect. 109, 1265–1270. Wellman, D. E., Reid, D. A., and Ulery, A. L. (1999). Elevated arsenic levels at a former crude oil storage facility-assessment, remediation, and possible sources. J. Soil Contam. 8, 329–341.
82
S. MAHIMAIRAJA ET AL.
Wenzel, W. W., Kirchbaumer, N., Prohaska, T., Stingeder, G., Lombi, E., and Adriano, D. C. (2001). Arsenic fractionation in soils using an improved sequential extraction procedure. Anal. Chim. Acta 436, 309–323. Watt, C., and Le, X. C. (2003). Arsenic speciation in natural waters. Biogeochem. Environ. Imp. Trace Elem. ACS Symp. Ser. 835, 11–32. WHO, World Health Organisation. (1981). Environmental Health Criteria. 18: Arsenic, World Health Organisation, Geneva. Wilkie, J. A., and Hering, J. G. (1996). Adsorption of arsenic onto hydrous ferric oxide: Effects of adsorbate/adsorbent ratios and co-occurring solutes. Colloids Surfaces A Physicochem. Engineer. Aspects 107, 97–110. Williams, M., Fordyce, F., Paijitprapapon, A., and Charoenchaisri, P. (1996). Arsenic contamination in surface drainage and groundwater in part of the southeast Asian tin belt, Nakhon Si Thammarat Province, southern Thailand. Environ. Geol. 27, 16–33. Woolson, E. (1983). Emissions, Cycling and effects of arsenic in soil ecosystems. In “Biological and Environmental Effects of Arsenic” (B. Fowler, Ed.). Elsevier Science, Amsterdam. Woolson, E. A. (1977). Generation of alkylarsines from soil. Weed Sci. 25, 412–416. Woolson, E. A., Axely, J. H., and Kearney, P. C. (1973). The chemistry and phytotoxicty of arsenic in soil. II. Effect of time and phosphorus. Soil Sci. Soc. Am. Proc. 37, 254–258. Xie, Z. M., and Huang, C. Y. (1998). Control of arsenic toxicity in rice plants grown on an arsenic-polluted paddy soil. Commun. Soil Sci. Plant Anal. 29, 2471–2477. Xu, H., Allard, B., and Grimvall, A. (1988). Influence of pH and organic substance on the adsorption of As(V) on geological materials. Wat. Air Soil Pollut. 40, 293–305. Yan-Chu, H. (1994). Arsenic distribution in soils. In “Arsenic in the Environment, Part I: Cycling and Characterization” (J. O. Nriagu, Ed.), pp. 17–49. Wiley, New York. Yeates, G. W., Orchard, V. A., Speir, T. W., Hunt, J. L., and Hermans, M. C. C. (1994). Impact of pasture contamination by copper, chromium, arsenic timber preservative on soil biological activity. Biol. Fertil. Soils 18, 200–208. Yih, L. H., Peck, K., and Lee, T. C. (2002). Changes in gene expression profiles of human fibroblasts in response to sodium arsenite treatment. Carcinogenesis 23, 867–876. Yokota, H., Tanabe, K., Sezaki, M., Akiyoshi, Y., Miyata, T., Kawahara, K., Tsushima, S., Hironaka, H., Takafuji, H., Rahman, M., Ahmad, S. A., Sayed, M. H. S. U., and Faruquee, M. H. (2001). Arsenic contamination of ground and pond water and water purification system using pond water in Bangladesh. Eng. Geology 60, 323–331. Yokota, H., Tanabe, K., Sezaki, M., Yano, Y., Hamabe, K., Yabuuchi, K., and Tokunaga, H. (2002). Arsenic contamination in groundwater of Samta, Bangladesh. Wat. Sci. Technol. 46, 375–380. Yoshida, I., Kobayashi, H., and Ueno, K. (1976). Selective adsorption of arsenic ions on silica gel impregnated with ferric hydroxide. Anal. Lett. 9, 1125. Yuan, T., Hu, J. Y., Ong, S. L., Luo, Q. F., and Ng., W. J. (2002). Arsenic removal from household drinking water by adsorption. J. Environ. Sci. Health A. 37, 1721–1736. Zhang, W. H., Cai, Y., Tu, C., and Ma, L. Q. (2002). Arsenic speciation and distribution in an arsenic hyperaccumulating plant. Sci. Total Environ. 300, 167–177. Zhao, F. J., Dunham, S. J., and McGrath, S. P. (2002). Arsenic hyperaccumulation by different fern species. New Phytol. 156, 27–32. Zou, B. J. (1986). Arsenic in soil. Tarangxue Jin Zhan. 14(2), 8–13.
THE CONTRIBUTION OF BREEDING TO YIELD ADVANCES IN MAIZE (ZEA MAYS L.) Donald N. Duvick Iowa State University Ames, Iowa 50011
I. Introduction A. Maize Yield Trends During the Past Century B. Factors Responsible for Upward Yield Trends II. Genetic Gains in Grain Yield of Hybrids A. Previously Reported Genetic Yield Gains B. Recent Estimates of Genetic Yield Gains C. Estimates of the Contribution of Breeding to Total Yield Gains D. Changes that Have Accompanied Genetic Yield Gains in Hybrids III. Genetic Gains from Population Improvement A. Comparisons with Genetic Gains in Hybrids B. Relative Contributions of Population Improvement and Pedigree Breeding IV. Analysis and Conclusions A. Possible Reasons for Genetic Yield Gains B. Potential Helps or Hindrances to Future Gains in Yield C. Predictions References
Maize (Zea mays L.) yields have risen continually wherever hybrid maize has been adopted, starting in the U.S. corn belt in the early 1930s. Plant breeding and improved management practices have produced this gain jointly. On average, about 50% of the increase is due to management and 50% to breeding. The two tools interact so closely that neither of them could have produced such progress alone. However, genetic gains may have to bear a larger share of the load in future years. Hybrid traits have changed over the years. Trait changes that increase resistance to a wide variety of biotic and abiotic stresses (e.g., drought tolerance) are the most numerous, but morphological and physiological changes that promote efficiency in growth, development, and partitioning (e.g., smaller tassels) are also recorded. Some traits have not changed over the years because breeders have intended to hold them constant (e.g., grain maturity date in U.S. corn belt). In other instances, they have not changed, despite breeders’ intention to change them 83 Advances in Agronomy, Volume 86 Copyright 2005, Elsevier Inc. All rights reserved. 0065-2113/05 $35.00
84
DONALD N. DUVICK (e.g., harvest index). Although breeders have always selected for high yield, the need to select simultaneously for overall dependability has been a driving force in the selection of hybrids with increasingly greater stress tolerance over the years. Newer hybrids yield more than their predecessors in unfavorable as well as favorable growing conditions. Improvement in the ability of the maize plant to overcome both large and small stress bottlenecks, rather than improvement in primary productivity, has been the primary driving # 2005, Elsevier Inc. force of higher yielding ability of newer hybrid.
I. INTRODUCTION A. MAIZE YIELD TRENDS DURING
THE
PAST CENTURY
Maize yields began to rise markedly in many countries during the past century, first in the United States in the 1930s and then in other parts of the world in the 1950s and 1960s. For example: • U.S. yields, level at approximately 1.5 mg ha1 in the first three decades of the 20th century, started to rise significantly in the 1930s, reaching 8.5 mg ha1 by the end of the century (USDA-NASS, 2003b). The U.S. yield gains averaged 63 kg ha1 year1 from 1930–1960 and 110 kg ha1 year1 during the next 40 years (Troyer, 2000). • Maize yields in Canada tripled during the period 1940–2000, increasing from 2.5 to 7.5 mg ha1, a linear increase of 80 kg ha1 year1 (Bruulsema et al., 2000). • Maize yields in Germany doubled in the period 1965–2000, going from 4 to 8 mg ha1 (Frei, 2000). • Maize yields in France quadrupled in the period 1950–1984, increasing from 1.5 to 6.0 mg ha1 (Derieux et al., 1987). • In Argentina, the national mean maize yield increased “at a rate of 2.3% per year from 1970–1992” (Eyhe´ rabide et al., 1994). Table I summarizes yield gain data for several regions of the world during the period 1961–2002. Globally, maize yields doubled during this time, from 1.9 to 4.3 mg ha1, a linear increase of 61 kg ha1 year1. Different regions varied in the size of annual gain, as well as in average yields at the beginning and the end of the interval, but all showed positive and significant gains with the exception of eastern Europe (highly variable in the past decade) and southern Africa (minimal gain and highly variable during entire period). Yields in south Asia did not start to rise significantly until the 1980s; annual gains since 1985 have averaged 38 kg ha1 year1.
ADVANCES IN MAIZE
85
Table I Maize Yield Trends in Selected Regions (1961–2002)a Region European Union (15) USA China Canada World South America Eastern Europe South Asia Southern Africa
1961 mean (mg ha1)
2002 mean (mg ha1)
Annual gainb (kg ha1 year1)
R2c
2.5 3.9 1.2 4.6 1.9 1.4 1.8 1.0 0.7
9.1 8.2 5.0 7.6 4.3 3.4 4.2 1.7 1.3
169 109 103 69 61 48 42 20 8
0.98 0.83 0.96 0.77 0.95 0.87 0.38 0.78 0.26
a
From FAO Statistical Databases (2004) http://apps.fao.org/default.htm. Linear regression coefficients, calculated from annual means, 1961–2002. c Coefficient of determination. b
These examples and other data show that maize yields have increased significantly in many regions of the world during the latter half of the 20th century, especially in those places where maize is grown as a commercial crop.
B. FACTORS RESPONSIBLE FOR UPWARD YIELD TRENDS 1.
Cultural Practices
Changes in cultural practices have been responsible for a significant portion of maize yield gains. Crop management practices, such as weed and pest control, timeliness of planting, and increased efficiency of harvest equipment, have improved over the years, especially (but not exclusively) in the industrialized countries (e.g., Cardwell, 1982; Edmeades and Tollenaar, 1990). Perhaps most importantly, the use of synthetic nitrogen fertilizers increased markedly starting in the years after World War II when plentiful and affordable supplies became available, first in the industrialized countries and then in many (but not all) of the developing countries (e.g., Cardwell, 1982; Edmeades and Tollenaar, 1990; Miquel, 2001). Total fertilizer applications on all crops worldwide increased fivefold during the period 1961–1992. The linear increase started from an average application of about 20 kg ha1 in 1962 and reached 105 kg ha1 in 1992 (USDA-ERS, 2003). However, in some countries, application amounts of synthetic nitrogen fertilizer did not fit this general trend; they began to level off in the 1980s. Application of
86
DONALD N. DUVICK
commercial nitrogen fertilizer to maize plantings in the United States rose from an average of 58 kg ha1 in 1964 to 157 kg ha1 in 1985, but since then has stabilized at approximately 145–150 kg ha1 (Daberkow et al., 2000; USDA-ERS, 2003). It would seem, therefore, that yield gains of U.S. maize since the mid-1980s cannot be attributed to application of increasing amounts of nitrogen fertilizer on maize plantings. Plant density—the number of maize plants per hectare—also increased steadily through the years following World War II in the United States as well as in other countries. The increase was more or less in step with increases in application amounts of fertilizer nitrogen. In the central U.S. corn belt, plant density averaged about 30,000 plants hectare1 (or less) in the 1930s; it began to increase in the late 1940s and 1950s, reaching about 40,000 plants hectare1 in the 1960s, 60,000 plants hectare1 in the 1980s, and is often at 80,000 plants hectare1 or higher at present (Duvick, 1977, 1984a, 1992; Duvick et al., 2004b; Paszkiewicz and Butzen, 2001; USDA, 1949–1992). During the past 50 years, plant density in the central U.S. corn belt has increased at an average rate of about 1000 plants hectare1 year1.
2.
Plant Breeding
a. Farmer Breeding. Genetic improvements, as well as cultural improvements, can contribute to an increased yield of maize. Farmer breeders, beginning with the people who first domesticated maize, have selected plants and cultivars to fit their wants and needs and, in so doing, have developed thousands of landraces adapted to a multitude of environments, as well as with a wide range of morphological and quality traits (e.g., Goodman and Brown, 1988; Grobman et al., 1961; Paterniani and Goodman, 1977). We can assume that a higher yield, or at least an acceptable and dependable level of yield, was always a desired trait for maize cultivars, as well as for those of other staple grain crops. Although long-term yield trends are not recorded for specific farmer breeding programs, a general observation indicates that when crop varieties are grown in a new environment (e.g., when migrants carry their favorite cultivars to a new land), the cultivars often do not perform as well as intended. Careful selection in the unadapted cultivars, often coupled with hybridization to cultivars from elsewhere, then is used to develop genetically different cultivars that are better adapted to the new environment and therefore yield more (and more dependably) than the first introductions. Examples in U.S. history are the 19th century development of hard red winter wheat (Triticum aestivum L.) cultivars for Kansas (Malin, 1944) and “Corn Belt Dent” maize open pollinated cultivars (OPCs) for U.S. corn belt states such as Illinois and Iowa (Wallace and Brown, 1988).
ADVANCES IN MAIZE
87
Figure 1 United States maize yields, annual average, 1900–2003. From USDA-NASS (2003a).
Farmers developed adapted maize OPCs for the U.S. corn belt states in a relatively short time (Hallauer and Miranda, 1988; Wallace and Brown, 1988). Within a few decades after settlement of the region in the early years of the 19th century, maize yields and general performance of the new “Corn Belt Dent” cultivars were at acceptable levels in most parts of the region. However, from then on, gains in yield were small or nonexistent. This is evidenced by the lack of gain in U.S. maize yields during the first three decades of the 20th century (Fig. 1). One could suppose that the lack of yield gain during those decades was because maize-growing areas in the country changed in location and extent over time and therefore were not always equivalent in productivity. However, in the states of Iowa and Illinois, where maize-growing areas and cultural practices were relatively constant during this period, yields were essentially unchanged also. Yields in those states were level at approximately 2.3 mg ha1 during the years 1900–1930 (USDA-NASS, 2003a). It would seem that farmer breeders in the corn belt, using selection techniques of that time [primarily mass selection based on individual plant performance (Sprague, 1952)], could not raise maize yields further than the levels attained in the initial development of adapted cultivars. New breeding methods were tried in the late 19th and early 20th centuries. The production of varietal hybrids (first generation crosses of two maize OPCs) was tried and abandoned because of unreliable results (e.g., Crabb, 1993; Richey, 1922). A few professional breeders in the public sector (USDA) worked on variety improvement in the 1920s using relatively
88
DONALD N. DUVICK
unsophisticated methods of mass selection or ear-to-row breeding (Crow, 1998; Russell, 1991; Sprague, 1946, 1994). Their efforts did not increase yields either, except when a program provided adaptation to a new environment. These breeders, working in the first decades of the 20th century, lacked access to the present-day knowledge of experimental design, statistical analysis, and quantitative genetics. Lack of these tools must have hindered their progress. b. Hybrids. U.S. maize yields started to increase when maize hybrids made from crosses of inbred lines were introduced in the early 1930s. During the next few years the increase in maize yield was correlated with the increase in the proportion of maize area planted to hybrids (USDA, 1944 – 1962; USDA-NASS, 2003a). Yields in Iowa increased from 2 mg ha1 to 3.5 mg ha1 in the period 1933–1943, as the percentage of maize area planted to hybrids went from 0.7 to 99%. U.S. maize yields rose from 1.5 mg ha1 in 1933 to 2.4 mg ha1 in 1950, as the percentage of area planted to hybrids went from 0.1 to 78%. In either case, yield gains took place before a significant increase in use of synthetic nitrogen fertilizers or chemical control of weeds and insects (Cardwell, 1982; USDA, 1956), so it seems likely that the yield gains primarily were caused by genetic improvements; the new hybrids yielded more than the OPCs that they replaced, and successive hybrids yielded even more. Maize yields began to rise in conjunction with the introduction of hybrids in other countries as well (Cunha Fernandes and Franzon, 1997; Derieux et al., 1987; Eyhe´ rabide et al., 1994; Frei, 2000; Tollenaar, 1989), although, as in the United States, improved crop management techniques usually accompanied the introduction of hybrid maize; plant breeding and crop management jointly contributed to the sharp increases in maize yields. The proportion of gain attributed to genetic improvements is treated in more detail in later sections, with emphasis on hybrids and how sequential changes in their breeding and genetics have contributed to increased on-farm yield. c. Improved Populations. In the United States, the first hybrids were made from inbreds that had been developed by selfing some of the better OPCs of the 1920s. Breeders then worked to develop a second generation of improved hybrids using new inbreds made by selfing the same OPCs. They found that the second round of hybrids yielded little or no more than the first; it seemed that breeders must have selected most of the superior genotypes in the initial round of selfing in the OPCs. Some of the breeders conjectured that it might be possible to make new “synthetic” OPCs, with a potential for production of a superior second generation of inbred lines, by intercrossing some of the best inbreds from the first round of OPC selfing (Baker, 1990).
ADVANCES IN MAIZE
89
To this end, the breeders made several “synthetics” by intercrossing the better inbreds of the day. Research in maize quantitative genetics had begun by this time, and some of the populations were subjected to various kinds of selection to make genetic improvements in the populations as such. The selection procedures were based on various assumptions about gene action and genetic variability (Hallauer and Miranda, 1988; Sprague, 1946, 1966). The Iowa State University Stiff Stalk Synthetic (BSSS) (Eberhart et al., 1973; Sprague, 1946) is one of the best known of these populations. Sprague (1946) lists the 16 progenitor inbred lines of this synthetic. Breeders practiced population improvement on other kinds of populations as well, such as locally adapted OPCs, exotic landraces, or composites of exotic landraces and/or inbred lines (e.g., Hallauer and Miranda, 1988; Sriwatanapongse et al., 1985). The name “recurrent selection” was coined (Sprague, 1952) to distinguish these kinds of population improvements from pedigree breeding (i.e., developing improved inbred lines from crosses of proven inbreds). Depending on the prospective end user, breeders intended to develop improved populations that would serve as sources of superior inbred lines or that could be used directly as productive cultivars per se. Results of their work are discussed in a later section.
II. GENETIC GAINS IN GRAIN YIELD OF HYBRIDS A. PREVIOUSLY REPORTED GENETIC YIELD GAINS Russell (1991) has summarized 16 independent estimates of genetic yield gains of sequentially released maize hybrids. Most of the estimates are based on comparisons of U.S. hybrids and were reported at intervals during the 20-year period of 1971–1991. Estimates ranged from 25–92 kg ha1 year1 with a mean of 57 kg ha1 year1. It seems likely that the wide range in values was caused, in part, by differing growing conditions among the several investigations and consequent differential interactions with old or new genotypes. Other factors, as well, might explain some of the variation, as follows: • Breeding might have been less effective in some regions (such as those with erratic and often severe abiotic stress) than in others. • Choice of the time series of hybrids for comparison could have had major effect on size of measured gain. For example, a short time series might show less improvement than a long one if the short time series happened to sample a period with small genetic gain.
90
DONALD N. DUVICK
• Plant density of trials could affect the results differentially; older hybrids could have been disadvantaged if the plant density at which their yield was measured was higher than that for which they were bred or new hybrids could be disadvantaged if the density was below that which they required for maximum yield. • Harvest technology could be another source of difference; e.g., combine harvested trials (as compared with hand-harvested trials) could underestimate yields of older hybrids if combines failed to pick up all downed stalks (and ears) of older hybrids with poor standability. Conversely, hand-harvested trials could overestimate yields of the older hybrids if a standard shelling percentage was used to convert ear corn weight to grain weight instead of shelling the ears and weighing the shelled grain. Use of a standard shelling percentage could inflate estimates of grain weight on poorly pollinated nubbins of the older hybrids. However, these possibilities must remain conjecture. The salient fact is that all of the experiments listed by Russell showed positive and linear genetic yield gains, fluctuating around a mean of about 60 kg ha1 year1. Additional estimates of genetic gain in hybrids have been made since Russell’s review and are summarized in the following section.
B.
RECENT ESTIMATES 1.
OF
GENETIC YIELD GAINS
Argentina
Elite experimental maize hybrids tested in 154 regional trials in the Argentine corn belt during the 1979–1991 period had an estimated linear genetic gain of 105 kg ha1 year1 (Eyhe´ rabide et al., 1994). Estimates were based on comparisons with a common check. A second series of estimates extended the period (1979–1998) and showed an estimated genetic gain of 107 kg ha1 year1, or 2.9% year1. Further analysis of these data indicated that gains were not linear during the entire period; gains were greater in the second decade than in the first, perhaps in part because of the introduction of single cross hybrids in the 1990s (Eyhe´ rabide and Damilano, 2001). 2.
Brazil
Analysis of 30 years of national maize trials in Brazil (1963–1993) indicates linear genetic progress of 123 kg ha1 year1 (Cunha Fernandes and Franzon, 1997). Trials were grown in three locations, and estimates were based on comparisons with a moving base of common entries.
ADVANCES IN MAIZE
3.
91
United States
Duvick (1997), updating previous reports for hybrids adapted to central Iowa in the U.S. corn belt, stated that a time series of hybrids and one OPC representing the period from 1930–1991 showed a linear gain for grain yield of 74 kg ha1 year1. The estimate was based on data from trials comparing 36 hybrids and one OPC, conducted over a period of 4 years, three locations per year, at three plant densities per location. Yield for each hybrid was for its “optimum density” in the trials: the plant density at which it gave its highest mean yield. A further update extended this time series through the year 2001; it showed an estimated linear gain of 77 kg ha1 year1 (Duvick et al., 2004b). This estimate applied “best linear unbiased predictors” (BLUPs) of hybrid grain yield; it was based on trials of 51 hybrids and four OPCs grown at three plant densities in the years 1991–2001, using yield of each hybrid at its “optimum density” as described earlier.
C. ESTIMATES
OF THE CONTRIBUTION OF TO TOTAL YIELD GAINS
BREEDING
Russell (1991) listed 14 estimates of genetic yield gain of hybrids as percent of total yield gain. (Total yield gain was defined as on-farm gain for appropriate regions during the time span of hybrids that were compared.) Most of the comparisons were for the U.S. corn belt but the list also included estimates for Ontario (Canada), France, and Yugoslavia. Estimates of genetic gain varied from 29 to 94%, with a mean of 66%. As noted by Russell (1991), several reasons can be advanced to show that this broad variability could be caused by inconsistencies in planning and executing the experiments, such as machine harvest vs hand harvest, or whether experimental estimates of genetic gain were adjusted to on-farm state averages. Nevertheless, all estimates agree in showing that hybrid maize breeding (i.e., genetic improvement) has played a major part in raising maize yields. Among the reports of genetic gain since Russell’s summary (reviewed in Section II.B), Cunha Fernandes and Franzon (1997) estimated that 57% of total gain in yield in Brazil was due to genetics. The other reports did not contain such an estimate, but further examination of data in Duvick et al. (2004b) provides an estimate of 51% for the contribution of genetics, when trial yields are adjusted to the equivalent of average on-farm yields for Iowa during the period 1930–2001. Based on these and earlier estimates, one can state that hybrid maize breeding during the past six or seven decades has been responsible for 50 to
92
DONALD N. DUVICK
60% of the total on-farm yield gain. However, one also must acknowledge that the interaction between breeding and management (cultural practices) is such that neither tool could have caused the gains without aid of the other; changes in breeding and management continually have interacted in positive fashion.
D. CHANGES THAT HAVE ACCOMPANIED GENETIC YIELD GAINS IN HYBRIDS Breeders have noted that genetic gains in grain yield of hybrids may be accompanied by changes in other traits, sometimes as a result of direct selection, sometimes without direct intention by the breeders. And some traits have stayed essentially unchanged over the generations. Three reviews in the previous decade (Edmeades and Tollenaar, 1990; Russell, 1991; Tollenaar et al., 1994) have given detailed accounts of such changes and are recommended as sources of information and informed commentary on the topic prior to the early 1990s. The following sections update those accounts, as well as provide summaries and commentary for some of the earlier research reports.
1.
Plant and Ear Traits
a. Plant and Ear Height. Plant and ear height were reduced in the second era but not thereafter in a study of single cross hybrids representing U.S. corn belt hybrids of three eras: 1930s, 1950s, and 1970s (Meghji et al., 1984). A study of 28 hybrids and four OPCs adapted to Iowa, representing seven decades culminating in the 1980s, found no trend to reduction in plant height but a continuing trend to reduction in ear height (Russell, 1984). Plant height for a 1930–2001 time series of 51 hybrids and four OPCs adapted to central Iowa likewise was essentially unchanged over the years, but ear height showed a weak trend toward reduced height, approximately 3 cm decade1 (Duvick et al., 2004b). b. Leaf Angle. Leaves became more upright in the 1970s era in a comparison of single crosses representing U.S. corn belt hybrids of three eras: 1930s, 1950s, and 1970s (Meghji et al., 1984). Leaf orientation below and above the ear became more upright with time in a study of U.S. Midwestern hybrids representing the decades from 1930–1970 (Crosbie, 1982; Russell, 1991). The trend to upright leaf orientation was greatest above the ear. Russell (1991) stated that the distinct increase in upright leaf orientation in the1970s decade was probably because inbred B73, with upright leaves for its time, was
ADVANCES IN MAIZE
93
a parent in the set of 1970s single cross hybrids. The previously mentioned 1930–2001 time series of 51 hybrids and four OPCs for central Iowa (Duvick et al., 2004b) showed a similar trend toward more upright leaf habit. Ratings (as scores) in this study were for the entire plant. c. Tassel Size. Tassel weight was least in the most recent era in comparisons of single cross hybrids representing U.S. corn belt hybrids of the 1930s, 1950s, and 1970s; tassel branch number decreased consistently over the eras (Meghji et al., 1984). Tassel branch number and tassel weight decreased over time in a 1930 to 1991 time series of hybrids for Iowa (Duvick, 1997). Tassel branch numbers in the series averaged 2.5 fewer branches per decade, and tassel dry weight declined, on average, 0.5 g per decade. Reduction of tassel size continued in hybrids released during the next 10 years, as evidenced by scores for tassel size of hybrids in the 1930–1991 time series extended to 2001 (Duvick et al., 2004b). d. Leaf Number. Number of leaves per plant neither increased nor decreased in a 1930–1991 time series of Iowa hybrids and OPCs (Duvick et al., 2004b). Leaf number increased from 12.2 in the 1930s to 13.8 in the 1970s in comparisons of single cross hybrids representing U.S. corn belt hybrids of the 1930s, 1950s, and 1970s (Meghji et al., 1984). e. Leaf Area Index (LAI). Russell (1991) suggests that changes in LAI “may be specific to the particular cultivars used, rather than a general occurrence of all germplasm representative of similar eras.” This statement is borne out by the lack of consistent trends across experiments conducted by different researchers. LAI tended to be higher for recent hybrids than for older ones in a time series of four hybrids grown in Ontario (Canada) from 1959 to 1989 (Dwyer et al., 1991; Tollenaar, 1991). In another investigation, LAI increased over time in a set of eight maize hybrids that were commercially important in central Ontario between 1959 and 1988 (Tollenaar, 1989). However, a set of Iowa hybrids (20 single cross hybrids) representing the decades of 1930–1970 showed no obvious trend (Crosbie, 1982; Russell, 1991), and a 1930–1991 time series of 36 commercial hybrids and one OPC for Iowa also showed no change in LAI over time (Duvick, 1997). f. Leaf Rolling. Leaf rolling of plants, especially in the vegetative stage, is often seen when plants are subjected to drought. Leaf rolling consistently increased in newer hybrids in a set of 18 commercial hybrids adapted to central Iowa and representing the period 1953–2001 when the hybrids (grown in a rain-free environment in Chile) were compared with and without managed drought stress at various stages of development (Barker et al., 2005; Edmeades et al., 2003). Commenting on this observation, Barker et al.
94
DONALD N. DUVICK
(2005) said “Apparently elite hybrids can reduce radiation interception and water use by leaf rolling, while generating sufficient assimilate flux to the ear to set adequate kernel numbers and conserving water for later in the season.” g. Staygreen. Staygreen, also called delayed leaf senescence, or resistance to premature death from unidentified causes, is consistently improved in newer hybrids (Crosbie, 1982; Duvick et al., 2004b; Meghji et al., 1984; Russell, 1991; Tollenaar, 1991). The improvement in each of these trials was greatest under environmental stress such as that induced (or accentuated) by high plant density. Staygreen improved over time in a set of 18 commercial hybrids adapted to central Iowa and representing the period 1953–2001 when the hybrids, grown in a rain-free environment in Chile, were (a) well watered, (b) subjected to induced drought at flowering time, or (c) subjected to drought during the final grain-fill period (Barker et al., 2005). Genetic gain for staygreen was greatest in the well-watered treatment and was least under drought imposed during the grain-fill period. h. Tillers. Number of tillers per 100 plants varied from hybrid to hybrid, but decreased slightly on average in a 1930–1991 time series of hybrids and OPCs for Iowa (Duvick et al., 2004b). i. Anthesis. Date of anthesis varied among decades, but showed no trend toward earlier or later dates in an Iowa series of four OPCs and 24 single cross hybrids representing the 1930s through 1980s (Russell, 1985). Similarly, heat units from planting to anthesis varied among hybrids but showed no general trend toward earlier or later in a 1930–2001 time series of 51 hybrids and four OPCs adapted to central Iowa (Duvick et al., 2004b). j. Silk Emergence. Silk emergence date trended toward earliness in an Iowa series of four OPCs and 24 single cross hybrids representing the 1930s through 1980s (Russell, 1985) and also in a 1930–1991 time series of 36 hybrids and one OPC for Iowa (Duvick, 1997). In the latter case, there was little or no trend to an earlier silk date in absence of stress, such as at low plant density, whereas higher plant density accentuated the trend, not because the new hybrids had earlier silking dates, but rather because the stress of high plant density delayed silk emergence of the older hybrids. k. Anthesis-Silking Interval (ASI). Anthesis usually precedes silk emergence, and the interval between the two events is called the anthesissilking interval. (The term “silk delay” is also used.) ASI was unchanged in hybrids representing earlier decades but was significantly shorter in hybrids
ADVANCES IN MAIZE
95
of the 1970s in a set of Iowa hybrids (20 single cross hybrids) representing the decades of 1930–1970 (Crosbie, 1982). ASI became shorter in each decade except the 1980s in an Iowa-adapted series of four OPCs and 24 single cross hybrids representing 1930–1980 (Russell, 1985). ASI was significantly shorter in each interval in comparisons of single cross hybrids representing U.S. corn belt hybrids of the 1930s, 1950s, and 1970s (Meghji et al., 1984). ASI showed a highly significant linear trend to shorter intervals in a 1930–2001 time series of 51 hybrids and four OPCs adapted to central Iowa (Duvick et al., 2004b). The trend was greater in trials grown at higher plant densities. ASI became shorter over time in a set of 18 commercial hybrids adapted to central Iowa and representing the period 1953–2001 (Barker et al., 2005; Edmeades et al., 2003). This trend was exhibited when the hybrids were well watered and was accentuated when they were subjected to induced drought at flowering time. The experiment was conducted in a rain-free environment in Chile. l. Ears per Plant. Both total and harvestable ears per plant increased over the decades in a set of Iowa hybrids (20 single cross hybrids) representing the decades of 1930–1970 (Crosbie, 1982). A 1930–2001 time series of 51 hybrids and four OPCs adapted to central Iowa showed a highly significant trend toward more ears per 100 plants (+3.6 ears decade1) (Duvick et al., 2004b). However, ears per plant showed no change over the decades in an Iowa-adapted series of four OPCs and 24 single cross hybrids representing 1930–1980 (Russell, 1985). The Russell experiment was planted at a single density (51670 plants ha1), whereas data for the other two experiments were expressed as means of three densities in which the medium and high densities were higher than optimum for the older hybrids and, therefore, were more likely to cause barrenness in those hybrids. The end result would be a trend toward more ears per 100 plants (i.e., fewer barren plants) in the newer hybrids. m. Grain-Filling Period. Newer hybrids had a longer period of grain fill, calculated as time from silk emergence to black layer (physiological maturity), in observations of four discrete time series of hybrids adapted to the Midwestern United States (Cavalieri and Smith, 1985; Crosbie, 1982; Meghji et al., 1984; Russell, 1985). The newer hybrids flowered at approximately the same time as the older hybrids, and although their grain-fill periods were longer, they also exhibited a faster final dry-down rate, and so had little or no delay in relative maturity (based on grain moisture percentage at harvest time). The newer hybrids thus have more time to devote to grain fill; they exhibit increased efficiency in use of the growing season in the U.S. Midwest, which is limited on either end by average date of last frost in spring and first frost in autumn.
96
DONALD N. DUVICK
n. Kernel Weight. Weight per 300 kernels increased in each decade except the final one in an Iowa-adapted series of four OPCs and 24 single cross hybrids representing seven eras from pre-1930s through the 1980s (Russell, 1985). Kernel weight did not change significantly over the decades except for a significant increase in the final (1970s) decade in comparisons of 20 single cross hybrids representing each decade from 1930–1970 (Crosbie, 1982). Weight per 100 kernels increased linearly in an Iowa-adapted series of 36 hybrids and one OPC representing the years 1930–1991, while number of kernels per ear decreased slightly but not significantly (Duvick, 1997). Weight per kernel exhibited a marked linear increase under well-watered conditions and also under drought stress at flowering, early, and midfill stages, but showed little or no increase under drought in late-fill and terminal stages in a set of 18 Iowa-adapted hybrids (evaluated in Chile) representing the period 1953–2002 (Barker et al., 2005; Edmeades et al., 2003). In appraisal of the results summarized in this section: the general trend to increased weight per kernel (and no increase in number of kernels per ear, in the one series where this was measured) may indicate that assistance in achieving genetic yield gain over time (and also gain in yield stability) is given more efficiently by plants with increased kernel weight than by plants with increased kernel number. o. Grain Protein Percentage. Grain protein percentage declined consistently in a series of 36 hybrids and one OPC for central Iowa spanning the period 1930–1991 (Duvick, 1997). The loss averaged 0.3% protein decade1, with a series mean of 9.8% protein. p. Grain Starch Percentage. Grain starch percentage increased consistently in a series of 36 hybrids and one OPC for central Iowa spanning the period 1930–1991 (Duvick, 1997). The increase averaged 0.3% starch per decade, with a series mean of 70.4% starch. Because production of starch requires less energy than production of protein, selection for yield without attention to protein or starch percentage may have indirectly selected genotypes with less grain protein and more grain starch, giving a net increase in efficiency of grain production. q. Harvest Index (HI). HI did not change consistently over time in comparisons of single cross hybrids representing U.S. corn belt hybrids of the 1930s, 1950s, and 1970s (Meghji et al., 1984). HI did not change consistently in a set of 20 single cross hybrids representing Iowa hybrids of the decades of the 1930s through 1970s (Crosbie, 1982) or in a series of nine hybrids representing three decades (1959–1988) of maize production in Ontario when hybrids were compared at optimum plant density for yield for each hybrid (Tollenaar, 1989). HI did not change consistently over the decades in an Iowa-adapted series of four OPCs and 24 single cross hybrids
ADVANCES IN MAIZE
97
representing 1930–1980 (Russell, 1985). HI improved on average to a small degree in successive hybrids in a 1934–1985 time series of Iowa-adapted hybrids (Duvick et al., 2004b) and in the same series extended to 1991 (Duvick, 1997). Higher plant density accentuated the trend. However, the superior HI of the newer hybrids at higher plant densities was not because of increased HI per se in the newer hybrids, but because of reduced HI in the older hybrids—the result of barrenness induced by stress. At higher densities the older hybrids maintained plant size but lost yield because of increased percentages of barren or partially barren plants. 2.
Resistance to Root Lodging
Hybrids improved over the years in resistance to root lodging in a series of Iowa hybrids representing the decades 1930–1960 (Russell, 1974). However, a later examination of a longer series of hybrids (1930–1980) showed no significant improvement in root lodging resistance, although all hybrids were decidedly more resistant than the OPCs of the 1920s (Russell, 1984). Four examinations of a successively lengthened time series of commercial hybrids for central Iowa showed linear increases in resistance to root lodging, in each examination (Duvick, 1977, 1984a, 1997; Duvick et al., 2004b). The four experiments contained hybrids released in the years 1939–1971, 1934–1978, 1934–1991, and 1934–2001, respectively. However, in two other trials of this series (for hybrids released in 1934–1989 and 1934–2000), improvement of root strength ceased in the final decade at about the 95% nonroot-lodged level (Duvick, 1992; Duvick et al., 2004a). The intensity of root lodging in a given trial can influence the differentiation between the older and the newer hybrids. Low levels of lodging (e.g., because of lack of the right combination of rain and wind or because of insufficient plant density) will make it impossible to differentiate among the more resistant hybrids because all will be in the range of 95 to 100% upright. Also, from time to time new hybrids might be released with less resistance to root lodging than intended, following which genetic improvements are implemented in newer releases. Either of these conditions (abiotic or genetic) could explain why apparent cessation of linear improvement in one time series is followed by a later time series in which gains are linear. 3.
Resistance to Stalk Lodging
Lodging resistance (not differentiated between root or stalk lodging) improved significantly in a series of hybrids grown in France from 1950–1985 (Derieux et al., 1987). Higher plant densities accentuated the difference between older and newer hybrids.
98
DONALD N. DUVICK
Stalk lodging resistance improved consistently in a time series of Iowaadapted hybrids representing 1930–1960 (Russell, 1974). However, in a longer series (1930–1980), resistance to stalk lodging improved until the 1970s but did not improve in the 1980s (Russell, 1984). Five different examinations of a successively lengthened time series of commercial hybrids for central Iowa showed linear improvements in resistance to stalk lodging (Duvick, 1977, 1984a, 1992, 1997; Duvick et al., 2004b). The five series contained hybrids released in 1939–1971, 1934–1978, 1934–1989, 1934–1991, and 1934–2001. However, in one other trial for this series (hybrids released from 1934–2000), improvement ceased in the final decade at about the 95% nonstalk-lodged level (Duvick et al., 2004a). So for both root lodging and stalk lodging, one can conclude that although the overall trend is toward improved resistance to lodging, improvement may seem to stop from time to time. Additional experiments involving further extensions of the time series will be needed to test the validity of such conclusions.
4.
Tolerance to Abiotic Stress
a. High Temperatures. A 1930–1991 time series of 36 hybrids and one OPC for Iowa exhibited a linear increase in grain yield in a low yield season with a “hot and dry” summer, as well as in two highly favorable (exceptionally high yield) seasons (Duvick, 1997). Weather records for the “hot and dry” year (1991) indicate that temperatures during the flowering period were higher than normal and precipitation was exceptionally low (Iowa State University, 2003). b. Low Temperatures. The aforementioned 1930–1991 time series also exhibited a linear increase in grain yield in a “very cool and wet season” (Fig. 2) (Duvick, 1997; Duvick et al., 2004b). Weather records for that trial year (1993) indicate that precipitation amounts during the summer months were at record-breaking high levels, and daytime high temperatures were well below normal (Iowa State University, 2003). Dwyer and Tollenaar (1989) stated that “genetic improvement in the response to cold stress . . . has significant consequences for yield of fieldgrown maize, since many Canadian seasons are subject to short seasons or cool growing periods.” They showed (for a series of eight hybrids, released during the years 1959–1988) that reduction in photosynthetic response to irradiance (PRI) following a cold period during kernel fill was greater in older than in newer hybrids. A subsequent study (Tollenaar et al., 2000) showed similar results for a series of eight U.S. maize hybrids representing the 1930s, 1950s, 1970s, and 1990s.
ADVANCES IN MAIZE
99
Figure 2 Grain yield per hybrid regressed on year of hybrid introduction for trials grown in 1992, 1993, and 2001. Seasons: 1992, highly favorable; 1993, cool and extremely wet; 2001, hot and dry. Yield per hybrid is for the density giving the highest average yield. From Duvick et al. (2004b). Copyright # 2004 by John Wiley & Sons, Inc. This material is used by permission of John Wiley & Sons, Inc.
Frei (2000), reporting for maize production in northwestern Europe, stated, “Breeding for adaptation to long and cool growing season has [led] to changes in growth behavior and yield physiology. . . . There is good evidence that low base temperature genotypes exist in northern breeding programs. . . . Breeding for lower base-temperature or cold tolerance can alleviate the stover versus grain antagonism.” [On a personal note, the author has seen some short-season inbred lines from the northern U.S. and Canada fail to make chlorophyll (white or striped seedling plants) in the cool early summer of central Germany, while locally bred lines in the same nursery were green and vigorous.] c. Drought. Russell (1974) stated that in high stress drought environments, a group of commercial hybrids representing the most recent era yielded considerably more than any other group. Hybrids and OPCs in the trial were adapted to Iowa and represented the approximate period 1930–1970. Examination of another set of hybrids comparing decades 1930–1980 showed that the hybrids of the 1970s and 1980s had superior
100
DONALD N. DUVICK
yields in all environments, which included two drought-stress locations and two high-yielding environments (Russell, 1991). Comparison of a series of U.S. corn belt hybrids and OPCs representing the decades 1930–1980 showed linear gains in grain yield under either drought stress or irrigated treatments (Castleberry et al., 1984). The mean yield of the 1930s group was equal to 60% of the mean yield of the 1980s group when both groups were subjected to drought stress, and 63% of the mean yield of the 1980s group when both groups were given full irrigation. In Ontario, Canada, a newer hybrid was more tolerant of short drought periods than an older hybrid (Dwyer et al., 1992). During a drought period, the newer hybrid continued photosynthesis for about 2 h longer than the older one before starting to decline. A further study indicated that the two hybrids might adopt different mechanisms to tolerate moisture stress (Nissanka et al., 1997). The newer hybrid maintained relatively higher rates of photosynthesis and transpiration at a lower stem water potential. Although one cannot consider a comparison of only two hybrids as a “time series” that demonstrates trends over the years, examinations like these can give hints of possible trends and suggest profitable fields for future investigation. Derieux et al. (1987), comparing 33 maize hybrids (of three maturity groups) grown in France from 1950–1982, stated that modern hybrids are more adapted to stress, such as low temperature and drought. Regressions of mean yield per decade of release on mean yield per location of trial consistently showed that the newer the decade, the higher the yield at all locations. Water stress limited yield in some locations, particularly for hybrids in the semiearly category. As noted earlier in the section “High Temperatures,” a 1930 to 1991 time series of 36 hybrids and one OPC for Iowa exhibited a linear increase in grain yield in a season (year 1991) with exceptionally low precipitation during flowering, as well as in highly favorable seasons (Duvick, 1997). The same series, further extended (1930–2001), showed a linear increase in grain yield in another season (year 2000) when “heat and drought at silking time caused reduced yields” (Duvick et al., 2004a), and also in a third season (year 2001) when yields were low “because of a season-long drought, especially severe at the sensitive anthesis-silking period” (Fig. 2) (Duvick et al., 2004b). Weather records show that rainfall was well below average during the anthesis-silking period in 2000, and also in 2001 (Iowa State University, 2003). A time series of 18 commercial hybrids adapted to central Iowa and representing the period 1953–2001 showed linear gains in grain yield in each of three different watering regimes. The hybrids (grown in a rain-free environment in Chile) were (a) well watered, (b) subjected to drought at
ADVANCES IN MAIZE
101
flowering time, or (c) subjected to drought during the grain-fill period (Barker et al., 2005; Edmeades et al., 2003). The hybrids in this experiment were a subset of the series studied by Duvick et al. (2004b), discussed previously. Gains in grain yield under optimal conditions were about twice as large as gains when stress coincided with flowering or grain filling. A time series of 2 OPCs and 52 hybrids adapted to central Iowa and representing the years (for the hybrids) 1934–2001 was subjected to a managed drought trial in Woodland, California (Barker et al., 2005). The hybrids in this experiment were the same as those studied by Duvick et al. (2004b), discussed previously. Watering regimes were similar to those described for the experiment in Chile (described earlier). Trials were grown at two plant densities. Both of the densities showed linear gains in grain yield for all three watering regimes. Figure 3 shows results of the trial at high density. Yield gain was greatest in the well-watered regime, although differences among the three regimes were not large. Annual genetic gains for all watering regimes were greater in the trial grown at the higher density, typical also of multidensity trials in rain-fed environments. All of the experiments described in this section show that genetic yield gains over time are expressed in drought as well as in favorable growing
Figure 3 Grain yield of two OPCs and 52 hybrids regressed against year of release. Hybrids were grown in Woodland, California, at 90,000 plants ha1 in three managed stress environments: full irrigation, flowering drought stress, or grain-filling drought stress. Adapted from Barker et al. (2005).
102
DONALD N. DUVICK
seasons. On average, the newer the hybrid, the greater is its drought tolerance. Discussion in later sections suggests possible causes of this improvement. d. Excessive Soil Moisture. As noted earlier in this section (“Low Temperatures”), a 1930 to 1991 time series of Iowa hybrids showed a linear gain in grain yield in a “very cool and wet season” (Duvick, 1997). That growing season, 1993, was one of the wettest on record for the state of Iowa (precipitation was two to three times normal), and soils were excessively moist (and in some cases flooded, although not in these trials) during much of the summer (Corrigan, 2003; Iowa State University, 2003). The linear gain in yield when the 1930 to 1991 time series was grown in this unusually wet growing season (see Fig. 2) indicates that although breeders had not selected directly for such abnormally wet growing conditions, they must have done so indirectly, perhaps through improvement in the ability of plants to set and develop kernels in the presence of reduced photosynthesis per plant or in the ability to tolerate a reduced uptake of key soil nutrients. One should note, however, that yield gain was least (59 kg ha1) in the flood year of 1993 (the lowest yielding year) and greatest (82 kg ha1) in the most favorable season, 1992. e. Deficiency of Soil Nitrogen. Maize cultivars (OPCs and commercial hybrids) typical of those grown in the U.S. corn belt in the decades 1930s through 1980s were compared at high and low soil fertility levels (in trials receiving approximately 200 kg ha1 N, 90 kg ha1 P2O5, and 150 kg ha1 K2O versus trials in an area that, for two decades, had received no fertilizer and was planted to continuous maize) (Castleberry et al., 1984). Yield gains by decade were linear and positive under both of the soil fertility treatments (high and low), although the average annual gain was greater in the high fertility trial. Single cross hybrids representing four decades (1940–1970) of U.S. corn belt hybrids were compared at three levels (70, 130, and 200 kg ha1) of nitrogen (N) fertilizer application (Duvick, 1984a). The l970s decade gave the highest grain yield at all N levels, and the 1960s decade produced the second highest yield at all N levels. A second experiment, described in the same report, compared five commercial hybrids spanning the period 1940–1978 at two treatments: high N, high plant density (215 kg N ha1 and 54,000 plants ha1) and low N, low plant density (70 kg N ha1 and 35,000 plants ha1) The two newest hybrids yielded more than the older ones at either treatment level. In both experiments, the interaction of hybrids N rates was not statistically significant. Carlone and Russell (1987) compared OPCs and a series of single cross hybrids at three plant densities (34,445, 51,661, and 68,889 plants ha1) and four N fertilizer levels (0, 80, 160, and 240 kg ha1). The single crosses were
ADVANCES IN MAIZE
103
chosen to represent hybrid genotypes of the decades 1930–1980. The trials suffered severe moisture stress in both years of trial, 1983 and 1984. Comparisons used the yield of each era at its optimum plant density. Hybrids of the older eras had their highest yield at lower densities; the newer hybrids had their highest yield at the higher densities. The 1980s era had the highest yield at each N level, and the 1970s era had the second highest yield at each N level. Carlone and Russell (1987) reported that levels of N fertilizer (0–240 kg ha1) interacted with plant densities and with hybrid genotype. The optimum N level (level with highest yield) for hybrids of the 1940s, 1950s, and 1960s was higher than the optimum N level for hybrids of the 1970s and 1980s. However, the highest yields of the older group were lower than those of the newer group at all N levels, so one might conclude that the newer hybrids used N more efficiently than the older hybrids. Carlone and Russell (1987) also showed that hybrids within an era differed in response to densities and N levels. Two hybrids of the 1970s group increased in yield as densities and N levels increased, but one hybrid increased in yield significantly more than the other, such that the greatest difference in yield between them was at the highest plant density and the highest N level. McCullough et al. (1994a) stated that when two hybrids were compared in controlled environment chambers, an old hybrid (release year 1959) was more sensitive than a new hybrid (release year 1988) to stress caused by low soil N (0.5 mM) during early development. The new hybrid also maintained a higher rate of leaf photosynthesis per unit of N regardless of N supply. A second experiment (McCullough et al., 1994b) indicated that the higher N-use efficiency of the new hybrid under low N supply “is associated with higher N uptake and a higher leaf N per unit leaf area.” Field trials confirmed that the new hybrid yielded more than the old hybrid under both high N and low N treatments (Tollenaar et al., 1994, 1997). The yield difference between hybrids was accentuated when weeds were present, as compared with weed-free conditions; one might conclude that the new hybrid was also more “weed tolerant” than the old hybrid. As stated earlier, comparison of only two hybrids representing “early” and “late” eras is not equivalent to the study of trends in a time series of several hybrids, but the comparison can indicate possible changes over time and may suggest profitable areas of future research to discover trait changes that accompany sequential improvements in hybrid performance. f. Unspecified Abiotic Stress—“Stress versus Nonstress” Environments. Yield trial results can be categorized according to the average yield at each test site. One can assume that the lower the yield at a given site (in absence of obvious disease or insect problems), the greater the amount
104
DONALD N. DUVICK
of “unspecified” abiotic stress. A stability analysis of the kind proposed by Eberhart and Russell (1966) can be used to compare yield response of individual hybrids or of groups of hybrids such as those released in a given decade. Yields of groups of hybrids (as in a decade) can be regressed on mean yield at each test site (e.g., Figs. 4 and 5). Russell (1991) cited such comparisons in several experiments. In general, all experiments showed that the newest groups of hybrids had the highest yields in all sites, regardless of average yield level at the site. However, linear regressions (b), showing degree of response of each group of hybrids to increasing site productivity, demonstrated no consistent trend of response. In some cases, b values were similar for all eras, whereas in other cases, b values were greater for new than for old eras, and in still other experiments, the b values differed randomly among eras (e.g., Fig. 4). Russell concluded that “there seems to be no distinct relationship between response and era of the hybrids. More likely, the responses were specific for the genotypes.” He
Figure 4 Yield response, b, for open-pollinated cultivars (OPCs) and six hybrid groups of 10-year eras, 1930–1980, to eight environment indexes (four locations 2 years). Reprinted from Russell (1991), # 1991, with permission from Elsevier, and also with permission from W. A. Russell and the Iowa State Journal of Research.
ADVANCES IN MAIZE
105
Figure 5 Mean grain yield of hybrids released within two-decade spans, and of three OPCs, regressed on mean yield of all hybrids per environment. Trials were grown in a total of 13 environments during the years 1996–2000. Means of three densities per environment: 30,000, 54,000, and 79,000 plants ha1. From Duvick et al. (2004b). Copyright # 2004 by John Wiley & Sons, Inc. This material is used by permission of John Wiley & Sons, Inc.
also considered the possibility that hand harvest vs machine harvest could have introduced a bias in some of the results that he reviewed. (Combines might fail to gather ears of lodged plants, and because older hybrids tend to lodge more than newer hybrids, yields of older hybrids would be underestimated.) Since the Russell (1991) report, Duvick et al. (2004b) presented results for 42 commercial hybrids and four OPCs tested in 13 environments in central Iowa during the years 1996–2000. They were grouped for stability analysis as follows: OPCs, 1930s and 1940s, 1950s and 1960s, 1970s and 1980s, and 1990s (Fig. 5). The regression for the OPCs was well below that for the hybrids, at 0.65. Regression values were similar for all hybrid eras (b ¼ ca. 1.0), although with a slight increase for the newest era. Thus, in this experiment the OPCs showed markedly less response than the hybrids to higher yield environments and the newest hybrid group gave the greatest response. However, in all cases, the newer the era, the higher the yield in any location, low yield or high. Stress tolerance increased significantly over the years. In conclusion, experiments described in this section (“Tolerance to Abiotic Stress”) have shown that hybrid tolerance to abiotic stress has
106
DONALD N. DUVICK
increased consistently over the years. Although results are contradictory regarding whether newer hybrids are more or less responsive than older hybrids to higher yield environments, newer hybrids tend to be the most responsive, sometimes strikingly so (e.g., Barker et al., 2005; Duvick, 1984a, 1992). Reasons for such variability in response are not known but probably depend, as do most results, on interactions of genotype and environment.
5.
Tolerance to Biotic Stress
a. Insects. A 1930–1991 time series of 36 hybrids and one OPC for Iowa exhibited a linear increase in resistance to second-generation European corn borer (Ostrinia nubilalis Hubner) (ECB2), as measured by tunnel length following artificial infestation and by scores for evidence of natural infestation in yield trials (Duvick, 1997). This improvement took place even though breeders had not selected directly for resistance to ECB2. The same series showed no improvement in resistance to first-generation borer. Breeders and entomologists in the United States have collaboratively produced inbreds and breeding populations with improved natural tolerance and/or resistance to both generations of borer (see review in Russell, 1991), but there is no record of how or if these materials were used in commercial hybrids. They were not the source of increased resistance in the aforementioned 1930–1991 time series. In recent years (starting in 1996), seed companies have commercialized transgenic maize hybrids that are resistant to ECB. These hybrids, commonly called Bt hybrids, have been genetically engineered to incorporate a gene of Bacillus thuringiensis (Bt). Most of the first Bt hybrids contained the gene that produces the insecticidal protein Cry1Ab. The toxic Bt protein is effective against larvae from both first and second ECB generations (Peferoen, 1992; Traore et al., 2000). When subjected to artificial infestation, Bt hybrids showed significantly less tunneling from second-generation borer than non-Bt hybrids (Traore et al., 2000). They also had 9.7% more total plant weight in 1997 and 9.4% more grain yield in 1998 than their non-Bt counterparts. However, the amount of difference depended on the cultivar. Under natural on-farm ECB infestation, Bt hybrids usually yield significantly more than their isogenic counterparts in seasons when infestation is relatively heavy, but not when infestation is light. In 14 yield trials in Iowa in 1997, nine Bt hybrids yielded 7% more than their near-isogenic counterparts (Rice, 1997). Similar advantage for Bt hybrids was measured in several other corn belt states in 1997, but in 1998 with lighter infestation the average yield advantage was about one-fifth that of the 1997 amount (Gianessi and Carpenter, 1999).
ADVANCES IN MAIZE
107
These data indicate that for Bt hybrids, as with other types of biotic and abiotic stress tolerance, the amount of yield gain contributed by the beneficial trait depends on the severity of the pertinent stress (in this case ECB infestation). Thus, James (2003a) stated that global yield gains due to Bt maize are currently estimated at 5% in the temperate maize-growing areas and 10% in the tropical areas. The tropical areas have more and overlapping generations of pests, leading to higher infestations and subsequently greater yield loss, in absence of resistance contributed by Bt. An additional restriction on potential yield gain from Bt hybrids is the need to plant “refuge areas” (perhaps 20–30% of total) of non-Bt maize to help prevent the development of resistance in the corn borer population (Ostlie et al., 2002). Biologists theorize that the pest population will eventually develop/increase new genotypes that are not susceptible to the Bt resistance genes, as has happened repeatedly (often in only a few seasons) in other instances of major gene resistance [also called vertical resistance (Simmonds, 1985; van der Plank, 1963)]. The use of refuge areas is intended to delay such genotypic change as long as possible. The refuge areas of course cannot provide the genetic yield advantage in the presence of corn borer that is provided by the Bt hybrids. More recently (in 2003), approval has been granted for use of a Bt transgene that prevents root injury by larvae of two different species of rootworm (Diabrotica barberi Smith & Lawrence, and Diabrotica virgifera virgifera LeCont) (James, 2003b; Rice et al., 2003). This Bt protein is called Cry3Bb1; it controls the rootworm larvae but not the adult beetle. As with Bt transgenic protection against corn borer, genetic yield advantage of rootworm resistance depends on the severity of infestation and its interaction with the environment. “Yield trials demonstrated that under heavy rootworm pressure and moisture stress the lack of corn rootworm larval injury in the [genetically engineered] corn resulted in substantially higher yields than [in] corn without the Bt protein. As rootworm pressure and moisture deficits declined, the yield advantage of . . . genetically engineered corn declined.” (Rice et al., 2003). Another similarity between the two kinds of Bt resistance is that refuge areas will be needed to delay the development of rootworm populations that are resistant to Cry3Bb1 (Rice et al., 2003). As with the corn borer Bt, Cry3Bb1 imparts vertical resistance and presumably will lose its effectiveness at some future date, thus necessitating replacement with a new genetic form (or forms) of resistance. b. Diseases. Frei (2000) stated that the minor presence of leaf diseases in northern Europe allows increased emphasis on selection for yield performance of maize hybrids for that area. This statement indirectly acknowledges that maize breeders in other regions must select hybrids with tolerance or resistance to locally prevalent diseases. The list of important diseases changes
108
DONALD N. DUVICK
from time to time, as new cultural methods and/or new genotypes encourage diseases that had been absent or relatively unimportant (Dodd, 2000; National Research Council (U.S.) Committee on Genetic Vulnerability of Major Crops, 1972; Tatum, 1971). Dodd (2000), speaking for maize in the United States, stated that during the past 40 years at least 14 diseases of maize have had significant increase in importance, although not all have endured or have proved to be widespread. Their emergence as a problem is often encouraged by changes in cultural practice, such as an increase in continuous maize growing and/or in minimum tillage. At other times, widespread planting of a single genotype will encourage spread of a particular disease. Breeders and farmers react promptly to new disease problems; susceptible hybrids are dropped in favor of resistant ones (if they are on hand) and further breeding ensures that new releases have the needed level of resistance to the problem disease(s). This battle will never end. Breeding for disease resistance shows its success (and yield-enhancing contribution) most clearly when the disease is active on susceptible hybrids (comparable to breeding for insect resistance). It would be difficult or impossible to plot gradual gains in yield due to gradual increases in disease resistance alone; nevertheless, the cumulative effects of successful breeding for disease resistance surely must contribute to the general increase in the level of on-farm yields. As Russell (1993) said, “Selection for disease resistance has been an integral component of maize breeding for many years, yet there are few data reflecting directly how the success of this selection affects grain yield.” However, he does note that “. . . improvement for stalk quality has been well documented . . . and stalk quality is highly dependent on plant health.” One must acknowledge that interactions of disease resistance traits with other beneficial genetic changes are perhaps the rule rather than the exception. For example, Clements et al. (2003) stated, “These results suggest that Bt transformation events like MON810 are a useful supplement to hybrid resistance to fumonisin contamination and fusarium [Fusarium spp.] ear rot.” They went on to say that such benefits (reduced borer damage and therefore less chance for disease entry) may accrue to susceptible hybrids but not to hybrids with a relatively high level of resistance to fusarium. The interactions of the two traits (disease resistance and insect resistance) determine the outcome. One cannot credit either trait by itself.
6.
Response to Changes in Plant Density
a. High Density. Genetic yield gain as a result of adaptation to continual increases in plant density is perhaps the most clear-cut and quantifiable change in maize hybrids over the years. Cardwell (1982) calculated that
ADVANCES IN MAIZE
109
increased plant densities contributed 21% of the gain in maize yield in Minnesota from 1930–1970. One must assume that not only the increase in plant density but also the introduction of maize genotypes that could withstand and profit from the higher densities was essential to achieving the gain. High plant density increases the deleterious effects of various kinds of stress —abiotic and biotic—and so increases the need for genetic improvements in stress tolerance (Troyer, 1996). Several examinations of U.S. hybrids, summarized by Russell (1991), showed that OPCs and old hybrids made their highest yields at lower densities typical of their era, whereas the newest hybrids yielded the most at the densities (always higher) typical of recent years. In other words, a hybrid usually gave the highest yield when grown at the density for which it was bred. Similar results have been shown for late-maturing hybrids in France and Ontario, Canada (essentially the same genotypes as grown in the United States), but not so for early maturing hybrids in France or Ontario (Derieux et al., 1987; Tollenaar et al., 1994). Tollenaar et al. (1994) suggested that because newer hybrids of the early maturity group have greater leaf area per plant as compared with the older hybrids of the same maturity, they do not respond to (or need) higher plant density. He stated that, in contrast, because the newer hybrids of the later maturity groups do not have increased leaf area per plant compared with older hybrids in their maturity group, the newer hybrids of the later maturity groups require more plants per hectare to increase leaf area (and thereby photosynthetic surface) per hectare. As mentioned in Section II.D.4, optimum plant density can be affected by the level of fertilizer N as well as by the hybrid genotype; hybrids differ within and between eras in their response to various combinations of N level and plant density (Carlone and Russell, 1987). A 1930–1991 time series of 36 hybrids and one OPC for Iowa (Duvick, 1997) showed the same general trends as in earlier trials of shorter versions of this series of hybrids (Duvick, 1977, 1984a, 1992); i.e., the older hybrids yielded more at lower densities typical of their era, whereas the newer hybrids yielded more at higher densities typical of their era. However, the newest hybrids in the 1930–1991 time series made only a very small gain in yield when planted at the highest density (79,000 plants ha1) as compared with their performance at the intermediate density (54,000 plants ha1). This suggests the possibility that future yield gains from breeding for adaptation to higher plant densities will come at a slower pace and/or will require more breeding effort, at least with the present breeding strategy. Breeders intending to increase genetic yield potentials may need to modify or replace current breeding strategies and/or selection criteria.
110
DONALD N. DUVICK
Figure 6 Grain yield per hybrid regressed on year of hybrid introduction at each of three plant densities: 10,000, 30,000, and 79,000 plants ha1. Best linear unbiased predictors (BLUPs) for hybrid grain yield based on trials grown in the years 1991–2001, three locations per year, one replication per density. From Duvick et al. (2004b). Copyright # 2004 by John Wiley & Sons, Inc. This material is used by permission of John Wiley & Sons, Inc.
b. Low Density. Duvick (1997) also showed that yields of the 1930–1991 series of hybrids did not significantly increase over time when the hybrids were planted at an extremely low plant density of 10,000 plants ha1. In this nearly stress-free environment, all hybrids were able to express maximum yield potential per plant (or at least a close approach) and, under these conditions, the older hybrids showed virtually as much yield potential per plant as the newer hybrids. A further report, with the hybrid time series extended to 2001 (Duvick et al., 2004b), showed essentially the same results (Fig. 6); breeders have not significantly increased yield potential per plant, even though they have greatly increased maize yield potential per unit area.
7.
Herbicide Tolerance
An experiment designed to compare eight maize hybrids representing three decades of yield improvement in Ontario, Canada, showed that the hybrids reacted differentially to the herbicide bromoxynil (4-hydroxy-3,5-dibromobenzonitrile)
ADVANCES IN MAIZE
111
(Tollenaar and Mihailovic, 1991; Tollenaar et al., 1994). The hybrids in this experiment, dating from 1959–1988, showed continuing improvement in tolerance to bromoxynil, with a statistically significant trend of decreasing phytotoxicity. They also showed continuing and significant improvement in grain yield, especially at higher plant densities. Commenting on these results, Tollenaar et al. (1994) suggested that improved antioxidant defense mechanisms may be associated with increased tolerance to bromoxynil and also with increased grain yield over the decades. They also said, “The small and, in particular, gradual nature of the increased bromoxynil tolerance suggests a highly complex, polygenic inheritance of this particular kind of stress tolerance.” Some maize hybrids now contain deliberately bred-in herbicide tolerance, primarily as transgenic resistance to broad-spectrum herbicides such as glyphosate (e.g., Hetherington et al., 1999). Herbicide-tolerant maize covered about 15% of the U.S. maize acreage in 2003 (ERS, 2003) and in 2002 was planted on about 4% of all land planted to transgenic crops globally (James, 2002). Strictly speaking, herbicide tolerance is intended to improve the efficiency of weed management and not necessarily to increase maize productivity, although better weed control could indirectly result in higher yields, if weed levels were high with other kinds of management. In extreme cases, herbicide tolerance/resistance can increase maize yield significantly and in strikingly large amounts. For example, initial experiments in several African countries indicated that when maize is bred to be resistant to a herbicide that normally is toxic to maize, seed coated with that herbicide can provide effective season-long control of Striga spp. (Kanampiu et al., 2003). Striga, a parasitic weed (sometimes called “witchweed”), can cause devastating crop loss in maize as well as other grain crops in those countries. The herbicide resistance (imadazolinone resistance, “IR”) is nontransgenic; it results from a mutation in an acetolactate synthase gene. The specific herbicides used in these experiments were imazapyr and pyrithiobac. When Striga density was high, the herbicide treatment resulted in a three- to fourfold increase in yield. Kanampiu et al. (2003) stated, “When the IR gene is incorporated into locally adapted varieties as in Kenya, this can result in improvements in maize growth and hence high maize yield benefits to small-scale farmers.” Herbicide tolerance, indirectly, can produce undesired results and lower yields. King and Hagood (2003) showed that postemergence control of johnsongrass with glyphosate increased the severity of maize chlorotic dwarf virus and maize dwarf mosaic virus in glyphosate-tolerant hybrids that were susceptible to those diseases. They said, “The increased disease severity resulted from greater transmission by insect vectors, which moved from dying johnsongrass to the crop.” However, disease severity did not
112
DONALD N. DUVICK
increase in a virus-tolerant (and glyphosate-tolerant) hybrid subjected to the same conditions. The authors concluded that for fields infested with johnsongrass, the hybrid choice should be primarily for disease resistance and secondarily for herbicide resistance.
8.
Other Physiological Traits
a. Photosynthesis. As noted in earlier sections, leaf photosynthesis seems to be more efficient in newer hybrids than in older hybrids when they are compared in a range of stress conditions such as drought, low temperature, or low N supply. Such an increase in efficiency could help maize plants recover more rapidly from transient stresses such as those induced by cold weather, overly wet soils, or drought. As summarized by Tollenaar et al. (p. 215, 1994), “[These] findings, some of them preliminary in nature, suggest that although hybrid differences in leaf photosynthesis under unstressed conditions may not be indicative of actual or potential yield, hybrid differences in response of leaf photosynthesis to stress conditions may be a useful physiological indicator of high stable yields. To date, selection for yield per se has apparently provided a selection pressure in favor of stress-tolerant leaf photosynthesis.” b. Canopy Gas Exchange, Temperature. Nissanka et al. (1997) compared an old and a new hybrid (1959 vs 1988) in regard to whole plant gas exchange and stem water potential throughout a water-deficit stress cycle and during the subsequent recovery period upon rehydration. Under moisture stress, the new hybrid maintained relatively higher rates of photosynthesis and transpiration at a lower stem water potential than the old hybrid. During the recovery day, canopy photosynthesis was 53% higher and canopy transpiration was 31% higher in the new hybrid than in the old hybrid. Respiration per unit CO2 fixed was lower in the new than in the old hybrid in all conditions. The authors concluded that the new hybrid was more tolerant to water stress and recovered faster upon rehydration than the old hybrid. (As noted previously, comparison of only two hybrids is not evidence of a long-term trend, but it may give useful suggestions for further investigation.) Canopy temperature under drought stress consistently decreased, going from older to newer hybrids, in measurements of a set of 18 commercial hybrids adapted to central Iowa and representing the period 1953 to 2001 (Barker et al., 2005). The hybrids were grown in a rain-free environment in Chile and were subjected to managed drought stress at various stages of development. Barker et al. (2005) suggested that the trend to lower canopy temperature under drought stress may result from the lower
ADVANCES IN MAIZE
113
radiation intensity on the more upright leaves of modern hybrids or on a greater capacity by these hybrids to capture soil water.
9.
Parentage and Genetic Diversity
As new hybrids replace those that preceded them, pedigrees and genotypes change. It can be instructive to know more about the nature of these changes. Which founder inbreds and/or OPCs remain in the pedigrees of current hybrids and which ones disappear? Has genetic diversity decreased, increased, or stayed about the same over the years? In a limited way some of these questions were answered by Duvick et al. (2004a,b) with respect to the time series of Iowa hybrids used for their studies of changes in hybrid performance over time. For example: • The 51 hybrids in the trials traced back to 53 founder sources: OPCs, synthetic populations, and inbred lines. The founders came primarily from the U.S. corn belt but a few were from the southeastern and northeastern United States or from Latin America. • Some families have persisted over the years and have contributed relatively large amounts (by pedigree) to present-day hybrids, whereas others appeared in pedigrees for only a few decades and then declined or disappeared. For example, Reid Yellow Dent, Iowa Stiff Stalk Synthetic, and Reid Iodent have been important contributors since they first appeared in pedigrees and they now contribute 33, 22, and 15%, respectively, to hybrids of the 2000s. However, Maryland Yellow Dent, Boone Country White, and the inbred Hy contributed briefly to pedigrees in the 1950s and 1960s but then disappeared completely from pedigrees of subsequent hybrids in the time series. Krug reached a peak use of 23% in the 1940s but declined rapidly to a steady level of about 3%, and some families appeared late but have made significant, if not large, contributions; for example, Argentinian Maiz Amargo appeared in the 1980s and has contributed 4 to 5% in each of the past two decades. • The pedigree information shows, therefore, that although certain lineages have predominated over the years of hybrid improvement and replacement, they have been supplemented significantly by additional, and diverse, lineages. On the whole, pedigree contributions have been broad and volatile; the record provides strong evidence for “genetic diversity in time” [as defined by Duvick (1984b)] in this 70-year time series of 51 hybrids for central Iowa. Genetic diversity, including genetic diversity in time (also called temporal genetic diversity), is widely acclaimed for its ability to provide beneficial genetic response (timely and wide-ranging) to new and/or unusual kinds of biotic or abiotic stress (FAO, 1996; Gollin and Smale, 1998;
114
DONALD N. DUVICK
National Research Council (U.S.) Committee on Genetic Vulnerability of Major Crops, 1972; Rosenow and Clark, 1987; Simmonds, 1962; Smale et al., 2002).
10.
Molecular Markers
Pedigree data, although informative, do not identify genetic materials in the pedigree lineages, either qualitatively or quantitatively. Molecular marker data, tracing a given DNA fragment from one generation to the next, can enable quantification of the amount of founder germplasm that persists in successive generations. Application of this technology to the Iowa time series of hybrids (Duvick et al., 2004a,b) using simple sequence repeats (SSR) showed that • The number of alleles fluctuated from decade to decade, with about 40 to 50% of the total number of alleles present in any one decade. The study identified 969 alleles at 100 SSR loci in the array of hybrids and OPCs. • The number of alleles per locus was similar for the female and male parents of hybrids. • Large-scale turnover of alleles took place in the first decades (1930s and 1940s) of hybrid breeding (which agrees with pedigree data), indicating that many of the first inbred lines were not useful as parents for the next generation of breeding. They were dropped from breeding pools, and new and more successful breeding materials were brought in. • The initial large-scale turnover of alleles was followed by a relatively steady state of replacement until the 1970s (corresponding with the changeover from double cross to single cross hybrids) when the number of new alleles per decade again declined to what may be a new and lower steady state of replacement. • The alleles of the inbred parents of the modern hybrids (primarily single crosses) could be separated (with multidimensional scaling analysis) into two groups (called “stiff stalk” and “nonstiff stalk”), whereas alleles of the older hybrids (primarily double crosses) sorted into an undifferentiated third group (Fig. 7). This confirms the general observation that breeders (using pedigree and experiential data) have established two breeding pools to balance important traits (including those that enable use of inbreds as either seed parent or pollen parent) and/or to maximize heterosis when inbreds from the two groups are crossed with each other. The primary goal of such divergent selection is to enable the development of economically producible hybrids with improved on-farm yield and general performance.
ADVANCES IN MAIZE
115
Figure 7 Scores for 94 inbreds contributing to Era hybrids on the first two dimensions of the multidimensional scaling analysis of SSR polymorphism data for 298 SSR loci (R2 ¼ 0.45 for the two-dimensional model). From Duvick et al. (2004b). Copyright # 2004 by John Wiley & Sons, Inc. This material is used by permission of John Wiley & Sons, Inc.
11.
Heterosis: Hybrid and Inbred Performance
a. Heterosis for Grain Yield. The phenomenon of heterosis in maize stimulated the initial research that led to the development and introduction of hybrid maize (e.g., Crow, 1998; Hayes, 1952; Shull, 1952), and maize breeders sometimes seem to have considered that breeding for higher yield is synonymous with breeding for increased heterosis. Few researchers have gathered data, however, to measure the degree to which the proportionate contribution of heterosis to grain yield has changed over the years of hybrid breeding (Duvick, 1999). b. Absolute Heterosis. Schnell (1974) summarized data comparing single cross yield and parental inbred yield from 17 experiments designed for other purposes. He stated that heterosis for the decades 1920–1970 showed “only a modest increase . . . as compared to the large simultaneous increase in the yields of inbreds. . . .” Schnell refers here to “absolute
116
DONALD N. DUVICK
heterosis,” the difference between the yield of a single cross and the mean yield of its parental inbreds (midparent yield). The amount of annual increase in midparent yield was nearly as great as that of the single cross hybrids. Schnell also calculated heterosis as “relative heterosis” (absolute heterosis divided by single cross yield) and noted that it decreased from 75% in the 1920s to about 50% in the 1970s. This is because the denominator (single cross yield) increased at a faster rate than the numerator (absolute heterosis). Meghji et al. (1984) studied changes in heterosis for inbreds and their single crosses representing three decades (1930s, 1950s, and 1970s) of U.S. corn belt hybrid germplasm. Six inbreds per decade represented the 1930s and 1950s (two inbreds, WF9 and Os420, were in both decades) and seven inbreds represented the 1970s (four of the six single crosses for the 1970s contained the inbred Mo17). The trials were grown in Illinois; the year(s) of the trial is not indicated. Yields of inbreds and single crosses (averaged across densities) increased simultaneously over the decades. Absolute heterosis also increased over the decades; the increase averaged 51 kg ha1 year1 (Experiment 1, Table II). The increase was greater at a high density typical of the 1970s than at a low density typical of the 1930s. Duvick (1984a) compared inbreds and single crosses representing successful Iowa hybrids for the decades 1930–1970. Five unrelated inbreds for each decade were crossed in all possible combination to give 10 single crosses per decade; trials were grown in 3 years (1977–1979) at three densities Table II Contributions of Absolute Heterosis and Relative Heterosis to Grain Yield in Three Experimentsa 1930s Experimentb
Categoryc
1
SX yield Het Abs Het (%) SX yield Het Abs Het (%) SX yield Het Abs Het (%)
2
3
a
1940s
1950s
1960s
1970s
1980s
1
bd
kg ha 7097 4112 (58) 4600 2700 (59) 5941 3787 (64)
5300 3200 (60) 6371 3754 (59)
7407 4438 (60) 6900 4100 (59) 6865 4143 (60)
7000 3600 (51) 7174 4188 (58)
9538 6138 (64) 7900 4300 (54) 7929 4102 (52)
61 51 83 36 9164 4658 (51)
60 13
Adapted from Duvick (1999), with permission from American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America. b Experiment 1: Means of two densities, data from Meghji et al. (1984). Experiment 2: Means of three densities grown in 1977–1979, data from Duvick (1984). Experiment 3: Means of three densities grown in 1992–1993, data from Duvick (1999). c SX yield, single cross yield; Het Abs, absolute heterosis; Het (%), relative heterosis. d Linear regression coefficient (kg ha1 year1).
ADVANCES IN MAIZE
117
(30,000, 47,000, and 64,000 plants ha1) representing densities of the 1930s, 1950s, and 1970s. Yields of inbreds and their single crosses (averaged across densities) increased simultaneously in each decade. Absolute heterosis increased in each decade, except in the 1960s. The increase averaged 36 kg ha1year1 (Experiment 2, Table II). In a second experiment, Duvick compared inbreds and single crosses representing successful Iowa hybrids for the decades 1930–1980. Results of the experiment are reported in Duvick (1999) and Duvick et al. (2004b). Seven representative single crosses per decade were compared with their parental inbreds during 2 years (1992 and 1993), at three locations per year, and three densities (30,000, 47,000, and 64,000 plants ha1) per location. In 1993, a very wet year, one location was lost because of flooding. Yields of inbreds and their single crosses (as averaged across densities and years) increased simultaneously and by nearly the same amount in each decade (Fig. 8). Absolute heterosis (SX–MP) increased minimally
Figure 8 Yields of single crosses (SX), their inbred parent means (MP), and heterosis (as SX – MP). Single cross pedigrees are based on heterotic inbred combinations in Era hybrids during the six decades, 1930–1980, 12 inbreds and six single crosses per decade. Means of trials grown in three locations in 1992 and two locations in 1993 at three densities, 30,000, 54,000, and 79,000 plants per hectare, one replication per density. From Duvick et al. (2004b). Copyright # 2004 by John Wiley & Sons, Inc. This material is used by permission of John Wiley & Sons, Inc.
118
DONALD N. DUVICK
(13 kg ha1year1) during the six-decade period (Fig. 8 and Experiment 3, Table II). However, the two growing seasons gave contrasting results. Absolute heterosis averaged over densities was constant over the decades in the 1992 trial, but it increased significantly (32 kg ha1 year1) in the 1993 trial (Fig. 9). The year 1992 was an optimal growing season with very high yields; whereas 1993 was a high-stress, low-yield year, with extremely wet and cool growing conditions. The trials of this experiment also exhibited contrasting outcomes when grown at different plant densities. Absolute heterosis, averaged over the two seasons, showed no trend over the decades at the lower density, increased slightly but unevenly at the medium density, and increased consistently (b ¼ 32 kg ha1 year1) at the higher density (Table III). Data from this experiment show, therefore, that yielding ability under stress has been improved to a greater degree in hybrids than in inbreds, even though (as shown in this and in other experiments) yield under stress is greatly improved over time for both of the categories (inbreds and hybrids).
Figure 9 Heterosis (as SX – MP) in two seasons: 1992 and 1993. Single cross pedigrees based on heterotic inbred combinations in Era hybrids during the six decades, 1930–1980, 12 inbreds and six single crosses per decade. Trials were grown in three locations in 1992 and two locations in 1993. Means of three densities: 30,000, 54,000, and 79,000 plants ha1, one replication per density. From Duvick et al. (2004b). Copyright # 2004 by John Wiley & Sons, Inc. This material is used by permission of John Wiley & Sons, Inc.
ADVANCES IN MAIZE
119
Table III Interaction of Plant Density with Decadal Changes in Absolute Heterosisa 1930s Densityb
Categoryc
Low
SX MP Abs. Het SX MP Abs. Het SX MP Abs. Het
Medium
High
1940s
1950s kg ha
6717 2062 4655 6569 2337 4232 5708 2308 3400
6703 2373 4330 7033 3065 3968 6648 3003 3645
7099 2395 4703 7960 3174 4786 6823 3063 3760
1960s
1970s
1980s
1
7387 2658 4730 8171 3493 4679 7192 3390 3802
bd 7187 3309 3877 8747 4352 4396 9098 4463 4635
8174 3875 4298 10024 4969 5055 10492 5476 5016
26 35 8 64 50 15 90 59 32
a
Adapted from Duvick (1999), with permission from American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America. Data for means of seven single crosses and corresponding midparents per decade grown in five locations over 2 years (1992 and 1993). b Plant densities: Low, 30,000 plants ha1; medium, 54,000 plants ha1; and high, 79,000 plants ha1. c SX, single cross; MP, midparent; Abs. Het, absolute heterosis. d Linear regression coefficient (kg ha1 year1).
These results agree with those of Meghji et al. (1984), who stated that the increase of absolute heterosis over the decades was greater at higher than at lower plant density. An intriguing area of research might be to look for the genetic and physiological changes that have accompanied the increases in absolute heterosis under stressful growing conditions. Can the needed genetic combinations be present only in heterozygous individuals or can they be gradually gathered into a single genome? Discoveries of intraspecific violation of genetic colinearity in maize (Fu and Dooner, 2002) may have implications for this area of investigation. To review this section, limited data indicate that absolute heterosis for grain yield has increased over the years to a small extent (more so under abiotic stress) but that its annual gain is less (sometimes much less) than total genetic gain in hybrid yield. Of course, one must recognize that one way to increase absolute heterosis would be to reduce inbred yields without increasing hybrid yields; this is not a desirable outcome. The current situation seems to be that yields of inbreds have increased in line with those of their hybrids, but at a slightly slower rate such that absolute heterosis is gradually increasing, especially when plants are grown under stress. c. Relative Heterosis. Calculations of relative heterosis (absolute heterosis as percentage of single cross yield) for the aforementioned three experiments indicate that although absolute heterosis increased over time
120
DONALD N. DUVICK
in most comparisons, relative heterosis did not increase in the two Iowa experiments (Experiments 2 and 3, Table II), although it did increase in the Illinois experiment (Experiment 1, Table II). Relative heterosis in the Iowa experiment of 1992–1993 (Experiment 3, Table II) declined from 64% in the 1930s to 52% in the 1970s and 51% in the 1980s. Although absolute heterosis increased minimally over the decades in this experiment, the gain in single cross yield was greater than the gain for absolute heterosis, and so, as with Schnell data, the proportionate contribution of heterosis to grain yield was reduced over time. Interestingly, also, Schnell’s estimate of 50% for relative heterosis in the 1970s corresponds closely with values of 54 and 52% for the 1970s in the two Iowa experiments. Data summarized in this section suggest that relative heterosis for grain yield has not increased markedly over the years; it more likely has stayed constant or declined. d. Heterosis for Other Traits. Plant height and flowering date exhibit heterosis to a large degree in maize; crosses between two inbreds are always taller and earlier than the mean of the parents. Heterosis for these traits may (or may not) be related to some of the genetic interactions that produce heterosis for grain yield and so it may be informative to examine changes over time in heterosis for plant height, or other plant size measurements, and flowering date. Mean values for inbreeding depression in ear height and plant height decreased over time, in comparisons made by Meghji et al. (1984). This indicates that absolute heterosis for ear height and plant height decreased over time. In the same experiment, means for inbreeding depression of tassel weight and tassel branch number increased in the 1950s but decreased in the 1970s to levels lower than in the 1930s. This would indicate reduced heterosis for tassel weight and tassel branch number in the 1970s genotypes. Small but statistically significant trends toward reduced absolute heterosis for plant height (3 cm 10 year1), ear height (4 cm 10 year1), and heat units to anthesis (11 heat units 10 year1) were exhibited in the aforementioned comparison of inbreds and single crosses representing successful Iowa hybrids for the decades 1930–1970 (Duvick, 1984a). Absolute heterosis was reduced to a small degree for plant height but was not reduced for heat units to anthesis in the aforementioned comparison of inbreds and single crosses representing successful Iowa hybrids for the decades 1930–1980 (Duvick et al., 2004b). Data from these experiments indicate that, to a small degree, the size and maturity differences between inbreds and their hybrid progeny were reduced over time. However, the differences between the two classes remain (and probably will remain) large. For example, reexamination of data for plant
ADVANCES IN MAIZE
121
height heterosis summarized in Duvick et al. (2004b) shows that the height difference between the two classes was reduced from about 85 cm in the 1930s to about 70 cm in the 1980s, primarily because of a reduction in height of the single crosses. If this rate of reduction in height difference could be maintained (0.3 cm year1 or 15 cm 50 year1), it would take about five more 50-year cycles of selection to equalize inbred and hybrid height. Data for changes over time in heterosis for plant size and maturity agree in one respect with those for grain yield—neither category has exhibited major increases (or decreases) in absolute heterosis. However, absolute heterosis for grain yield has increased to a small degree (at least, under stress), whereas absolute heterosis for plant size and maturity has decreased to a small degree. Superficially, the two categories of heterosis do not seem to answer to the same genetic stimuli.
III. GENETIC GAINS FROM POPULATION IMPROVEMENT A. COMPARISONS
WITH
GENETIC GAINS IN HYBRIDS
Although the emphasis of this review is on maize hybrids and how successive changes in their breeding and genetics have contributed to increased on-farm yield, recurrent selection to make improved populations has interacted with and sometimes contributed to genetic improvements in hybrids (e.g., via useful inbred lines bred from improved populations). Additionally, for some farmers in some parts of the world, annual purchase of hybrid seed is not an option. For these people, improved populations, maintained by saving seed, are the only practical option for access to improved cultivars. Therefore, this section briefly summarizes genetic gains achieved by recurrent selection for population improvement in comparison with genetic gains in hybrid performance and comments briefly on the contribution of improved populations to hybrid maize yield. A comparison of genetic gain for four recurrent selection experiments with genetic gain for two time series of hybrids indicated that both methods produced about the same annual genetic gain for grain yield: 71 kg ha1 year1 for recurrent selection and 68 kg ha1 year1 for the hybrids (Duvick, 1977). Duvick described the two sets of experiments as follows: “Both took place in central Iowa; both occurred in about the same time frame; both had as a primary goal maximum improvement in yield.” For these reasons he thought it appropriate to compare the outcomes of the two kinds of breeding program. He also suggested, however, that selection pressure in the recurrent selection programs might have placed less weight on nonyield
122
DONALD N. DUVICK
traits (e.g., root and stalk lodging) than was the case in the hybrid breeding programs, thus allowing greater progress for yield per se in the recurrent selection programs. So the programs were not completely equivalent in selection goals. Duvick (1977) called for more comparisons of recurrent selection with hybrid breeding (which typically is based on pedigree breeding), suggesting that “Maize breeding probably could be helped by the results of quantitative genetics studies specifically designed to compare ‘recurrent selection using the pedigree method’ and ‘recurrent selection using the population pool method.’ Good data, demonstrating the strong and weak points of these two related and proven methods, would help the entire hybrid maize breeding effort in its goal of producing good hybrids as quickly and efficiently as possible.” This request was easier to make than to grant, however, and to the author’s knowledge, no such paired breeding programs have been designed and executed. Even to compare estimates of genetic gain for hybrids with estimates of genetic gain for recurrent selection is difficult because most reports of progress in recurrent selection express yield gains in units cycle1 rather than units year1, and number of years per cycle usually is not stated and must be inferred (if possible) from descriptions of the breeding cycle. However, Edmeades and Tollenaar (1990) summarized 17 estimates of genetic gain in temperate environments and 10 estimates of genetic gain in tropical environments, with all estimates expressed as kg ha1 year1. With one or two exceptions, the estimates for temperate environment are for time series of commercial hybrids and the estimates for tropical environments are for recurrent selection programs intended to produce improved populations. Genetic gain in grain yield for the temperate programs averaged 66 kg ha1 year1, whereas genetic gain for the tropical programs averaged 145 kg ha1 year1. The average gain for the temperate experiments is identical to the average from the previously mentioned summary by Russell (1991). This result is not too surprising because for the most part the two lists cite the same reports, although the two lists are not identical. The list of tropical experiments shows a broad range of estimates, from 51–310 kg ha1 year1. Edmeades and Tollenaar (1990) explained the high values for the estimates of the tropical programs (primarily recurrent selection) as follows: “The higher average rate of gain reported from the tropics is generally the result of one, sometimes two and very occasionally three selection cycles per year, the use of relatively unimproved germplasm with a broad genetic base, and a selection scheme that was based on family performance.” Although it is probably true that the first gains are the easiest in any professional breeding program that starts with landrace materials, it also seems likely that some of the high gains reported for the recurrent selection programs in
ADVANCES IN MAIZE
123
the tropics result from effective selection technology for important traits such as drought tolerance (e.g., Ba¨ nziger et al., 1999; Bolan˜ os and Edmeades, 1992), combined with rapid turnaround of breeding generations. Edmeades and Tollenaar (1990) presented evidence for the possibility that first gains are the easiest. They said that “. . . tropical maize . . . in its unimproved state is tall, leafy, lodging-prone and has a harvest index of about 0.35.” They stated that initial selection in unimproved populations in the tropics results in shorter plants and reduced lodging, improved HI, and reduction in barrenness. The present review has shown that temperate hybrids have shown little or no change over the years in plant height, leaf number, or HI when hybrids are grown at the plant density for which they were bred. This may indicate that farmer selectors had already changed these traits to a close approximation of “optimum” levels when they developed the corn belt OPCs that were the basis of hybrid breeding. However, the first hybrids clearly surpassed parent OPCs in root and stalk strength and in resistance to barrenness. Although the stress of constantly increasing plant density continues to sort out hybrid genotypes with increasingly greater resistance to lodging and barrenness, these improvements are not as dramatic as those of the initial hybrids compared with their parent OPCs. Taking into account these changes (or lack thereof) in temperate breeding programs, one might predict that annual yield gains of population improvement programs for tropical materials eventually will move to lower (but nevertheless acceptable) levels, similar to those for hybrid improvement in the temperate zones. Following the relatively easy gains resulting from initial reductions in plant size and lodging and from increases in HI, yield gains will need to come from more gradual improvements in tolerance to locally important kinds of abiotic and biotic stress. Leaving aside comparisons between hybrid breeding and recurrent selection, numerous published reports testify clearly that recurrent selection for grain yield and/or general performance can give positive results in temperate as well as in tropical materials, and it can do so with acceptable investments of time and effort (e.g., Hallauer and Miranda, 1988; Lamkey, 1992; Russell, 1991; Sriwatanapongse et al., 1985). Several methods have been tried; some worked well and some did not (e.g., Edwards and Lamkey, 2002; Lamkey, 1992), much as has been true for various kinds of pedigree breeding applied to hybrid development. One difference between recurrent selection and hybrid breeding has been that breeders sometimes have used recurrent selection primarily to intensify expression of a single trait such as resistance to ECB or high grain oil percentage (Klenke et al., 1986; Sprague, 1952). They often discovered that other important traits such as grain yield or stalk digestibility could deteriorate, sometimes as an indirect response to strong selection for the intensified
124
DONALD N. DUVICK
trait or sometimes simply because of insufficient selection pressure for the other important trait(s) (Hallauer and Miranda, 1988; Ostrander and Coors, 1997; Russell, 1991). Such populations could have only limited value as sources of commercially useful inbred lines because inbreds (and the hybrids they produce) must be as well balanced as possible for a full selection of important traits. Transfer of the improved trait to useful inbred lines may require a long-term effort or may not even be worth doing if its intensification depends on the deterioration of another important trait.
B. RELATIVE CONTRIBUTIONS OF POPULATION IMPROVEMENT AND PEDIGREE BREEDING “Pedigree selection is the most widely used breeding method to develop inbred lines for use as parents of hybrids.” . . . “Pedigree selection will always be an important component of modern corn-breeding programs (Hallauer et al., 1988, pp. 470–471),” “Pedigree selection was and is the most commonly used selection method of line development (Hallauer and Miranda, 1988, p. 10).” These statements, although in agreement with the general experience of hybrid maize breeders, should not be construed to say that recurrent selection for population improvement has not made vital contributions to hybrid maize breeding. Inbred lines such as B14, B37, and B73 have, in their time, been parents of hybrids that dominated maize plantings in the U.S. corn belt, and they also have been important sources of germplasm for further breeding by pedigree selection. These inbreds are direct products of the previously described population, BSSS (Russell, 1991). Other elite inbreds derived from improved populations could be named as well (e.g., Hagdorn et al., 2003), although their impact has been more limited than that of the “big three.” Nevertheless, as stated earlier, pedigree breeding has been and remains the backbone of hybrid maize breeding. [See Troyer (1996) for a detailed and first-hand description of pedigree breeding in action.] Planned single crosses, successful commercial hybrids, and crosses (or backcrosses) of an elite line to an improved population are typical starting points for pedigree selection to produce new inbred lines. The author knows of no detailed, data-based exposition of the reasons for predominance and persistence of pedigree breeding to develop inbred lines and so can only speculate about reasons for this situation. Perhaps one reason for the predominance of the pedigree method is that its parental materials—inbred lines—have been very widely tested, not only in breeders’ performance trials, but also by thousands of farmers. Breeders typically have started selfing and pedigree selection from crosses of inbreds (or their progeny) that were used in widely successful hybrids. Each hybrid can be considered as a “test cross” for its parental inbreds. An unexpected
ADVANCES IN MAIZE
125
weakness in one or both parents is more likely to be discovered when a hybrid is grown by thousands of farmers over a period of years, as compared with growing an improved population in a relatively small number of yield trials for one or two seasons. Conversely, one also can be more confident of identifying genotypes that can give top performance over a wide variety of environments if the replication number is in the thousands rather than in the dozens (or fewer). (Commercial seed companies have a unique advantage here in that they can gather trial data from hundreds or thousands of onfarm “strip trial” comparisons. Although these data may be intended primarily for potential use in sales promotion, they also provide the company’s breeders with a deep mine of information about comparative performance of a given genotype over a broad range of environments.) A second consideration is that the odds of getting a superior inbred—with good balance for all traits—from selfing a cross of two currently topperforming inbreds are probably higher than the odds of getting a line of equivalent merit from selfing an excellent but relatively heterogeneous improved population, even though truly superior (and novel) genotypes may well exist in the improved population. This second possibility leads to further speculation that a new program (such as some of those in the tropics) may have greater success in extracting useful inbred lines from improved populations than has been exhibited in mature temperate breeding programs. One reason could be that some of today’s improved populations for the tropics are very high quality; they have benefited from years of experience in designing and/or choosing proper environments for performance trials and in the development of improved designs for recurrent selection. The potential utility of such populations (e.g., for production of stress tolerant hybrids) has been predicted by Edmeades et al. (1997), who said “The probability of obtaining a hybrid that yielded 40% greater than the trial mean under severe stress was 4-fold greater when lines were extracted from a drought-tolerant source population than from its conventional counterpart. . . . We conclude that droughtor N-tolerant elite source populations provide a greater proportion of drought- or N-tolerant inbred lines and hybrids.” A second reason could be that there will be few or no truly superior adapted inbred lines for a new breeding locality. If improved populations (improved for the traits needed in the area where hybrids are to be grown) are on hand, they may be the best available source of germplasm for the development of inbred lines. They will be better than local OPCs and better than crosses of unadapted inbreds that may be “elite” elsewhere but are not well suited to the local environment. (This scenario contrasts with the course of events in the U.S. corn belt, where two or three decades of pedigree breeding had brought out many superior inbreds by the time improved populations from recurrent selection programs were ready for use.)
126
DONALD N. DUVICK
In years to come, pedigree selection based on planned crosses of inbred lines may become more competitive in new breeding localities (such as in the tropics); however, further expertise in population improvement via recurrent selection may enable continued development of populations that are valuable sources of elite inbred lines for those environments. A final speculation about the utility of improved populations is that as expertise grows in molecular biology, breeders may learn how to “mine” improved populations for valuable genes or gene combinations with more precision and better odds of success than can be done with current methods of breeding and selection. The improved populations could provide a wider range of useful genetic diversity than can be achieved with pedigree breeding and would have the diversity in much more useful forms than exist in unadapted exotic cultivars and landraces. These latter materials are high in genetic diversity but are also high in “useless” genetic diversity. Remarks from Hallauer et al. (1988, pp. 531–532) make an appropriate conclusion for this section. They said “Recurrent selection and pedigree selection . . . should not be considered in opposition to one another. Rather, the two systems should complement each other. The goals of the two systems are different, but the ultimate objective is the same—contribute to genetic gain.” It seems safe to predict that as experience and technology progress, breeders will find increasingly profitable ways to utilize recurrent selection for development of a genetically diverse assortment of superior inbred lines that can be parents of successful hybrids.
IV. ANALYSIS AND CONCLUSIONS A. POSSIBLE REASONS
FOR
GENETIC YIELD GAINS
This review has shown that hybrid maize breeders have consistently increased the yielding ability of hybrids during the past 70 years and that genetic gains in grain yield are still linear. As the yielding ability of the hybrids has increased, other traits have changed as well, in directions that were sometimes intended and sometimes unintended or at least unplanned. Conversely, some traits have not changed (or have changed very little), sometimes at the intention of the breeders and sometimes despite the breeders’ intent to make a change. It will be instructive to categorize the various trait changes (or stabilities) as described in Section II.D. They can be categorized as (1) trait changes that promote the efficiency of grain production, (2) trait changes that increase tolerance to biotic and abiotic stress, (3) intended trait stabilities, and (4)
ADVANCES IN MAIZE
127
unintended (or unplanned) trait stabilities. The various trait changes are briefly summarized and arbitrarily sorted into these four groups as follows.
1.
Trait Changes that Promote Efficiency of Grain Production
• Leaf angle has become significantly more upright, especially since about the 1960s. • Tassel size has been markedly reduced. • Newer hybrids have longer period of grain fill but faster dry down; therefore they are not later in harvest maturity and make better use of the latter part of the growing season. • Kernel weight is greater in newer hybrids except under drought stress at late or terminal periods of grain fill. • Newer hybrids have lower percentage grain protein. • Newer hybrids have higher percentage grain starch. • In some experiments, newer hybrids are markedly more responsive to favorable environments (they make more efficient use of bountiful inputs), although results are not consistent in this regard.
2.
Trait Changes that Increase Tolerance (or Exhibit Evidence of Increased Tolerance) to Biotic and Abiotic Stress
• Grain yield has increased in linear fashion; increases are greatest at high plant density and are exhibited in high stress as well as low stress environments, in poorly fertilized as well as in well-fertilized environments. • Leaf rolling during drought stress is increased, perhaps because of changed leaf orientation; potentially this can help maintain lower leaf temperature and reduce water use. • Staygreen (resistance to stress-induced premature death) is markedly improved. • Anthesis-silking interval is shortened, especially when hybrids are subjected to conditions of abiotic stress such as drought or high plant density. • Newer hybrids show increased resistance to barrenness when trials are subjected to abiotic stress such as drought at flowering time or higher plant density. • Newer hybrids have higher HI than older ones if trials are subjected to biotic stresses that induce barrenness. • Hybrids show linear improvements in resistance to root lodging, although some trials indicate a ceiling at about 95% nonroot-lodged plants.
128
DONALD N. DUVICK
• Hybrids show linear improvements in resistance to stalk lodging, although some trials indicate a ceiling at about 95% nonstalk-lodged plants. • Hybrids show linear improvement in yield in seasons with above-average temperature during the growing season. • Hybrids show linear improvement in yield in seasons with below-average temperature during the growing season. • Newer hybrids are more drought tolerant. • Newer hybrids are more tolerant of excessive soil moisture (water-logged soils). • Newer hybrids are more tolerant of soil nitrogen deficiency. • Newer hybrids are more tolerant of unspecified abiotic stresses (“lowyield” sites). • Newer hybrids are more tolerant of ECB2, and, recently, transgenic hybrids have expressed a sharply increased level of resistance to both generations of European corn borer and (separately) to two species of rootworm. • Circumstantial records show that new kinds of disease resistance are added continually in response to new disease problems, but the contributions to the yield of sequential changes in disease resistance are not documented. • Newer hybrids, more tolerant of the stresses of higher plant density, enable the use of higher plant density to maximize yield and therefore the grain yield potential per unit area is increased. • Newer hybrids show increased tolerance to a specific herbicide, apparently correlated with increased antioxidant defense mechanisms; hybrids can be bred (using conventional genetics) to be resistant to another specific herbicide; and in recent years transgenesis has been used to impart tolerance to a third herbicide. • A newer hybrid has more efficient photosynthesis than an older one, especially under stress, and shows an improved capacity to recover the photosynthetic rate after stress. • A newer hybrid has more efficient canopy gas exchange, stem water potential, transpiration, and respiration than an older hybrid when plants are subjected to water stress. • Canopy temperature under drought stress is decreased. • Pedigree contributions have been diverse and are in a state of constant change, although with a steady core of persistent lineages. The consequent increase in genetic diversity (in time and in place) theoretically will provide an increased stability of performance in the presence of diverse biotic and abiotic stresses. • Molecular marker studies validate the pedigree information and show, further, that parent inbreds in recent years of the hybrid time series can be divided into two genetically different groups called stiff stalk and nonstiff stalk.
ADVANCES IN MAIZE
129
• Absolute heterosis for grain yield usually has increased to a small degree, with the greatest increase when trials are grown under stress of high plant density or drought.
3.
Intended Trait Stabilities
• Plant and ear height have been relatively stable with a weak trend to lower plant and ear height. • Anthesis date is relatively unchanged. • Date of silk emergence is unchanged over time in absence of stress, but newer hybrids silk earlier than older hybrids when trials are subjected to abiotic stress such as drought or high plant density; this is because silk emergence is delayed (or fails entirely) in the older hybrids when subjected to these kinds of abiotic stress. • Tillering ability is rarely expressed at modern plant densities; tillering is slightly reduced at low plant density.
4. Unintended (or Unplanned) Trait Stabilities • Leaf number is unchanged. • LAI is unchanged in the U.S. corn belt, but may be increased in early maturity regions of Ontario (Canada). • Number of ears per plant is not changed in absence of stress. • HI of temperate hybrids is not changed if hybrids are grown at the density for which they were bred. • Yield potential per plant is not increased; i.e., newer hybrids do not yield more than older hybrids at super-low plant density in absence of abiotic stress. • Relative heterosis for grain yield has not increased and, in some cases, is slightly reduced. • Absolute heterosis for plant height, ear height, and flowering date has not increased and, in some cases, is slightly reduced. Comparison of the categories in this summary shows that the list of improvements in stress tolerance is by far the longest. This may indicate that yield advances in hybrid maize depends primarily on increase in stress tolerance—more specifically, on increased tolerance to the stresses that typically occur in the environments where the hybrids are grown. It is true that yields of both older and newer hybrids are reduced in the presence of stress, either biotic or abiotic, but it also is true that the newer hybrids
130
DONALD N. DUVICK
always yield more than the older ones in the presence of stress and so, by this definition, are more stress tolerant. Changes that impart efficiency, such as smaller tassels, more upright leaves, faster dry down of grain (to enable longer kernel-fill period), and lower percentage grain protein, may have also enabled higher yields. These changes were not selected directly, but they might have been selected indirectly as a consequence of selection for increased grain yield per unit area because they (presumably) improve the efficiency of transforming sunlight, CO2, and soil nutrients into plant constituents, and so help increase yields. The author is not surprised by the stability of maturity and plant size over the years despite the fact that increased plant size and later maturity correlate positively with increased grain yield. The average date of first frost and farmer prejudice against tall plants have automatically set the limits to which breeders can select for harvest maturity and plant height, at least for U.S. corn belt hybrids. Some of the unintended (or unplanned) stabilities present surprises, or at least do not agree with conventional wisdom. Probably the breeders’ intended constraint on plant and ear height has indirectly held leaf number and leaf area index constant. In theory, the more leaf area per unit land area (up to a maximum LAI of about four or five), the more photosynthesis and consequently the higher the grain production per unit land area. However, additional leaf area per unit land area has been achieved not by increasing leaf area per plant, but by crowding more plants together. One could hypothesize that shorter internodes could allow more leaves and more leaf area per plant without increasing plant and ear height. This could substitute for (or extend) the effect of increased plant density. However, plant architecture (e.g., leaf dimensions, shape, or angle) might need to be changed if such an approach were used. Because leaf number usually is positively correlated with time to flowering, it might not be possible to increase leaf number and still hold the flowering date constant (a requirement for nearly all temperate maize production). Some may express surprise that HI of maize hybrids in the United States has not increased (in absence of stress-induced barrenness) because of frequent statements that an increase in HI was an important reason for the higher yields of green revolution wheat and rice cultivars (e.g., Donald, 1968; Peng et al., 1999; Reynolds et al., 1999; Swaminathan, 1998). However, the first maize hybrids (as well as their parent OPCs) had about the same HI as the initial high yield rice and wheat cultivars. It would appear that farmer selection (at least in U.S. corn belt OPCs) had already brought maize plants to an approximation of the 40–50% HI of initial green revolution rice and wheat cultivars. Interestingly, rice and wheat breeders at the international centers now suggest that an increase in biomass should receive major (perhaps primary) emphasis as they strive to effect further increases in yield for
ADVANCES IN MAIZE
131
their crops (e.g., Peng et al., 1994, 1999; Reynolds et al., 1999). “Clearly, there are limits to how far HI can be further increased in improved varieties [of rice] (Peng et al., 1994).” The lack of increase in yield potential per plant is surprising until one reflects on the fact that up until now, the sole method of increasing yield per unit area has been to increase plant density while maintaining constant ear size (grain weight per plant). Although theoretically it may be possible to raise yields per unit area by increasing yield per plant while holding population constant (at lower densities than present norms), for one reason or other this has not been done except in experimental studies (e.g., Fasoula and Fasoula, 2000; Tokatlidis et al., 2001). Such a goal might be practical, however, for hybrids suited for drought-prone environments, where planting at lower density is prudent but the ability to utilize occasional higher rainfall by increasing yield per plant would be desirable. The relative lack of increase in heterosis, either absolute or relative, also will surprise those of us who have supposed that the primary way to increase hybrid yield is to increase heterosis for grain yield. It would appear that comments by Hallauer (1999, p. 486) about recurrent selection can apply to hybrids as well. He stated, “the additive effects of alleles with partial to complete dominance were of greater importance but dominant and epistatic effects could not be discounted.” It is also intriguing to note that absolute heterosis for grain yield is greatest under conditions of stress, in company with the knowledge that an increase in stress tolerance of all kinds has strongly accompanied gains in hybrid yield over the years. Both inbreds and hybrids are greatly improved in stress tolerance, but hybrid gains are greater that those of their parental inbreds.
B. POTENTIAL HELPS OR HINDRANCES GAINS in YIELD
TO
FUTURE
For the past 70 years, breeders have improved the yielding ability of hybrid maize by selecting new genotypes with adaptation to the ever-increasing stresses of constantly increasing plant density as well as to other kinds of prevalent abiotic and biotic stress. They have done so by exposing these hybrids to an increasing diversity of environments as the capacity for wide area testing (especially in the commercial sector during the past few decades) has been dramatically increased by investments in mechanization and information management. Breeders consistently have selected for higher average yield with acceptable grain moisture, improved standability, resistance to local diseases and insect pests (that often change in kind or intensity over the years), and tolerance to increased plant density. (Farmers continually plant the newest hybrids at a density higher than recommended, thus forcing
132
DONALD N. DUVICK
breeders to continually raise the density at which they select the next round of hybrids.) Newer hybrids have managed the stress of high plant density (which usually accentuates other kinds of stress) in two ways: increased stress tolerance and increased efficiency of grain production. Efficiencies in grain production that presumably are provided by more upright leaves, smaller tassels, and lower grain protein percentage may have gone about as far as they can go; leaves cannot be much more vertical without clasping the stem, tassels on some hybrids have no side branches at all, and it is possible that livestock feeders will not be willing to accept grain with a smaller percentage of protein than is now at hand. Therefore, breeders will need to make even greater progress in improving traits for stress tolerance if they are to continue the linear increase in grain yield that has prevailed during most of the past 70 years of hybrid maize breeding. Increased emphasis on such traits as tolerance to extremes of temperature, to drought, to excess soil moisture, and to deficiency of soil nitrogen (without neglecting simultaneous selection under nonstressed, highyield conditions) will probably be worth the effort. For each of these traits, development and deployment of managed stress levels (such as managed irrigation in a rain-free climate) may increase the precision and reliability of selection for the trait. However, widespread on-farm testing in a full range of “natural” environments, from high yield to high stress, must be the rock on which all other selections are based. Past experience has shown that overemphasis on a single trait may produce correlated changes in other traits, sometimes with unintended (and unwanted) results. There is no substitute for extensive trials, in both small plot and strip test format, of hybrids in final stages before release (and also just after release). Although one may not be able to identify the different kinds of stress (or nonstress) in those trials with precision, one can be sure that such testing will subject the genotypes to a multitude of stresses common to the area of adaptation. Some of the stresses will be severe to catastrophic and some will be so minimal that only the hybrids can sense them. In the end the genotypes with the broadest tolerance to these stresses will give the highest yields in both low-yield and high-yield environments.
C.
PREDICTIONS
Mark Twain is supposed to have said, “Prophecy is a good line of business but it is full of risks.” This caution surely applies to predictions about prospects for future yield gains in hybrid maize, or in other kinds of maize cultivars.
ADVANCES IN MAIZE
133
Increased grain yield per se may be less important in the future than in the past for an increasingly large part of maize production. Specialty products such as maize with altered or higher oil or protein content, or high extractable starch, or maize bred for use as a biofuel may rise in importance in addition to or sometimes in place of commodity feed grain production (e.g., Lambert et al., 1998; Ng et al., 1997; Whitt et al., 2002). Conceivably, some of these novel kinds of maize could command a premium price. Although yield of some kind would be important for these specialty categories of maize, grain yield per se would take second place to yield of a specified product—yield of a specific kind of oil, protein, starch, etc. However, despite these possible new markets for specialty types of maize, the demand for maize as a feed grain will increase robustly if the developing world continues to improve its economy and therefore its appetite for meat, milk, and eggs (Rosegrant et al., 2001; Taha, 2003). Consequently, production of maize as a commodity grain for animal feed will continue to dominate commercial maize production. Increasing the on-farm grain yield of maize hybrids will persist as a primary goal of maize breeders. Farmers will require (and demand) hybrids that dependably produce maximum yield with minimum inputs—a key requirement for the profitable production of commodity maize grain. Although the past does not necessarily predict the future, 70 years of linear genetic yield gain in many parts of the maize-growing world would seem to predict that similar gains will continue for at least the next few decades. This prediction is more likely to hold for regions that have recently adopted hybrids and complementary intensive management practices. Production in those regions will not be as near to the theoretical maximum yield potential (Cassman, 1999; Duvick and Cassman, 1999) as may be true for regions with a longer period of constantly increasing yields. However, even in regions with longer periods of yield increase such as the U.S. corn belt, continuation of the long-standing practice of remolding adapted germplasm, and slowly and carefully importing useful pieces of exotic germplasm, will guarantee increased yield potential and increased stability of yield for years to come. On-farm yields may not always rise in line with genetic improvements, however. In the coming decades, residents of the wealthier countries may force their farmers to reduce (or, in some cases, eliminate) applications of synthetic fertilizers and/or pesticides, with the intention to improve environmental and human health. Such reductions could reduce maize yields. Even though the newer hybrids would yield more than the older ones (including the OPCs) following such a reduction of inputs, their yields could very well be lower than before reductions were put into effect, although the amount of loss (if any) would depend on the extent of the input reductions.
134
DONALD N. DUVICK
If this speculated mandatory reduction of management inputs should come to pass, one can envision a future in which genetic yield gains will continue but on-farm yields will stagnate or decline. In other words, the genetic improvement of critical traits might be needed simply to maintain yields at an even level or to minimize reductions in yield. Without genetic improvements, yields would drop even more. This scenario highlights the importance of breeding for resistance to disease and insect pests, tolerance to deficiency of soil nutrients, and tolerance to other locally important kinds of abiotic stress. One should note, however, that breeders have already made good progress in breeding for such kinds of tolerance/resistance. As shown repeatedly in this review, breeding for increased resistance to abiotic and biotic stress has been the basis for 70 years of yield increase and dependability in hybrid maize. With stimulus from farmers and the marketplace, it seems reasonable to suppose (and predict) that breeders can and would increase the intensity of breeding for stress tolerance, with special emphasis on specific stresses that were amplified by the reduction of specific inputs. For example, data reviewed in earlier sections of this report indicate that successive hybrids have shown steady and significant genetic improvement in the efficient use of soil nitrogen, although the selection for improvement was indirect (and unintentional). If application of nitrogen fertilizers should be curtailed, maize breeders could select directly for efficient use of soil nitrogen, and should be able to make even faster progress than in earlier years. Such progress might enable a continued increase (or at least prevent a decline) in on-farm yields, despite mandated reductions in fertilizer use. Likewise, with regard to pesticides, as noted earlier, a small number of effective aids from biotechnology for genetic improvements in disease and insect resistance are already in place and more are contemplated (Fernandez-Cornejo and McBride, 2002; James, 2003a; Rice et al., 2003; Runge and Ryan, 2003). Continuation and enhancement of these transgenic breeding efforts could incrementally and significantly increase the ability of farmers to maintain yields without the use of pesticides. (An example of enhancement of transgenic protection would be to progress from the construction of vertical resistance genotypes to building systems of horizontal— more durable—resistance.) Importantly, the transgenic improvements will be most useful if they are integrated with continuing achievements in conventional breeding for pest resistance (which itself will be enhanced by new knowledge and tools, contributed by molecular biology). In time, there will be no distinction between “biotechnology protection” and “conventional protection.” For some maize producers, an even more drastic reduction in inputs may occur in future years, as water for irrigation becomes less available in regions where irrigation is important (and sometimes essential) for maize production
ADVANCES IN MAIZE
135
(Rosegrant et al., 2002). As with breeding for tolerance to nutrient imbalance or biotic stresses, breeders in regions of water shortage will need to increase their emphasis on breeding for drought tolerance, either by moving from indirect to direct selection or by increasing the emphasis on already existing direct selection for drought tolerance. Past experience indicates that this trait can be improved without sacrificing the ability to produce high yields with favorable water balance, and evidence now accumulating indicates that these breeding efforts may someday be made even more efficient because of new insights provided by molecular biology investigations. So, as with postulated reductions in soil nutrient and pesticide application, breeders should be able to mitigate yield reductions, or even maintain on-farm yields, in many of the areas where irrigation is reduced or eliminated. A less optimistic prediction, based on past experience, is that the price of successive increases in genetic yielding ability will continue to rise, just as it has risen during the past 70 years. In the United States, for example, today’s crew of maize breeders (broadly defined to include those who work in biotechnology applied to plant breeding) is many times larger than the crew that made advances during the first two or three decades of hybrid maize breeding (Crosby et al., 1985; Duvick, 1984a; Fernandez-Cornejo, 2004; Frey, 1996), yet the genetic yield gain per year is no larger than in past times—the gain is linear. (And therefore the expenditure per unit gain is many times larger now than in early years.) Judging from this past experience, today’s plant breeding crew will need to be enlarged even further if future gains are to be made at the same pace as is now achieved unless much more efficient methods of breeding are developed and implemented. Some have predicted that significant gains in breeding efficiency will occur as various tools of biotechnology are employed, utilizing new genetic discoveries and new knowledge of genes and gene action in the maize plant. However, the current state of the art primarily is in the development of techniques and data (e.g., Emrich et al., 2004; Jansen et al., 2003; Lawrence et al., 2004). Biotechnology is not the primary tool for the development of improved cultivars. Although useful new traits have been added via transgenesis, the number actually in use is still small (but growing) and is limited to such defensive traits as pest resistance and/or herbicide tolerance. Breeders must continue to use routine empirical breeding methods to effect broad-scale yield improvements in the “base germplasm,” which they then can enhance with useful transgenes. Today’s maize breeders cannot reduce the effort devoted to “conventional plant breeding” without also reducing the rate of progress in the development and release of higher-yielding cultivars, cultivars with a full and well-balanced spectrum of the important traits that in sum govern maize yield. Additionally, the present large investment in the testing of transgenic products for safety adds significantly to the cost of their use for genetic
136
DONALD N. DUVICK
improvement. Comments by soybean breeders on the use of biotechnology for soybean breeding are relevant to its use in maize breeding, as follows: “Despite all the opportunities, biotech soybeans face numerous challenges. Because of the cost of technology and regulatory clearance, it is difficult for developers to earn sufficient returns on research investment for many biotech traits. Gaining acceptance of crops and grain derived through biotechnology, particularly in Europe, is yet another challenge. Although biotechnology acceptance is increasing around the world, significant challenges will be faced by those wanting to bring new transgenic traits to market (Soper et al., 2003).” One hesitates to predict how much time must pass before the current investment in biotechnology can bring about significant savings in time and money per unit gain in maize yield, even though it is obvious that biotechnology research and development indeed will give significant and innovative support to maize breeding. New techniques, based on new knowledge in molecular biology, are increasing breeding efficiency incrementally, and with time the number and use of such new techniques can only grow. As stated by Runge and Ryan (2003), “Plant biotech R&D in the pipeline as of 2001 through mid-2003 indicates almost a hundred new traits in testing. Represented in these activities are about 40 universities (mainly land grants) and about 35 private sector companies. Without question, more research and development as measured by field tests has been devoted to biotech traits in corn than to any other crop, attracting scores of public and private institutions. Among the traits in testing for corn were 19 new agronomic properties, four traits for fungal resistance, seven for herbicide tolerance, four for insect resistance, ten trials focusing on some form of marker genes, and over 30 for output and other end-use traits.” This leads to the author’s final prediction, that despite the efforts of some segments of society (e.g., Turning Point Project, 1999a,b) to stop or otherwise hinder the use of biotechnology as an aid in food production (for detailed description of this subject, see Charles, 2001), the tools of molecular biology increasingly will aid maize breeders in their efforts to develop superior hybrids. The most important and long-reaching aid will come not so much from transgenes per se as from the use of a wide range of biotechnology-based tools to give breeders a deeper knowledge of the genetics and physiology of the maize plant; with this knowledge they will be able to fine-tune maize genomes to achieve desired ends with much greater speed and efficiency. Valuable assistance will continue to come directly from some classes of transgenes, e.g., new and more broadly effective versions of Bt transgenes (James, 2003a; Rice et al., 2003). Breeders in the tropics and subtropics will have particular use for transgenes that impart effective resistance to the wide array of disease and insect problems in those regions, especially for instances
ADVANCES IN MAIZE
137
in which adequate genetic resistance cannot be found in maize itself. In all cases, these transgenes will prevent yield loss in the presence of epidemics and infestations of the pest in question. In those places where pest depredation is chronic, yield levels on average will advance. However, as noted earlier, the use of transgenes for pest resistance must move beyond changes that contribute vertical resistance to those that impart more durable kinds of horizontal resistance, not an easy task but one that will become possible as biotechnology-based knowledge and insights accumulate. As biologists move beyond genomics to proteomics, metabolomics, and other related disciplines (some as yet unnamed or undiscovered), they will help maize breeders identify key genes and gene systems/interactions in the maize plant, and then, working together, the molecular biologists and breeders will learn how to regulate or reconstruct them in ways that will intensify the expression of key traits, whether for tolerance to heat and/or drought, to chronic disease problems, or to tolerance of deficiency of soil nutrients such as nitrogen. Breeders will learn how to mine genetically diverse exotic maize populations for improved versions of key genes or (more likely) their regulator systems and, with aid of molecular markers, to move them into elite germplasm with precision and efficiency (see Tuberosa et al., 2002). (Marker-assisted selection is already used extensively in some crops, including maize, to move useful genes or linkage groups from exotic to adapted germplasm and/or elite cultivars.) In some cases, knowledge of the identity and/or function of important genes in other species (such as for tolerance to certain kinds of abiotic stress) will enable biologists to identify their counterparts in maize, enabling breeders and molecular biologists, collaboratively, to fine-tune the actions of the maize genes, sometimes by using key regulatory sequences from the exotic organism (e.g., Appenzeller et al., 2004; Shou et al., 2004). Of course, individual genetic changes will be useful only when they efficiently interact with the complete genetic complex in ways that improve the overall performance of the plant according to goals set by the breeder. Knowledge of the ways in which expression of a key gene affects the action and products of other genes (i.e., the pleiotropic and epistatic effects), and consequent interactions with the environment, will be critical to the success of improving the action of any individual gene. Finally, beyond all these conjectured advances, the maize breeders can always hope for the Holy Grail of plant physiologists, major improvement in the efficiency of the primary steps of converting intercepted solar radiation into stored carbon, effected without disrupting the rest of the infinitely complicated network of interacting genetic systems and ensuing physiological processes that operate the functioning maize plant or any other organism.
138
DONALD N. DUVICK
However, in the end, after all the modern tools have been employed to the maximum degree, maize breeders will still need to walk the fields, observing their latest creations under the widest possible range of conditions that commonly occur in the intended region of adaptation. (One could describe this activity as “personal perusal and evaluation of the genotype environment interaction.”) The breeders will collate this subjective and highly personal information with objective information obtained from widespread performance trials, laboratory analyses, and other factual tests of performance. In brief, maize breeders will need to practice the art as well as the science of breeding if they are to continue the genetic progress that has been achieved by their predecessors during the past three-quarters of a century. “As the joy of artistic creation begins to assert itself we may expect many interesting developments in the newer methods of corn breeding (H. A. Wallace, 1930, unpublished manuscript).”
REFERENCES Appenzeller, L., Doblin, M., Barreiro, R., Wang, H., Niu, X., Kollipara, K., Carrigan, L., Tomes, D., Chapman, M., and Dhugga, K. S. (2004). Cellulose synthesis in maize: Isolation and expression analysis of the cellulose synthase (CesA) gene family. Cellulose 11, 287–289. Baker, R. F. (1990). D. N. Duvick interview with R. F. Baker. Pioneer Hi-Bred International, Inc., Johnston, Iowa. Ba¨ nziger, M., Edmeades, G. O., and Lafitte, H. R. (1999). Selection for drought tolerance increases maize yields across a range of nitrogen levels. Crop Sci. 39, 1035–1040. Barker, T., Campos, H., Cooper, M., Dolan, D., Edmeades, G., Habben, J., Schussler, J., Wright, D., and Zinselmeier, C. (2005). Improving drought tolerance in maize. Plant Breed. Rev. 25, 173–253. Bolan˜ os, J., and Edmeades, G. O. (1992). Eight cycles of selection for drought tolerance in lowland tropical maize. I. Responses in grain yield, biomass, and radiation utilization. Field Crop Res. 31, 233–252. Bruulsema, T. W., Tollenaar, M., and Heckman, J. R. (2000). Boosting crop yields in the next century. Better Crops 84, 9–13. Cardwell, V. B. (1982). Fifty years of Minnesota corn production: Sources of yield increase. Agron. J. 74, 984–990. Carlone, M. R., and Russell, W. A. (1987). Response to plant densities and nitrogen levels for four maize cultivars from different eras of breeding. Crop Sci. 27, 465–470. Cassman, K. G. (1999). Ecological intensification of cereal production systems: Yield potential, soil quality, and precision agriculture. Proc. Natl. Acad. Sci. USA 96, 5952–5959. Castleberry, R. M., Crum, C. W., and Krull, C. F. (1984). Genetic yield improvement of U.S. maize cultivars under varying fertility and climatic environments. Crop Sci. 24, 33–36. Cavalieri, A. J., and Smith, O. S. (1985). Grain filling and field drying of a set of maize hybrids released from 1930 to 1982. Crop Sci. 25, 856–860. Charles, D. (2001). “Lords of the Harvest: Biotech, Big Money, and the Future of Food”. Perseus Press, Cambridge, MA.
ADVANCES IN MAIZE
139
Clements, M. J., Campbell, K. W., Maragos, C. M., Pilcher, C., Headrick, J. M., Pataky, J. K., and White, D. G. (2003). Influence of Cry1Ab protein and hybrid genotype on fumonisin contamination and fusarium ear rot of corn. Crop Sci. 43, 1283–1293. Corrigan, P. (2003). The Flood of ‘93: Comprehensive review of the 1993 flood and how it impacted Iowa. [Online]. Available by National Weather Service, Des Moines, IA http:// www.crh.noaa.gov/dmx/WX_events/93floodsummary.shtml (viewed November 3, 2003). Crabb, R. (1993). “The Hybrid Corn-Makers”, 2nd Ed., West Chicago Publishing Company, West Chicago, IL. Crosbie, T. M. (1982). Changes in physiological traits associated with long-term breeding efforts to improve grain yield of maize. In “Proceedings of 37th Annual Corn & Sorghum Research Conference, 5–9 Dec., 1982” (H. D. Loden and D. Wilkinson, Eds.), Vol. 37, pp. 206–223. American Seed Trade Association, Washington, DC, Chicago, IL. Crosby, E. A., Eddleman, B. R., Kalton, R. R., Ruttan, V. W., and Wilcke, H. L. (1985). “A Survey of U.S. Agricultural Research by Private Industry III”. Agricultural Research Institute. Crow, J. F. (1998). 90 years ago: The beginning of hybrid maize. Genetics 148, 923–928. Cunha Fernandes, J. S., and Franzon, J. F. (1997). Thirty years of genetic progress in maize (Zea mays L.) in a tropical environment. Maydica 42, 21–27. Daberkow, S., Taylor, H., and Huang, W.-Y. (2000). “Agricultural Resources and Environmental Indicators: Nutrient use and Management”. Rep. No. AH722. USDA/ERS, Washingon, DC. Derieux, M., Darrigrand, M., Gallais, A., Barriere, Y., Bloc, D., and Montalant, Y. (1987). Estimation du progre`s gene´ tique re´ alise´ chez le maı¨s grain en France entre 1950 et 1985. Agronomie 7, 1–11. Dodd, J. (2000). How to foresee corn disease outbreaks. In “55th Annual Corn & Sorghum Industry Research Conference, December 6–8, 2000”, Vol. 55, pp. 91–98. American Seed Trade Association, Chicago, IL. Donald, C. M. (1968). The breeding of crop ideotypes. Euphytica 17, 385–403. Duvick, D. N. (1977). Genetic rates of gain in hybrid maize yields during the past 40 years. Maydica 22, 187–196. Duvick, D. N. (1984a). Genetic contributions to yield gains of U.S. hybrid maize, 1930 to 1980. In “Genetic Contributions to Yield Gains of Five Major Crop Plants. CSSA Special Publication No. 7” (W. R. Fehr, Ed.), pp. 15–47. Crop Science Society of America, American Society of Agronomy, Madison, WI. Duvick, D. N. (1984b). Genetic diversity in major farm crops on the farm and in reserve. Econ. Bot. 38, 161–178. Duvick, D. N. (1992). Genetic contributions to advances in yield of U.S. maize. Maydica 37, 69–79. Duvick, D. N. (1997). What is yield? In “Developing Drought- and Low N-Tolerant Maize. Proceedings of a Symposium, March 25–29, 1996, CIMMYT, El Batan, Mexico” (G. O. Edmeades, M. Ba¨ nziger, H. R. Mickelson, and C. B. Pen˜ a–Valdivia, Eds.), pp. 332–335. CIMMYT, Me´ xico, D.F. Duvick, D. N. (1999). Heterosis: Feeding people and protecting natural resources. In “The Genetics and Exploitation of Heterosis in Crops” (J. G. Coors and S. Pandey, Eds.), pp. 19–29. American Society of Agronomy, Inc., Crop Science Society of America, Inc., Soil Science Society of America, Inc., Madison, WI. Duvick, D. N., and Cassman, K. G. (1999). Post-green revolution trends in yield potential of temperate maize in the north-central United States. Crop Sci. 39, 1622–1630. Duvick, D. N., Smith, J. S. C., and Cooper, M. (2004a). Changes in performance, parentage, and genetic diversity of successful corn hybrids, from 1930 to 2000. In “Corn: Origin,
140
DONALD N. DUVICK
History, Technology and Production” (C. W. Smith, J. Betran, and E. Runge, Eds.), pp. 65–97. Wiley, New York. Duvick, D. N., Smith, J. S. C., and Cooper, M. (2004b). Long-term selection in a commercial hybrid maize breeding program. In “Plant Breeding Reviews” (J. Janick, Ed.), Vol. 24, pp. 109–151. Wiley, New York. Dwyer, L. M., Stewart, D. W., and Tollenaar, M. (1992). Analysis of maize leaf photosynthesis under drought stress. Can. J. Plant Sci. 72, 477–481. Dwyer, L. M., and Tollenaar, M. (1989). Genetic improvement in photosynthetic response of hybrid maize cultivars, 1959 to 1988. Can. J. Plant Sci. 69, 81–91. Dwyer, L. M., Tollenaar, M., and Stewart, D. W. (1991). Changes in plant density dependence of leaf photosynthesis of maize (Zea mays L.) hybrids, 1959 to 1988. Can. J. Plant Sci. 71, 1–11. Eberhart, S. A., Debela, S., and Hallauer, A. R. (1973). Reciprocal recurrent selection in BSSS and BSCB1 maize varieties and half sib selection in BSSS. Crop Sci. 13, 451–456. Eberhart, S. A., and Russell, W. A. (1966). Stability parameters for comparing varieties. Crop Sci. 6, 36–40. Edmeades, G. O., Ba¨ nziger, M., and M. Cortes, C. (1997). From stress-tolerant populations to hybrids: The role of source germplasm. In “Developing Drought– and Low N-Tolerant Maize: Proceedings of a Symposium, March 25–29, 1996, CIMMYT, El Bata`n, Mexico” (G. O. Edmeades, M. Ba¨ nziger, H. R. Mickelson, and C. B. Pen˜ a-Valdiva, Eds.), pp. 263–272. CIMMYT, Mexico, D.F. Edmeades, G. O., Schussler, J., Campos, H., Zinselmeier, C., Habben, J., Collinson, S., Cooper, M., Hoffbeck, M., and Smith, O. (2003). Increasing the odds of success in selecting for abiotic stress tolerance in maize. In “Versatile Maize—Golden Opportunities: Proceedings 5th Australian Maize Conference, Toowoomba, 18–20 Feb. 2003” (C. J. Birch and S. R. Wilson, Eds.), pp. 16–28. Maize Association of Australia. Edmeades, G. O., and Tollenaar, M. (1990). Genetic and cultural improvements in maize production. In “Proceedings of the International Congress of Plant Physiology” (S. K. Sinha, P. V. Sane, S. C. Bhargava, and P. K. Agrawal, Eds.), pp. 164–180. Society for Plant Physiology and Biochemistry, New Delhi, India. Edwards, J. W., and Lamkey, K. R. (2002). Quantitative genetics of inbreeding in a synthetic maize population. Crop Sci. 42, 1094–1104. Emrich, S. J., Aluru, S., Fu, Y., Wen, T.-J., Narayanan, M., Guo, L., Ashlock, D. A., and Schnable, P. S. (2004). A strategy for assembling the maize (Zea mays L.) genome. Bioinformatics 20, 140–147. ERS (2003). Data: Adoption of genetically engineered crops in the U.S. [Online]. Available by Economic Research Service/USDA http://www.ers.usda.gov/data/BiotechCrops/ (viewed November 20, 2003). Eyhe´ rabide, G. H., and Damilano, A. L. (2001). Comparison of genetic gain for grain yield of maize between the 1980s and 1990s in Argentina. Maydica 46, 277–281. Eyhe´ rabide, G. H., Damilano, A. L., and Colazo, J. C. (1994). Genetic gain for grain yield of maize in Argentina. Maydica 39, 207–211. FAO (1996). “The State of the World’s Plant Genetic Resources for Food and Agriculture”. Food and Agriculture Organization of the United Nations, Rome, Italy. Fasoula, V. A., and Fasoula, D. A. (2000). Honeycomb breeding: Principles and applications. In “Plant Breeding Reviews” (J. Janick, Ed.), Vol. 18, pp. 177–250. Wiley, New York. Fernandez-Cornejo, J. (2004). “The Seed Industry in U.S. Agriculture: An Exploration of Data and Information on Crop Seed Markets, Regulation, Industry Structure, and Research and Development”. Rep. No. 786. USDA–ERS, Washington, DC. Fernandez-Cornejo, J., and McBride, W. D. (2002). “Adoption of Bioengineered Crops”. Rep. No. 810. Economic Research Service/USDA, Washington, DC.
ADVANCES IN MAIZE
141
Frei, O. M. (2000). Changes in yield physiology of corn as a result of breeding in northern Europe. Maydica 45, 173–183. Frey, K. J. (1996). “National plant breeding study. I. Human and financial resources devoted to plant breeding research and development in the United States in 1994”. Rep. No. 98. Iowa Agriculture and Home Economics Experiment Station, Ames, IA. Fu, H., and Dooner, H. K. (2002). Intraspecific violation of genetic colinearity and its implications in maize. Proc. Natl. Acad. Sci. USA 99, 9573–9578. Gianessi, L. P., and Carpenter, J. E. (1999). “Agricultural Biotechnology: Insect Control Benefits”. National Center for Food and Agricultural Policy, Washington, DC. Gollin, D., and Smale, M. (1998). Valuing genetic diversity: crop plants and agroecosystems. In “Biodiversity in Agroecosystems” (W. Collins and C. O. Qualset, Eds.), pp. 237–265. CRC Press LLC, Boca Raton, FL. Goodman, M. M., and Brown, W. L. (1988). Races of corn. In “Corn and Corn Improvement” (G. F. Sprague and J. W. Dudley, Eds.), pp. 33–79. American Society of Agronomy, Madison, WI. Grobman, A., Salhuana, W., Sevilla, R., and Mangelsdorf, P. (1961). “Races of Maize in Peru: Their Origins, Evolution, and Classification”. National Academy of Sciences – National Research Council, Washington, DC. Hagdorn, S., Lamkey, K. R., Frisch, M., Guimaraes, P. E. O., and Melchinger, A. E. (2003). Molecular genetic diversity among progenitors and derived elite lines of BSSS and BSCB1 maize populations. Crop Sci. 43, 474–482. Hallauer, A. R. (1999). Heterosis: What have we learned? What have we done? Where are we headed? In “Genetics and Exploitation of Heterosis in Crops” (J. G. Coors and S. Pandey, Eds.), pp. 483–492. American Society of Agronomy, Inc., Crop Science Society of America, Inc., Soil Science Society of America, Inc., Madison, WI. Hallauer, A. R., and Miranda, J. B. (1988). “Quantitative Genetics in Maize Breeding”, 2nd Ed., Iowa State University Press, Ames, IA. Hallauer, A. R., Russell, W. A., and Lamkey, K. R. (1988). Corn breeding. In “Corn and Corn Improvement” (G. F. Sprague and J. W. Dudley, Eds.), pp. 469–565. American Society of Agronomy, Inc., Crop Science Society of America, Inc., Soil Science Society of America, Inc., Madison, WI. Hayes, H. K. (1952). Development of the heterosis concept. In “Heterosis” (J. W. Gowen, Ed.), pp. 49–65. Iowa State College Press, Ames, IA. Hetherington, P., Reynolds, T., Marshall, G., and Kirkwood, R. (1999). The absorption, translocation and distribution of the herbicide glyphosate in maize expressing the CP-4 transgene. J. Exp. Bot. 50, 1567–1576. Iowa State University (2003). University Extension Weather Information: Iowa Averages, Climatology Data [Online]. Available by ISU Extension http://mesonet.agron.iastate.edu/ climodat/ (viewed November 6, 2003). James, C. (2002). “Preview: Global Status of Commercialized Transgenic Crops 2002. ISAAA Briefs”. Rep. No. 27. ISAAA, Ithaca, NY. James, C. (2003a). “Global Review of Commercialized Transgenic Crops: 2002 Feature: Bt maize”. The International Service for the Acquisition of Agri-biotech Applications (ISAAA), Ithaca, NY. James, C. (2003b). “Preview: Global Status of Commercialized Transgenic Crops: 2003. ISAAA Briefs”. Rep. No. 30. ISAAA, Ithaca, NY. Jansen, R. C., Jannink, J.-L., and Beavis, W. D. (2003). Mapping quantitative trait loci in plant breeding populations: Use of parental haplotype sharing. Crop Sci. 43, 829–834. Kanampiu, F. K., Kabambe, V., Massawe, C., Jasi, L., Friesen, D., Ransom, J. K., and Gressel, J. (2003). Multi-site, multi-season field tests demonstrate that herbicide seed-coating herbicideresistance maize controls Striga spp. and increases yields in several African countries. Crop Prot. 22, 697–706.
142
DONALD N. DUVICK
King, S. R., and Hagood, E. S. (2003). The effect of johnsongrass (Sorghum halepense) control method on the incidence and severity of virus diseases in glyphosate-tolerant corn (Zea mays). Weed Technol. 17, 503–508. Klenke, J. R., Russell, W. A., and Guthrie, W. D. (1986). Recurrent selection for resistance to European corn borer in a corn synthetic and correlated effects on agronomic traits. Crop Sci. 26, 864–868. Lambert, R. J., Alexander, D. E., and Han, Z. J. (1998). A high oil pollinator enhancement of kernel oil and effects on grain yields of maize hybrids. Agron. J. 90, 211–215. Lamkey, K. R. (1992). Fifty years of recurrent selection in the Iowa Stiff Stalk Synthetic maize population. Maydica 37, 19–28. Lawrence, C. J., Dong, Q., Polacco, M. L., Seigfried, T. E., and Brendel, V. (2004). MaizeGDB, the community database for maize genetics and genomics. Nucleic Acids Res. 32, 393–397. Malin, J. C. (1944). “Winter Wheat in the Golden Belt of Kansas”. University of Kansas Press, Lawrence, KS. McCullough, D. E., Aguilera, A., and Tollenaar, M. (1994a). N uptake, N partitioning, and photosynthetic N-use efficiency of an old and a new maize hybrid. Can. J. Plant Sci. 74, 479–484. McCullough, D. E., Giradin, P., Mihajlovic, M., Aguilera, A., and Tollenaar, M. (1994b). Influence of N supply on development and dry matter accumulation of an old and a new maize hybrid. Can. J. Plant Sci. 74, 471–477. Meghji, M. R., Dudley, J. W., Lambert, R. J., and Sprague, G. F. (1984). Inbreeding depression, inbred and hybrid grain yields, and other traits of maize genotypes representing three eras. Crop Sci. 24, 545–549. Miquel, P. (2001). “La France et ses paysans: Une histoire du monde rural au XXe sie`cle”. L’Archipel, Paris. National Research Council (U.S.) Committee on Genetic Vulnerability of Major Crops (1972). “Genetic Vulnerability of Major Crops”. National Academy of Sciences, Washington, DC. Ng, K.-Y., Pollak, L. M., Duvick, S. A., and White, P. J. (1997). Thermal properties of starch from 62 exotic maize (Zea mays L.) lines grown in two locations. Cereal Chem. 74, 1997. Nissanka, S. P., Dixon, M. A., and Tollenaar, M. (1997). Canopy gas exchange response to moisture stress in old and new maize hybrid. Crop Sci. 37, 172–181. Ostlie, K. R., Hutchison, W. D., and Hellmich, R. L. (2002). “Bt Corn and European Corn Borer: Long-Term Success Through Resistance Management”. Rep. No. BU-07055-GO 1997. Regents of the University of Minnesota, St. Paul, MN. Ostrander, B. O., and Coors, J. G. (1997). Relationship between plant composition and European corn borer resistance in three maize populations. Crop Sci. 37, 1741–1745. Paszkiewicz, S., and Butzen, S. (2001). Crop Insights Vol. 11, No. 6: Corn Hybrid Response to Plant Population [Online]. Available by Pioneer Hi-Bred International, Inc. http://www. pioneer.com/usa/agronomy/corn/1106.htm (viewed January 13, 2004). Paterniani, E., and Goodman, M. M. (1977). “Races of Maize in Brazil and Adjacent Areas”. CIMMYT, Mexico D.F. Peferoen, M. (1992). Engineering of insect-resistant plants with Bacillus thuringiensis crystal protein genes. In “Plant Genetic Manipulation for Crop Protection” (A. M. R. Gatehouse, V. A. Hilderand, and D. Boulter, Eds.), Vol. 7, pp. 135–153. CAB International, Wallingford, UK. Peng, S., Cassman, K. G., Virmani, S. S., Sheehy, J., and Khush, G. S. (1999). Yield potential trends of tropical rice since the release of IR8 and the challenge of increasing rice yield potential. Crop Sci. 39, 1552–1559. Peng, S., Khush, G. S., and Cassman, K. G. (1994). Evolution of the new plant type for increased yield potential. In “Breaking the Yield Barrier: Proceedings of a Workshop on
ADVANCES IN MAIZE
143
Rice Yield Potential in Favorable Environments, IRRI, 29 November–4 December 1993”, pp. 5–20. IRRI, Manila, Phillipines. Reynolds, M. P., Rajaram, S., and Sayre, K. D. (1999). Physiological and genetic changes of irrigated wheat in the post-green revolution period and approaches for meeting projected global demand. Crop Sci. 39, 1611–1621. Rice, M. E. (1997). Perception and performance of Bt corn. In “Proceedings of the 52nd Annual Corn and Sorghum Research Conference”, Vol. 52, pp. 144–156. American Seed Trade Association, Washington, DC. Rice, M. E., Tollefson, J. J., and Oleson, J. (2003). Corn rootworm management with YieldGard rootworm corn: This changes everything. In “Proceedings 15th Annual Integrated Crop Management Conference, Iowa State University, Ames, Iowa”, Vol. 15, pp. 75–84. Iowa State University, University Extension, Ames, IA. Richey, F. D. (1922). The experimental basis for the present status of corn breeding. J. Am. Soc. Agron. 14, 1–17. Rosegrant, M. W., Cai, X., and Cline, S. A. (2002). “Global Water Outlook to 2025: Averting an Impending Crisis”. International Food Policy Research Institute and International Water Management Institute, Washington, DC and Colombo, Sri Lanka. Rosegrant, M. W., Paisner, M. S., Meijer, S., and Witcover, J. (2001). “Global Food Projections to 2020: Emerging Trends and Alternative Futures”. International Food Policy Research Institute, Washington, DC. Rosenow, D. T., and Clark, L. E. (1987). Utilization of exotic germplasm in breeding for yield stability. In “Fifteenth Biennial Grain Sorghum Research and Utilization Conference, 15–17 Feb. 1987” (R. R. Duncan, Ed.), pp. 49–56. National Grain Sorghum Producers Association, Abernathy, TX, Lubbock, TX. Runge, C. F., and Ryan, B. (2003). The Economic Status and Performance of Plant Biotechnology in 2003: Adoption, Research and Development in the United States [Online]. Available by Council for Biotechnology Information (CBI) report at http://www.apec. umn.edu/faculty/frunge/plantbiotech.pdf (viewed February 2, 2004). Russell, W. A. (1974). Comparative performance for maize hybrids representing different eras of maize breeding. In “29th Annual Corn and Sorghum Research Conference”, Vol. 29, pp. 81–101. American Seed Trade Association, Chicago, IL. Russell, W. A. (1984). Agronomic performance of maize cultivars representing different eras of breeding. Maydica 29, 375–390. Russell, W. A. (1985). Evaluations for plant, ear, and grain traits of maize cultivars representing seven eras of breeding. Maydica 30, 85–96. Russell, W. A. (1991). Genetic improvement of maize yields. Adv. Agron. 46, 245–298. Russell, W. A. (1993). Achievements of maize breeders in North America. In “International Crop Science I” (P. S. Baenziger, Ed.), pp. 225–233. Crop Science Society of America, Inc., Madison, WI. Schnell, F. W. (1974). Trends and problems in breeding methods for hybrid corn. In “XVI British Poultry Breeders Roundtable”, Vol. 16, pp. 86–98. Birmingham, England. Shou, H., Bordallo, P., Fan, J.-B., Yeakley, J. M., Bibikova, M., Sheen, J., and Wang, K. (2004). Expression of an active tobacco mitogen-activated protein kinase enhances freezing tolerance in transgenic maize. Proc. Natl. Acad. Sci. USA 101, 3298–3303. Shull, G. H. (1952). Beginnings of the heterosis concept. In “Heterosis” (J. W. Gowen, Ed.), pp. 14–48. Iowa State College Press, Ames, IA. Simmonds, N. W. (1962). Variability in crop plants, its use and conservation. Biol. Rev. 37, 422–465. Simmonds, N. W. (1985). A plant breeder’s perspective of durable resistance. FAO Plant Prot. Bull. 33, 13–17.
144
DONALD N. DUVICK
Smale, M., Reynolds, M. P., Warburton, M., Skovmand, B., Trethowan, R., Singh, R. P., OrtizMonasterio, I., and Crossa, J. (2002). Dimensions of diversity in modern spring bread wheat in developing countries from 1965. Crop Sci. 42, 1766–1779. Soper, J., Judd, D., Schmidt, D., and Sullivan, S. (2003). The future of biotechnology in soybeans. Ag. Bio. Forum 6, 8–10. Sprague, G. F. (1946). Early testing of inbred lines of corn. J. Am. Soc. Agron. 38, 108–117. Sprague, G. F. (1952). Early testing and recurrent selection. In “Heterosis” (J. W. Gowen, Ed.), pp. 400–417. Iowa State College Press, Ames, IA. Sprague, G. F. (1966). Quantitative genetics in plant improvement. In “Plant Breeding” (K. J. Frey, Ed.), pp. 315–355. Iowa State University Press, Ames, IA. Sprague, G. F. (1994). Recorded Interview with George F. Sprague (D. N. Duvick, Ed.). Johnston, IA. Sriwatanapongse, S., Chutkaew, C., Jampathong, S., Paisuwan, B., Aekatasanawan, C., and Jihnayon, S. (1985). Long-term improvement of a maize composite, Suwan I. In “Proceedings of the Fifth International Congress of Society for the Advancement of Breeding Researches in Asia and Oceania (SABRAO), November 25–29, 1985”, pp. 257–261. Kasetsart University, Bangkok, Thailand. Swaminathan, M. S. (1998). Crop production and sustainable food security. In “Crop Productivity and Sustainability: Shaping the Future: Proceedings of the 2nd International Crop Science Congress” (V. L. Chopra, R. B. Singh, and A. Varma, Eds.), pp. 3–18. Oxford & IBH Publishing Company. Pvt. Ltd., New Delhi. Taha, F. A. (2003). The Poultry Sector in Middle-Income Countries and Its Feed Requirements: The Case of Egypt [Online]. Available by Economic Research Service, USDA http://www. ers.usda.gov/publications/WRS03/dec03/wrs0302/ (viewed January 7, 2004). Tatum, L. A. (1971). The southern corn leaf blight epidemic. Science 171, 1113–1116. Tokatlidis, I. S., Koutsika-Sotiriou, M., and Fasoulas, A. C. (2001). The development of density-independent hybrids in maize. Maydica 46, 21–25. Tollenaar, M. (1989). Genetic improvement in grain yield of commercial maize hybrids grown in Ontario from 1959 to 1988. Crop Sci. 29, 1365–1371. Tollenaar, M. (1991). Physiological basis of genetic improvement of maize hybrids in Ontario from 1959 to 1988. Crop Sci. 31, 119–124. Tollenaar, M., Aguilera, A., and Nissanka, S. P. (1997). Grain yield is reduced more by weed interference in an old than in a new maize hybrid. Agron. J. 89, 239–246. Tollenaar, M., McCullough, D. E., and Dwyer, L. M. (1994). Physiological basis of the genetic improvement of corn. In “Genetic Improvement of Field Crops” (G. A. Slafer, Ed.), pp. 183–236. Dekker, New York. Tollenaar, M., and Mihailovic, M. (1991). Bromoxynil tolerance during the seedling phase is associated with genetic grain yield improvement in maize. Can. J. Plant Sci. 71, 1021–1027. Tollenaar, M., Ying, J., and Duvick, D. N. (2000). Genetic gain in corn hybrids from the northern and central corn belt. In “55th Annual Corn and Sorghum Research Conference”, Vol. 55, pp. 53–62. American Seed Trade Association, Chicago. Traore, S. B., Carlson, R. E., Pilcher, C. D., and Rice, M. E. (2000). Bt and non-Bt maize growth and development as affected by temperature and drought stress. Agron. J. 92, 1027–1035. Troyer, A. F. (1996). Breeding widely adapted, popular maize hybrids. Euphytica 92, 163–174. Troyer, A. F. (2000). Temperate corn: Background, behavior, and breeding. In “Specialty Corns” (A. R. Hallauer, Ed.), pp. 393–466. CRC Press, Washington, DC. Tuberosa, R., Salvi, S., Sanguineti, M. C., Landi, P., Maccaferri, M., and Conti, S. (2002). Mapping QTLs regulating morpho-physiological traits and yield: Case studies, shortcomings and perspectives in drought-stressed maize. Ann. Bot. 89, 941–963.
ADVANCES IN MAIZE
145
Turning Point Project (1999a). Biotechnology ¼ Hunger (paid advertisement by Turning Point Project on behalf of 22 non–governmental organizations). In “The New York Times”, p. A5, New York. Turning Point Project (1999b). Who plays God in the 21st century? (paid advertisement by Turning Point Project on behalf of 19 non-governmental organizations). In “The New York Times”, p. A11, New York. USDA (1944–1962). “Agricultural Statistics”. United States Government Printing Office, Washington, DC. USDA (1949–1992). “Agricultural Statistics”. United States Government Printing Office, Washington, DC. USDA (1956). “Agricultural Statistics 1956”. United States Government Printing Office, Washington, DC. USDA-ERS (2003). Fertilizer Use and Price Statistics [Online]. Available by USDA Economics and Statistics System http://usda.mannlib.cornell.edu/ (viewed February 3, 2004). USDA-NASS (2003a). Agricultural Statistics Data Base [Online]. Available by USDA–NASS http://www.nass.usda.gov:81/ipedb/ (viewed February 2, 2004). USDA-NASS (2003b). Corn: Acreage, Utilization, Price, and Value of Production. United States, 1866 to Date [Online]. Available by USDA-National Agricultural Statistical Service http://www.usda.gov/nass/pubs/trackrec/track03a.htm (viewed February 2, 2004). van der Plank, J. E. (1963). “Plant Diseases: Epidemics and Control”. Academic Press, New York. Wallace, H. A. (1930). Practical Aspects of the Newer Methods of Corn Breeding, p. 5 (unpublished manuscript). The Hi-Bred Corn Company, Des Moines, IA. Wallace, H. A., and Brown, W. L. (1988). “Corn and Its Early Fathers”. Revised/Ed. Iowa State University Press, Ames, IA. Whitt, S. R., Wilson, L. M., TenaIllon, M. I., Gaut, B. S., and IV, E. S. B. (2002). Genetic diversity and selection in the maize starch pathway. Proc. Natl. Acad. Sci. USA 99, 12959–12962.
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS Oliver Yu1 and Brian McGonigle2 1
2
Donald Danforth Plant Science Center, St. Louis, Missouri 63132 Crop Genetics, E.I. du Pont de Nemours and Company, Wilmington, Delaware 19880
I. II. III. IV. V.
VI.
VII. VIII. IX. X. XI.
The Health Benefits of Isoflavones in Soybeans Biological Functions of Isoflavonoids in Plants Targets of Isoflavone Engineering The Pathway Leading to Isoflavone Biosynthesis The Entry Point Enzyme: Isoflavone Synthase A. The Discovery of Isoflavone Synthase B. The Mode of Action of Isoflavone Synthase Other Key Enzymes in Isoflavone Biosynthesis A. Chalcone Isomerase B. Chalcone Reductase Transcriptional and Posttranscriptional Regulation of Related Pathways Metabolic Engineering of Isoflavone Accumulation in Legumes Metabolic Engineering of Isoflavone Accumulation in Nonlegumes The Bottleneck of Isoflavone Pathway Engineering: “Metabolic Channeling?” Conclusions References
Isoflavones are phenolic secondary metabolites found mostly in legumes. These compounds play key roles in many plant–microbe interactions and are associated with the health benefits of soy consumption. Because of their biological activities, metabolic engineering of isoflavonoid biosynthesis in legume and nonlegume crops have significant agronomic and nutritional impact by enhancing plant disease resistance and providing dietary isoflavones for the improvement of human health. This review first outlines the current understanding of isoflavone biosynthetic pathways, with focus on key structural enzymes and transcription factors that directly relate to the pathways. Then it summarizes recent progress on metabolic engineering of isoflavone biosynthesis in both legume and nonlegume plants. The major limitations of these approaches, as well as the “metabolic channeling” theory, which is proposed to explain some of the results from the engineering ß 2005, Elsevier Inc. works, are also discussed.
147 Advances in Agronomy, Volume 86 Copyright 2005, Elsevier Inc. All rights reserved. 0065-2113/05 $35.00
148
O. YU AND B. MCGONIGLE
I. THE HEALTH BENEFITS OF ISOFLAVONES IN SOYBEANS Soybean (Glycine max) has been cultivated in China for nearly 4000 years. Soybeans, along with rice, wheat, and two kinds of millet, were regarded as the five sacred crops. For centuries in China, soybean provided the majority of dietary proteins. Along with the spread of Chinese culture, soybean became a staple legume in almost all of the eastern and southeastern Asian countries. Today in these countries soybean is consumed in a variety of traditional forms, such as tofu (bean curd), doufugan (dried bean curd), doujiang (soy milk), tempeh (fermented bean cake), and miso (fermented bean paste). The daily per capita consumption of soybean in east Asia ranges from 12 to 36 g depending on the study (Fukutake et al., 1996; Holt, 1998; Khor, 1997). Even as the standard of living has improved significantly in this part of the world since the mid-1950s and animal-derived protein has become more accessible, soybean continues to be an important food crop. In contrast, in the United States, soybean was not grown on a large scale until World War II, and then only as a result of the search for alternative vegetable oils. Soybean used for human consumption was mostly limited to certain ethnic minorities at that time. However, since the mid-1990s, the food consumption of soybean has increased dramatically from approximately $900 million in the early 1990s to $3,100 million in 2001 (data from the United Soy Board). In North America and western Europe, soy protein is more typically consumed as a highly flavored soymilk, in meat replacements, or as a soy protein isolate, which can be added to a variety of foods. Among the many reasons for the phenomenal growth of the soy foods market, the perceived health benefits of soy and soy-related food products is the most important factor. Epidemiological studies have long revealed striking differences in the occurrence of hormone-dependent cancers between east Asian and Western populations. For example, the incidence of breast cancer is approximately six-fold lower in east Asia than in the United States, and prostate cancer rates are 12-fold lower (Messina, 1999a). These differences must arise from multiple factors, such as genetics and lifestyles. However, many studies suggest that dietary differences, particularly in the consumption of soybean, are one of the major contributors to the prevention of certain cancers in Asian populations. More specifically, one of the unique components of soybean that might play a vital role in health benefits is the isoflavones. Isoflavones are a group of diphenolic secondary metabolites produced by a limited number of higher plants. Although there are at least 22 families of plants that produce and accumulate isoflavones (Dewick, 1993), they occur most frequently in the Papilionoideae subfamily of the Leguminosae.
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
149
Isoflavonoids are compounds derived from the basic 3-phenylchroman backbones of isoflavones by various modifications, such as methylation, hydroxylation, or polymerization. These modifications lead to simple isoflavonoids, such as isoflavanones, isoflavans, and isoflavanols, as well as more complex structures, such as rotenoids, pterocarpans, and coumestans (Dewick, 1993). In soybean, the three major types of isoflavones are daidzein, genistein, and glycitein (Fig. 1). The amount of each isoflavone in soybean seeds varies significantly, with the ratio of daidzein:genistein:glycitein typically being 4:5:1 (Wang and Murphy, 1994b). Most isoflavones in soybean seeds are conjugated with glucose or malonyl-glucose at the C7 position (Fig. 1). Acetyl-glucose
Figure 1 Structure of common isoflavones, estrodiol, and conjugates. (A–H) Genistein, daidzein, glycitein, glyceollin I, coumestan, estrodiol, genistin (glucose conjugate), and malonyl-genistin.
150
O. YU AND B. MCGONIGLE
conjugates have been detected, but mostly in processed or fermented soybean products, which suggests that they may be degradation products of malonylglucose conjugates (Reinli and Block, 1996). Like many conjugated flavonoid compounds, the conjugated isoflavones are stored in vacuoles and are immobile without enzymatically removing the conjugated moieties. Among common edible legumes, soybean contains the highest level of isoflavones, which is more than 100-fold higher than many other legumes (Table I). The physiological function of isoflavones in humans and animals continues to be the subject of intense investigation. As a glimpse of the breadth of public interest to this subject, a keyword search of “isoflavone” in Medline returned 3701 references as of February 2004. Previous reviews have covered the health Table I List of Isoflavone Content in Common Legumesa Legumes Kidney beans Navy beans
Isoflavones (mg/100 g) 0.06 0.21
Pinto beans
0.27
Broad beans Chickpeas
0.03 0.10
Cowpeas
0.03
Lentils
0.01
Lima beans
0.03
Mung beans
0.19
a
Reference Mazur et al. (1998) Franke et al. (1995); Mazur et al. (1998) Franke et al. (1995); Mazur et al. (1998) Mazur et al. (1998) Franke et al. (1995); Mazur et al. (1998) Franke et al. (1995); Mazur et al. (1998) Franke et al. (1995); Mazur et al. (1998) Franke et al. (1995); Mazur et al. (1998)
Legumes
Isoflavones (mg/100 g)
Peanuts
0.26
Peas
2.42
Pigeon peas
0.56
Reference Franke et al. (1995); Mazur et al. (1998) Franke et al. (1995); Mazur et al. (1998) Franke et al. (1995); Mazur et al. (1998)
Soybean, Korea
144.99
Choi et al. (1996)
Soybean, Japan
118.51
Soybean, Taiwan
59.75
Franke et al. (1995); Wang and Murphy (1994a) Franke et al. (1995)
Soybean, U.S. food quality
Franke et al. Soybean, (1995); Mazur U.S. et al. (1998) commodity
128.35
153.40
Franke et al. (1995); Mazur et al. (1998); Wang and Murphy (1994a,b) Eldridge and Kwolek (1983); Franke et al. (1995); Wang and Murphy (1994a)
Data adopted from USDA—Iowa State University Database on the Isoflavone Content of Foods, Release 1.3 – 2002. http://www.nal.usda.gov/fnic/foodcomp/Data/isoflav/isoflav.html).
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
151
benefits of isoflavones extensively and thus are not discussed in detail here (Davis et al., 1999; Messina, 1999b; Messina et al., 1994; Nestel, 2003; Setchell and Cassidy, 1999; Taylor, 2003; Watanabe et al., 2003). In summary, the health benefit claims can be grouped into four main categories. 1. Isoflavones may reduce the occurrences of certain types of cancers. In vitro, animal and epidemiological data have established a relationship between isoflavone intake and breast, prostate, and colon cancer occurrence. Inhibition of tyrosine kinases and DNA topoisomerases by isoflavones may contribute to cancer preventions (Messina et al., 1994). 2. Isoflavones may reduce postmenopausal symptoms. Many animal and human studies have evaluated the health effects of isoflavones on menopause-related symptoms and diseases related to menopause/aging. Data are inconclusive regarding whether the observed health effects in humans are attributable to isoflavones alone or to isoflavones plus other components in whole foods. Although some data seem to support the efficacy of isoflavones in reducing the incidence and severity of hot flashes, many studies have not found any difference between isoflavone recipients and controls. Still, the consensus opinion of the North American Menopause Society concludes that foods or supplements that contain isoflavones have some physiologic effects (Taylor, 2003). Since the discovery of increased cancer risks associated with estrogen-based hormone replacement therapy, the use of isoflavones as an alternative for menopausal women has received renewed public and scientific interest (Albertazzi and Purdie, 2002; Barnes, 2003). 3. Isoflavones may prevent coronary heart disease by reducing low-density lipoprotein (LDL) and increasing high-density lipoprotein (HDL). In 1999, the Food and Drug Administration (FDA) approved a health claim that “diets low in saturated fat and cholesterol that include 25 g of soy protein a day may reduce the risk of heart disease” that can be included on packages containing at least 6.25 g of soy protein per serving. It remains the only health claim associated with soybean that has been approved by the FDA. However, recent reviews suggest that soy protein may have more of a significant hypocholesterolemic effect than isoflavones, especially in humans (Demonty et al., 2003). In fact, one study has shown that proteins from other isoflavone-poor legumes can also reduce cholesterolemia, at least in rats (Sirtori et al., 2004). 4. Isoflavones may have positive effects on other physiological processes such as neurobehavioral activities. With the well-known significance of estrogens in neuron and brain functions, studies have started on the neurobehavioral effects of isoflavone consumption in animals and humans (Lephart et al., 2002).
152
O. YU AND B. MCGONIGLE
Despite wide-ranging efforts, the exact nature of the function and mode of action(s) of isoflavones are still not clear. Most investigators appear convinced that the molecular structure of isoflavone, which mimics the hormone estrogen, drives at least part of their physiological functions (Fig. 1). Therefore, many isoflavonoids are commonly categorized as “phytoestrogens.” Indeed, isoflavones can be a ligand of estrogen receptors (Kuiper et al., 1998). The function of isoflavone, however, appears to be both agonistic and antagonistic of estrogen depending on the tissues (Doerge and Sheehan, 2002). It is necessary to emphasize that not all studies demonstrate that isoflavone intake is correlated with health benefits. In fact, a significant portion of research suggests that even some well-known health claims are hard to prove under defined experimental conditions. Like many other nutrient supplements, current understanding of the health benefits derived from isoflavone consumption is based on the meta-analysis of extensive literature, which summarizes the majority of the research carried out on a specific subject. Most importantly, additional research is required to address the molecular nature of the function of isoflavones in animals and humans. Although the health benefits of isoflavones are generally accepted, they are not without controversy. Some observations and experiments have even demonstrated adverse effects under specific conditions (Doerge and Sheehan, 2002). For example, during the early 1980s, when a group of cheetahs at the Cincinnati Zoo were fed a soy-based diet, all of the females became sterile. Other cheetahs of the same family that remained in South Africa and were fed animal-derived protein reproduced normally (Setchell et al., 1987). Later it was discovered that when zoos in North America switched to soy-based diets for cheetahs during the same period, less than 10% of adult females produced live cubs, compared with 60–80% in other countries, suggesting isoflavones in the soy protein may have had a dramatic effect on some carnivores’ reproductive systems. One particular area in which isoflavones have stirred public concern is soy-based infant formula. Approximately 7% of infants born in the United States are lactose intolerant, thus requiring formula that is not milk based. However, due to various reasons, about 25% of the infant formula sold in the United States is soy based (Mendez et al., 2002). When an infant consumes 8 ounces of soy formula, his or her blood isoflavone level can increase up to 22,000-fold, a much higher increase than in adults because of the infant’s lower body weight (Irvine et al., 1995). As a result, the public has been left to ponder the long-term effects of isoflavone on child development. On June 9, 2000, the ABC news program “20/20” reported a story “The Dark Side of Soy” that emphasized the role of soy-based infant formula in promoting the early onset of puberty. More recently,
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
153
Strom et al. (2001) reported the results of a long-term study in which 248 individuals fed soy formula and 563 individuals fed cow milk formula during infancy and now between the ages of 20 and 34 were examined. They concluded that exposure to soy formula does not appear to lead to different general health or reproductive outcomes than exposure to milk formula. Further long-term studies need to be carried out. In the meantime, it is clear that developing an isoflavone-null soybean, in addition to a high-isoflavone soybean, may have a significant economic outcome for soy farmers and industries. In addition to concerns about isoflavone intake for infants, there are concerns surrounding the safety levels of isoflavones, especially when isoflavones are taken in relatively pure form as a food supplement (Barnes, 2003). However, the levels under discussion are almost certainly beyond the levels that plants can biologically synthesize (Munro et al., 2003), as discussed in this review.
II. BIOLOGICAL FUNCTIONS OF ISOFLAVONOIDS IN PLANTS The function of isoflavones in plants is somewhat better understood than the effects of their consumption in animals. Two of the best-studied functions involve plant–microbe interactions: defense and symbiosis. In defense responses, isoflavones are involved in phytoalexin production. Phytoalexins are a group of chemically diverse, low-molecular mass natural products that possess antimicrobial and/or antiherbivore activities. They are the main chemical compounds plants deploy to combat pathogens and disease (Dixon et al., 1995; Graham, 1995; Hammerschmidt, 1999). Different families of plants often produce different types of phytoalexins. Many leguminous plants synthesize isoflavonoid phytoalexins. Isoflavones themselves have antimicrobial effects when tested in vitro (Graham and Graham, 1996). However, the more potent phytoalexins are other isoflavonoids that are derived from isoflavones, such as coumestans and pterocarpans (Hammerschmidt and Dann, 1999; Heath, 2000). Generally, phytoalexins are not detectable in healthy tissue. They are produced by cells immediately adjacent to infected sites and accumulate in dead and dying cells within this localized region. These compounds are synthesized rapidly after infection due to the de novo activation of secondary metabolic pathways, which divert primary metabolic precursors into the production of phytoalexins. In soybean, phytoalexins known as glyceollins increase drastically within hours of elicitor treatment (Graham, 1991, 1995). Analyses of the temporal and spatial distributions of induced
154
O. YU AND B. MCGONIGLE
glyceollins suggest a complex regulation network that controls the hydrolysis of conjugated daidzein pool, de novo synthesis of isoflavones and glyceollins, and the establishment of competency in neighboring cells (Graham and Graham, 1994, 2000; Guo et al., 1998). As part of the hypersensitive response, the increased production of isoflavonoids is correlated with increased disease resistance (for reviews, see Dixon, 2001; Dixon and Paiva, 1995; Mansfield, 2000). However, little molecular data exist to explain how isoflavonoids inhibit a microbial invasion. Symbiosis is the intimate association of two dissimilar organisms. Legumes and symbiotic soil rhizobia communicate using small diffusible molecules (for reviews, see Dixon et al., 1996; Gualtieri and Bisseling, 2000; Pueppke, 1996). The signal molecules that plants excrete from roots are flavonoids and isoflavonoids, which are chemotaxic to rhizobia and other microbes (Barbour et al., 1991; Dakora, 2000). When rhizobia encounter these compounds, the nodD protein in the bacterial membrane binds them and transcriptionally activates the nod operons (Pueppke, 1996; Smit et al., 1992). The specific binding of nodD with particular flavonoids/ isoflavonoids is the main determinant of host specificity of rhizobia (for a review, see van Rhijn and Vanderleyden, 1995). The proteins encoded by the nod operons synthesize and release a group of lipo-chitooligosaccharides called Nod factors. These compounds can induce a series of physiological changes in plants that eventually lead to nodule morphogenesis and N2 fixation (Downie and Walker, 1999; Pueppke, 1996). Among the biological effects of Nod factors, isoflavonoid biosynthesis is highly induced, presumably creating positive signal feedback cycles between the plant and the microbe (Recourt et al., 1991; Schmidt et al., 1994; van Brussel et al., 1990). This feedback response is essential in establishing a symbiotic relationship because other plants may also secrete flavonoid compounds from the roots, but only legumes distinguish themselves by responding to Nod factors with increased flavonoid/isoflavonoid secretion. The exact mechanisms and pathways that respond to Nod factors are not clear, although a receptor-mediated signal transduction pathway may exist (Bloemberg and Lugtenberg, 2001; Bonfante et al., 2000; Gualtieri and Bisseling, 2000; Stougaard, 2001; van Rhijn and Vanderleyden, 1995). In experiments using Arabidopsis thaliana, two nitrogen-fixing bacteria, Azorhizobium caulinodans and Herbaspirillum seropedicae, colonized the roots by entering through lateral root cracks at the junction of primary and lateral roots. This colonization was significantly enhanced by the application of isoflavone daidzein and genistein (Gough et al., 1997). It is intriguing to note that these bacteria regularly colonize rice and wheat roots as well (Cocking et al., 1995).
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
155
III. TARGETS OF ISOFLAVONE ENGINEERING Due to the traditional limited consumption of soybean in Western cultures (currently about 2 g per capita per day; Reinli and Block, 1996), the idea of increasing isoflavone concentrations to provide health benefits has gained widespread interest in recent years (Dixon and Steele, 1999; Forkmann and Martens, 2001). If a higher isoflavone soybean exists, people may get sufficient healthful compounds without drastically changing their dietary habits. In addition, the isoflavone levels in soybean vary significantly among different varieties. Even for the same variety in the same field, different crop years might bring about more than a threefold difference in isoflavone levels (Wang and Murphy, 1994a). The food industry requires stable and predictable isoflavone contents in soybean-derived products if its health benefits are to be a selling point (Head et al., 1996). However, reducing or eliminating isoflavones in soybean may be valuable for certain sectors of soy food markets, such as infant formulas. Some of the activities attributed to isoflavones, such as inhibition of tyrosine-specific protein kinases, are specific to genistein (Akiyama et al., 1987). The transcriptome of human gut epithelial cell lines challenged with either daidzein or genistein was characterized using microarray technology, and this characterization shows further evidence that the biochemical activities of genistein and daidzein are distinct (Gillies et al., 2003). Independent and, in some instances, opposite responses are found, although there is some degree of overlap in the transcriptomes. Because of this, at times, it may be desirable for some individuals to consume genistein and not daidzein and thus there is a desire to produce functional foods with altered ratios of isoflavones, i.e., the ratio of daidzein to genistein. The most desirable way to do this is to engineer soybeans that do not produce daidzein. Thus, in addition to year-to-year and location-to-location consistency, both the total content and the composition or ratios of isoflavones are targets of metabolic engineering. It is difficult to define specific benchmarks for each of the targeted areas because different populations assign different values to the products. There are two approaches toward these engineering goals: traditional breeding and molecular metabolic engineering. Traditional breeding currently focuses on biochemical and genetic analysis, including quantitative trait locus (QTL) analysis, to determine the factors that contribute to isoflavone production and accumulation, as well as screening for extraordinary isoflavone phenotypes from different varieties and mutants. This review, however, is dedicated to the second approach, which utilizes recombinant DNA technology to modify isoflavone biosynthesis. As discussed
156
O. YU AND B. MCGONIGLE
later, the molecular metabolic engineering approach has been successful in changing isoflavone content and composition. One significant advantage of metabolic engineering over traditional breeding is the prospect of engineering nonlegume food crops to produce isoflavones. As part of the learned cultural differences toward food preferences, many soy foods currently on the market have an unpleasant taste to most mainstream consumers in North America and western Europe. To overcome this major obstacle, engineering isoflavone biosynthesis in nonlegume plants, such as maize (Zea mays) or rice (Oryza sativa), may provide an alternative source of dietary isoflavones that are more acceptable. At the same time, the enhanced nutritional value of the engineered crops will bring additional value to farmers and industries. In addition to modifying plants for enhanced nutritional value, it may also be desirable to modify plants to alter plant–microbe interactions. These efforts will likely be distinct from efforts to enhance nutritional value as they are likely to require the accumulation of isoflavones in novel spatial and temporal patterns, i.e., where and when the plant–microbe interactions occur. Previous research demonstrates that the production of foreign phytoalexins in transgenic plants can dramatically enhance pathogen resistance (Coutos-Thevenot et al., 2001; Hain et al., 1990, 1993; Hipskind and Paiva, 2000; Sparvoli et al., 1994). For example, heterologous expression of stilbene synthase results in accumulation of the phytoalexin resveratrol in both tobacco (Nicotiana tobaccum) and alfalfa (Medicago sativa) that improves disease resistance toward the fungal pathogens Botrytis cinerea and Phoma medicaginis (Hain et al., 1993; Hipskind and Paiva, 2000). These experiments indicate that native pathogens may have difficulty in detoxifying novel phytoalexins. While isoflavones are not typically potent phytoalexins, other simple flavonoids, such as maysin, provide a high level of protection against certain pathogens (Lee et al., 1998) and it remains possible that isoflavones expressed in certain tissues may be of significant value. For instance, the expression of isoflavones in maize silks may provide protection against silk-burrowing earworms. Additionally, engineering isoflavone production in nonlegume plants is a necessary precondition to engineering of more derived isoflavone phytoalexins with greater toxicity. In addition to providing plant protection to disease and herbivory, it is possible that isoflavones may also enhance symbiosis between roots and rhizobial bacteria. Engineering plants to have the ability to fix nitrogen has long been a desire of plant scientists, but given the large number of genes necessary for nodulation, it remains beyond our current technological ability. However, some rhizobia are able to fix nitrogen as free-living organisms. If these bacteria are associated with a plant root, the levels of nitrogen fixation are sufficient to be of value (Cocking et al., 1995). Previous work has shown that specific flavonoids, including daidzein, promote intercellular
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
157
root colonization of Arabidopsis. It is an intriguing possibility that the accumulation of isoflavones in the root of nonlegume plants may lead to increased colonization by rhizobia, leading to nitrogen fixation. While engineering crops to produce isoflavones may not only enhance the nutritional value of the crop but also increase disease resistance, it should be noted that current attempts to engineer the isoflavone levels in soybean have all used seed-specific expression systems to limit the pleotropic effects of altering secondary metabolite biosynthesis.
IV. THE PATHWAY LEADING TO ISOFLAVONE BIOSYNTHESIS Isoflavonoids are synthesized from a branch of the phenylpropanoid pathway (Fig. 2). The phenylpropanoid pathway is ubiquitous throughout the plant kingdom and, in addition to isoflavonoids, produces a variety of phenolic compounds, such as lignans, lignins, flavones, flavonols, condensed tannins (also known as proanthocyanidins), and anthocyanins. Because flower color is such an intriguing attraction to humanity and colored compounds serve as visual markers for analysis, the phenylpropanoid pathway is by far the best-studied secondary metabolic pathway. For years, genetic and biochemical studies, particularly radioisotope-labeled precursor-feeding analyses, have revealed many steps and enzymes involved in flavonoid biosynthesis. Starting from the amino acid phenylalanine, the enzyme phenylalanine ammonia-lyase (PAL) removes the amine group from the amino acid and produces cinnamic acid. The first of several cytochrome P450 monooxygenases in this pathway, cinnamic acid 4-hydroxylase (C4H), adds a hydroxyl group to form p-coumarate. The enzyme 4-coumarate:coenzyme A ligase (4CL) further activates the p-coumarate by attaching a CoA at the three-carbon side chain. Next, chalcone synthase (CHS) carries out the condensation of p-coumaroyl-CoA with three molecules of malonyl-CoA to form the C15 flavonoid skeleton. In most species, this compound is naringenin-chalcone (4,2,4,6´-tetrahydroxychalcone). The chalcone synthesized by CHS can be converted to the flavanone naringenin (5,7,4´-trihydroxyflavanone) by the enzyme chalcone isomerase (CHI). Naringenin is one of the shared substrates between flavonoid and isoflavonoid pathways. Further modifications of naringenin lead to the production of various flavonoid compounds. The most common reaction using naringenin as a substrate is the addition of a hydroxyl group at the C3 position to form dihydrokaempferol as catalyzed by flavanone 3-hydroxylase (F3H), a 2-oxoglutartate-dependent dioxygenase (Deboo
158
O. YU AND B. MCGONIGLE
Figure 2 Outline of the phenylpropanoid pathway (including a list of enzymes and their abbreviations).
et al., 1995). The modification at the C3 position is essential for the production of anthocyanins and condensed tannins, which requires enzymes including dihydroflavonol reductase (DFR), leucoanthocyanidin dioxygenase (LDOX), UDPG-flavonoid glucosyl transferase (UFGT), and others. There are several other enzymes that utilize naringenin as a substrate. In maize, there are five known enzymes that use naringenin. In addition to F3H, maize flavone synthase (FNSII), a cytochrome P450 monooxygenase, uses naringenin to produce flavones. The flavonoid 3´-hydroxylase (F3´H) and flavonoid 3´5´-hydroxylase (F3´5´H) both modify naringenin with additional hydroxylations on the B ring, which are the precursors of the flavone maysin. DFR can also directly use naringenin as a substrate to initiate the tissue-specific production of phlobaphenes (Grotewold et al., 1994).
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
159
Although the activity of all these enzymes has been previously observed, the exact nature of how these enzymes partition the common flavanone substrate is not clear. The flux of the substrate toward each pathway has not been measured. For metabolic engineering of isoflavone synthesis in nonlegume plants, naringenin is a critical metabolite because heterologously expressed IFS must directly compete with other endogenous enzymes for this substrate. Understanding the mechanism and regulation of pathway branching will be essential for metabolic engineering projects. In species that synthesize isoflavones, the enzyme isoflavone synthase (IFS), another cytochrome P450 monooxygenase, acts as the key metabolic entry point for the formation of all isoflavonoids. This enzyme plays two roles: it diverts naringenin formed by CHI into genistein production and, in conjunction with another legume-specific enzyme, chalcone reductase (CHR), forms daidzein. In this case, CHR, CHS, and CHI work in concert to produce isoliquiritigenin and then liquiritigenin, which is the precursor for daidzein. The IFS enzyme is discussed in detail later. Both daidzein and genistein can be conjugated sequentially with glucosyl and malonyl side chains and sequestered in vacuoles. The glycitein biosynthetic pathway is still largely unknown, although it has been suggested that the flavanone liquiritigenin is hydroxylated by flavanone 6-hydroxylase (F6H) to serve as a precursor (Latunde-Dada et al., 2001). It is not known whether methylation precedes or follows isoflavone production. Isoflavones can also be further metabolized to downstream isoflavonoids, such as pterocarpanoids, by a series of legume-specific enzymes. The pterocarpanoid pathway has been studied most extensively in legumes such as Medicago truncatula and alfalfa. In these plants, an isoflavone O-methyltransferase (IOMT) adds a methyl group to the 4´ position (Akashi et al., 2003). This modification occurs prior to isoflavone synthesis (Akashi et al., 2000). Similar pathways for the synthesis of pterocarpans exist in other legumes, including chickpea (Cicer arietinum), pea (Pisum sativum), licorice (Glycyrrhiza echinata), and soybean (Barz and Welle, 1992). In soybean, pterocarpans, known as glyceollins, are synthesized through a well-defined pathway for which many of the genes have been cloned. Daidzein is hydroxylated by isoflavone hydroxylase (I2´H) to form 2´-hydroxydaidzein. I2´H has been cloned from chickpea and alfalfa, and ESTs with high homology exist from other legumes, including soybean (Akashi et al., 1998; Liu et al., 2003). 2´-Hydroxydaidzein is then reduced by isoflavone reductase (IFR) to form 7,2´-dihydrodaidzein. A gene encoding IFR has also been cloned from alfalfa, and ESTs with high homology from other legumes have been found (Paiva et al., 1991). However, it is important to note that homologous enzymes must be characterized biochemically before reliable predictions of gene functions can be made.
160
O. YU AND B. MCGONIGLE
The dihydrofuran ring was originally thought to be formed by a pterocarpan synthase, but more recent work shows that this is in fact catalyzed by two enzymes that have been identified and characterized from alfalfa: vestitone reductase (VTR) and 7,2´-dihydroxy-4´-methoxyisoflavanol dehydratase (DMID) (Guo et al., 1994). This compound, 3,9-dihydroxypterocarpan, is then hydroxylated by pterocarpan 6a-hydroxylase to form 3,6a,9-trihydroxypterocarpan. This reaction is catalyzed by another cytochrome P450, and the gene encoding this protein was cloned from induced soybean (Schopfer et al., 1998). Finally, a prenyltransferase catalyzes the addition of a prenyl group and the resulting compound is cyclized by the enzyme glyceollidin cyclase to form the glyceollin isomers I, II, and III (Welle and Grisebach, 1988). Genes encoding the enzymes that catalyze the last two steps have yet to be characterized. Most of the extensively modified isoflavonoid phytoalexins, such as pterocarpans and coumestrols, are derived from 5-deoxyisoflavones (such as daidzein). Surprisingly, the exact function of genistein (5-hydroxyisoflavone) in soybean disease resistance and other plant–microbe interactions is not very clear, even if it makes up approximately 50% of isoflavone contents in seeds. Although kievetone, a prenylated genistein phytoalexin, exists in a few legumes, including garden bean (Phaseolus vulgaris) and lupines (Lupinus albus), it has not been discovered in soybean (Goossens et al., 1987). It is not clear to what extent the transport of phenylpropanoid compounds through different plants tissues affects their eventual accumulation. Previous reports suggest that the flavonoids that accumulate after UV-light irradiation and the furanocoumarin induced by fungal pathogens are produced by the specific cells exposed to the induction instead of being transported from other tissues (Asthana et al., 1993; Siegrist et al., 1998). However, more recent work suggests that while seeds are the site of much isoflavone synthesis, some of the accumulation of isoflavones is due to transport from other plant tissues, including maternal tissues (Dhaubhadel et al., 2004). These unknowns present an additional layer of complexity to metabolic engineering projects. In many other cases of metabolic engineering, controlling the degradation or catabolism of the targeted products is very important. As an example, free lysine is degraded by lysine-ketoglutarate reductase and saccharopine dehydrogenase, and the development of high-lysine maize was hindered until these enzymes were cloned and their activities understood (Epelbaum et al., 1997). However, several reasons argue against the necessity for this approach in isoflavone engineering: in soybean seeds, the majority of the isoflavonoids produced are isoflavones. The downstream products only accumulate to high levels under stress conditions. It may also significantly reduce the agronomic values if the conversion of isoflavones to other phytoalexins is blocked, thus weakening the disease resistance of the crop.
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
161
However, a greater understanding of the biological fate of isoflavones may necessitate the reevaluation of this argument.
V. THE ENTRY POINT ENZYME: ISOFLAVONE SYNTHASE A. THE DISCOVERY OF ISOFLAVONE SYNTHASE IFS is the first committed enzyme in the isoflavone pathway and converts flavanone substrates to isoflavone products. In 1984, Grisebach’s group at the University of Freiburg in Germany first reported the enzyme activity in elicitor-treated soybean suspension cultures (Hagmann and Grisebach, 1984; Kochs and Grisebach, 1986). In an in vitro microsomal assay using the (radioisotope-labeled (2S)-naringenin substrate, the group was able to demonstrate that intramolecular aryl-migration is catalyzed by a NADPH- and oxygen-dependent enzyme located at the endoplasmic reticulum (ER) membranes. Because specific cytochrome P450 monooxygenase inhibitors such as carbon monoxide and ancymidol could inhibit this enzyme, IFS was thought to be a cytochrome P450 monooxygenase. Like many other cytochrome P450 monooxygenase enzymes, the lipophilic nature and relative low abundance of the protein hindered the isolation and identification of IFS. There was approximately 16 years between the identification of IFS as a cytochrome P450 monooxygenase and the eventual cloning of its DNA sequence. When the gene was finally cloned, it was reported by three independent groups, all of whom took a functional genomics approach to identifying the gene. Dixon’s group at the Noble Foundation in Oklahoma screened two cytochrome P450 monooxygenases, selected from soybean EST libraries, using microsomes purified from insect cells carrying a baculovirus expression vector (Steele et al., 1999). One of the gene products was able to convert liquiritigenin and naringenin to daidzein and genistein, respectively. The gene was named 2-hydroxyisoflavanone synthase (2-HIS). Based on sequence homology and cytochrome P450 monooxygenase nomenclature, it was placed in the CYP93C subfamily of cytochrome P450s. The sequence revealed all the features of a functional cytochrome P450 monooxygenases, including the oxygen binding “I” helix, the heme-binding motifs, and the conserved “PERF” domain. A Northern blot analysis suggested that the homologs of this gene in alfalfa were highly induced upon elicitor treatment in suspension cultures. Ayabe’s group at Nihon University in Japan reported a sequence, also a member of the CYP93C subfamily, that encoded 2-hydroxyisoflavanone synthase activity from licorice (Akashi et al., 1999). Using degenerate
162
O. YU AND B. MCGONIGLE
primers that targeted conserved cytochrome P450 monooxygenase domains, the group isolated eight cytochrome P450 monooxygenase genes from elicitor-induced licorice cultures (Akashi et al., 1999). When expressed in yeast carrying a cytochrome P450 reductase, one of the genes produced a protein that converted flavanone substrates to isoflavones. Radioisotope-labeled thin layer chromatography (TLC) and liquid chromatography-mass spectrometry (LC-MS) confirmed that this enzyme could synthesize 2-hydroxyisoflavanones and eventually isoflavones. An additional report of the discovery of IFS genes came from researchers at the DuPont Company (Jung et al., 2000). Starting from the DuPont EST database, the group focused on disease-induced cytochrome P450 monooxygenase genes. Using a similar yeast expression system as described earlier, they identified two highly homologous soybean EST sequences that encoded proteins that exhibited IFS activity as confirmed by gas chromatography-mass spectrometry (GC-MS). Subsequently, they cloned 11 IFS genes from nine different species using degenerate primers. These sequences included two sequences from Beta vulgaras (sugar beet), which is not a legume plant but had been previously reported to accumulate isoflavones (Geigert et al., 1973). Additionally, the genomic sequence of the two soybean IFS genes was cloned using polymerase chain reaction. One intron was found at the same location of both genes. The in vitro identification of IFS was further confirmed when the soybean IFS was expressed in Arabidopsis and the heterologously expressed IFS was able to utilize the endogenous naringenin accumulated during the Arabidopsis flavonoid biosynthesis and convert it to the isoflavone genistein. Currently, there are 30 IFS sequences in public databases, all of which belong to the CYP93C subfamily of cytochrome P450 monooxygenases (Table II).
B. THE MODE
OF
ACTION
OF ISOFLAVONE
SYNTHASE
IFS is an intriguing enzyme because it catalyzes at least two unusual reactions with one protein: a hydroxylation reaction and an intramolecular aryl migration reaction. Because crystallization of a membrane-bound protein is notoriously difficult, the extensive efforts to resolve the IFS structure have not yet been successful. Without a definitive structure, the exact mechanism of this enzyme reaction is still speculative. It has been proposed that the flavanone is first converted to 2-hydroxyisoflavanone and then to isoflavone by three steps (Hashim et al., 1990). First, a radical at the C3 is generated and then an intramolecular rearrangement moves the aryl group from C2 to C3 and leaves a hydroxyl group still attached to C2. Finally, another enzyme, named isoflavanone dehydratase, then converts 2-hydroxyisoflavanone to isoflavone (Fig. 3).
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
163
Table II List of Cloned IFS Genes in GenBank (http://www.ncbi.nlm.nih.gov/entrez/query.fcgi? db¼Nucleotide) and P450 Database (http://drnelson.utmem.edu/CytochromeP450.html) GenBank number
Species
P450 (CYP)
AF022462 AF195799
Glycine max (soybean) G. max (soybean)
93C1 93C1v1
AF135484, AB023636 AJ243804 AF089850a
G. max (soybean) Glycyrrhiza echinata (licorice) Cicer arietinum (chickpea) Glycine max (soybean)
93C1v2 93C2 93C3 93C4
AF195812 AF195806 AF195807 AF195808 AF195809 AF195801 AF195802 AF195800 AY253284a
G. max (soybean) Vigna radiata (mung bean) V. radiata (mung bean) V. radiata (mung bean) V. radiata (mung bean) Medicago sativa (alfalfa) M. sativa (alfalfa) M. sativa (alfalfa) Trifolium pratense (red clover)
93C5 93C6v1 93C6v3 93C6v3 93C6v4 93C7v1 93C7v2 93C8 93C9
AF195810 AF195811 AF195814 AF195815 AF195817 AF195816 AF195805 AF195804 AF195812 AF195803 AF195813 AB024931
T. pratense (red clover) T. pratense (red clover) T. repens (white clover) T. repens (white clover) Beta vulgaris (sugar beet) B. vulgaris (sugar beet) Lens culinaris (lentil) L. culinaris (lentil) Pisum sativum (pea) Vicia villosa (hairy vetch) Lupinus albus (white lupine) Lotus japonicus P. sativum (pea)
93C9v1 93C9v2 93C10v1 93C10v2 93C11v1 93C11v2 93C12 93C13 93C14 93C15 93C16 93C17 93C18b
AY167424a
Medicago truncatula
AF462633a
Pueraria montana var. lobata
Reference Siminszky et al. (1999) Akashi et al. (1999); Jung et al. (2000) Steele et al. (1999) Akashi et al. (1999) Overkamp et al. (2000) Wu and Verma, direct submission Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Kim et al. direct submission Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Jung et al. (2000) Shimada et al. (2000) Direct submission to P450 Butelli et al., direct submission Jeon and Kim, direct submission
a
Direct submission to GenBank. Direct submission to P450 database.
b
There is much experimental evidence supporting this hypothesis. It has been documented repeatedly that instead of isoflavone, the final product of the IFS enzyme is the 2-hydroxyisoflavanone, a compound not stable at
164
O. YU AND B. MCGONIGLE
Figure 3 IFS reaction scheme outline.
ambient conditions. Both liquiritigenin and naringenin were shown to become 2-hydroxyisoflavanones after incubating with IFS containing microsomes. For this reason, “2-hydroxyisoflavanone synthase” is a more accurate name for this enzyme than “isoflavone synthase.” An enzyme that catalyzes the dehydration of 2-hydroxyisoflavanone to isoflavone has been enriched to apparent homogeneity from Pueraria lobata (Hakamatsuka et al., 1998). The production of isoflavone from 2-hydroxyisoflavanone is most likely assisted by the homologs of this enzyme in other legumes. Based on the X-ray structure of the eukaryotic P450BM3 structure (Ravichandran et al., 1993) and the alignment of several CYP93 family cytochrome P450 monooxygenase proteins, including flavanone 2-hydroxylase (CYP93B1) and FNSII (CYP93B2), two amino acids in the IFS protein were identified that were thought to be involved in the aryl migration reaction (Sawada et al., 2002). When mutated, the resulting proteins produced largely (Ser310 to Thr) or only (Lys375 to Thr) 3-hydroxyflavanone instead of 2-hydroxyisoflavanone. When 3-hydroxyflavanone was fed to microsomes in vitro, IFS failed to convert it to 2-hydroxyisoflavanone. This suggests that 3-hydroxyflavanone is not a substrate for IFS. This further suggests that radical generation and aryl migration are catalyzed by different regions of the protein and that the reaction mechanism suggested by Hashim et al. (1990) is correct. Nonlegume plants accumulate isoflavone instead of 2-hydroxyisoflavanone when only the IFS gene is present even though the suggested reaction mechanism would suggest that 2-hydroxyisoflavanone should accumulate. It remains to be tested whether the autoconversion of 2-hydroxyisoflavanone to isoflavone occurs fast enough to prevent the former compound from accumulating in these plants or whether there is a general “flavonoid dehydratase” enzyme that carries out dehydration of multiple flavonoid/isoflavonoid compounds. Furthermore, there is no experimental evidence to exclude the possibility that IFS assists the conversion of 2-hydroxyisoflavone to isoflavone, perhaps at a much slower rate than the aryl migration reaction. A greater understanding of the IFS mechanism can lead to targeted enzyme engineering with enhanced IFS activities, which may be crucial for isoflavone pathway engineering across species. In the future, generating new
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
165
cytochrome P450 monooxygenases may also produce unique phenolic compounds with special physiological and biological functions.
VI. OTHER KEY ENZYMES IN ISOFLAVONE BIOSYNTHESIS Many of the upstream phenylpropanoid pathway enzymes have been well characterized, including PAL, C4H, 4CL, CHS, and CHI. Two of the most important enzymes for any metabolic engineering attempts are CHI and CHR. CHI is important because it catalyzes a reaction that is a branch point between where the pathway diverts its substrates toward flavonoid and isoflavonoid production. CHR, which is not found in nonlegume species, is responsible for the synthesis of 6´-deoxychalcone, which is the precursor for daidzein and glycitein. Furthermore, the majority of the phytoalexins are derived from 6´-deoxychalcone instead of 6´-hydroxylchalcone, making CHR crucial for metabolic engineering directed toward improved disease resistance. Additional structural enzymes are important for pathway engineering but they are not covered in detail here.
A. CHALCONE ISOMERASE CHI (EC 5.5.1.6) catalyzes the stereospecific isomerization of chalcones into corresponding (2S)-flavanones (Fig. 4). Some chalcones in aqueous solution can be spontaneously isomerized into (2RS)-flavanones with a fairly high turnover rate (Jez et al., 2000b). The in vitro enzyme kinetic assay indicates that CHI operates at the upper limit of the turnover rate for all enzymes, approaching the diffusion limit. CHI ensures that the reaction produces only (2S)-flavanones, which are the biological substrates for downstream enzymes. The only functional CHI gene in Arabidopsis (tt5) is essential for the biosynthesis of anthocyanin and other flavonoid compounds (Shirley et al., 1992).
Figure 4
CHI reaction scheme.
166
O. YU AND B. MCGONIGLE
In all higher plants, isomerization of naringenin-chalcone into naringenin by CHI occurs rapidly. However, the conversion of 2´,4,4´-trihydroxychalcone (isoliquiritigenin) to 7,4´-dihydroxylflavanone (liquiritigenin), a reaction occurring mostly in legumes, has relatively slower kinetics because of the intramolecular hydrogen bond in the substrate molecule (Jez et al., 2002). CHI enzymes isolated from nonlegume plants are unable to use isoliquiritigenin as a substrate. Therefore, CHIs are classified into two types and their distribution is highly family specific. Type I CHIs are found in both legumes and nonlegumes and isomerize only naringenin-chalcone to naringenin. Type II CHIs are found exclusively in leguminous plants and have activities toward both naringenin-chalcone and isoliquiritigenin, yielding naringenin and liquiritigenin, respectively. The genes that encode both types of CHIs have been cloned from various plant species (Kimura et al., 2001; Shimada et al., 2003), and the deduced amino acid sequences within the same type of CHI shared high identity (>70%), while the identity between type I and II CHIs is only about 50% homologous. The antigenic cross-reactivity of the proteins and primary protein sequence suggests that the different substrate specificities of CHIs between leguminous and nonleguminous plants result from the different structures of CHI proteins. Furthermore, both type I and II CHIs are differentially regulated after elicitor treatment (Shimada et al., 2003). X-ray crystallography of an alfalfa type II CHI showed the stereo structure of the protein and revealed the dynamic reaction mechanism of the catalysis (Jez and Noel, 2002; Jez et al., 2000b, 2002): The substrate bound to the active site is forced to form a constrained configuration and is efficiently converted into the product by a general acid–base catalysis mechanism. The amino acid residues possibly affecting the accessibility of naringenin-chalcone and isoliquiritigenin at the active site cleft have been suggested, but the exact structural basis of substrate specificity of CHI is still unclear. The Arabidopsis CHI structure has been resolved and should help determine the structural constrains of the broad and narrow substrate specificity (Noel and Winkel-Shirley, personal communications). Considering potential enzyme interactions, e.g., the interaction of type I CHI with the flavonoid pathway and type II CHI with the isoflavonoid pathway, introducing specific CHIs may be one of the key factors for metabolic pathway engineering (see later).
B. CHALCONE REDUCTASE CHR belongs to the aldo-keto-reductase super family that catalyzes the NAD(P)H-dependent reduction of a variety of carbonyl compounds. All higher plants produce chalcones via the action of CHS, which synthesizes a
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
167
tetraketide by condensing three molecules of malonyl-CoA sequentially to one molecule of the p-coumaroyl-CoA starter. CHR removes the hydroxyl group of the second malonyl-CoA, resulting in 6´-deoxychalcone. Earlier research demonstrated that CHR and CHS are transcriptionally coactivated. In an in vitro assay where purified CHR protein and CHS protein were added in a 2:1 ratio, 6´-deoxychalcone and 6-hydroxychalcone were produced in a 1:1 ratio (Welle and Schroder, 1992). It has been suspected that CHS and CHR form an enzyme complex that carries out the two reactions in tandem. However, colocalization and yeast two-hybrid assays so far have been negative, suggesting that the enzymatic association of the two proteins in vivo may be more complicated and may require additional protein factors. In addition, CHR and CHS proteins exist as multigene families in M. truncatula and soybean (Dixon et al., 2002). There are at least eight sequences from the M. truncatula EST database (TC78137, TC78138, TC82770, TC85519, TC85520, TC85521, BG586880, and AW774745; http://www.medicago.org/MtDB2/Queries/SimilarityDB2.html) that are homologous to CHR. These genes were determined to be homologs by querying the gene indices and selecting sequences with at least 50% similarity, on a DNA basis, to known genes. There are three sequences from the TIGR soybean gene index homologous to CHR (TC173540, TC180190, and TC192014; http://www.tigr.org/tigr-scripts/tgi/T_index.cgi? species¼soybean). There are also at least 13 CHS homologs in the M. truncatula database (TC76767, TC76768, TC76765, TC76884, TC79323, TC79835, TC83930, TC85138, TC85145, TC85146, TC85150, TC85169, and TC85174) and 10 CHS homologs in the soybean database (TC174579, TC175254, TC178994, TC179641, TC183485, TC189879, TC179880, TC190528, TC192493, and BI893708.). The large number of protein sequences would allow for a multitude of different combinations, making the prediction of specific interactions between CHS and CHR extremely difficult. The three-dimensional structure of Arabidopsis chalcone synthase is available (Jez et al., 2000a), but the structure of CHR has only recently been resolved (Noel, personal communications). This structural information will be useful in generating models for the putative CHS–CHR interaction. CHR is necessary for daidzein biosynthesis and, in soybean, for glycitein biosynthesis as well. Thus, engineering efforts targeted at CHR will be essential to control the ratio of isoflavone composition. In soybean, CHR expression needs to be silenced to produce low daidzein lines, and it may be necessary to overexpress CHR to produce larger amounts of daidzein either for nutritional purposes or as a precursor for phytoalexins. CHR is only found in legumes and thus to synthesize daidzein in nonlegume plants, the transformation of CHR genes is required. If the specific interaction between CHS and CHR is essential for producing 6´-deoxychalcone, additional
168
O. YU AND B. MCGONIGLE
“legume-specific” CHS may also need to be introduced for a high-level accumulation of daidzein in nonlegume plants.
VII. TRANSCRIPTIONAL AND POSTTRANSCRIPTIONAL REGULATION OF RELATED PATHWAYS The increased accumulation of flavonoids and isoflavonoids observed after fungal induction is mostly the result of transcriptional activation of biosynthetic genes (Weisshaar and Jenkins, 1998). Transcriptional regulation is therefore important for soybean and nonlegume plant metabolic engineering. As a general strategy, transcriptionally activating the upstream pathway may increase the flow of intermediates and provide more substrates. The transcriptional regulation of the phenylpropanoid and flavonoid pathways are among the most thoroughly analyzed regulatory pathways in plant systems (Fahrendorf et al., 1995; Ni et al., 1996; Weisshaar and Jenkins, 1998). In addition to tissue-specific expression, the phenylpropanoid pathway can be induced by various environmental factors, including both abiotic stress (such as high-intensity light, UV light, and nutrient deficiency) and biotic stress (such as pathogen attack and wounding). Thus various signal transduction pathways can eventually lead to phenylpropanoid pathway activation. The maize C1 gene regulates the tissue-specific biosynthesis of anthocyanin in the aleurone layers of the kernel by binding to a consensus cis element of the promoters of many phenylpropanoid pathway genes (Cone et al., 1986). Together with another transcription factor, R, C1 recognizes a conserved “CAACCACC” element and activates the transcription of the entire pathway (Grotewold et al., 2000; Sainz et al., 1997). A chimeric protein, CRC, consisting of the C1 DNA-binding domain, the complete R gene, and the C1 activation domain, is sufficient to drive anthocyanin production in maize and other plant species (Bruce et al., 2000). C1 belongs to the R2R3-Myb-like transcription family. In Arabidopsis, at least 145 R2R3-Mybs constitute the second largest transcription factor family (Kranz et al., 1998). Twenty-two subfamilies of Mybs have been identified. In animal systems, Mybs regulate essential cell functions such as cell cycle and cell differentiation; however, in plants the R2R3-Myb transcription factors have evolved to regulate diverse plant-specific processes, such as trichome development, ABA and GA hormone responses, and biotic/abiotic stress responses (Braun and Grotewold, 1999; Kranz et al., 1998; Meissner et al., 1999; Rabinowicz et al., 1999). Although most of these Myb-like transcription factors have not been functionally analyzed, the C1
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
169
homologs form a distinct subfamily that specializes in regulating (activating or repressing) anthocyanin production in various species (Rabinowicz and Grotewold, 2000). For example, the C1 homolog in Arabidopsis, the PAP1 gene, activates anthocyanin synthesis in many tissues when it is overexpressed under a constitute promoter (Jin and Martin, 1999; Martin and Paz-Ares, 1997; Moyano et al., 1996). Some Myb-like transcription factors function as suppressors of the pathway. For example, Antirrhinum AmMyb308 and AmMyb330 not only inhibit the transcriptional activation of the phenylpropanoid pathway in Antirrhinum, but also inhibit the pathway activation in tobacco when introduced as transgenes (Tamagnone et al., 1998). Additionally, a group of defense-induced Myb-like transcription factors have been reported. They belong to a more diverse subfamily than the highly conserved C1 subfamily and play important roles in mediating the plant defense response (Lee et al., 2001; Sugimoto et al., 2000; Vailleau et al., 2002; Yang and Klessig, 1996). Because isoflavones are induced by plant defense mechanism, the prospect that this group of Myb-like genes may activate the isoflavonoid pathway in response to defense signals is intriguing. The promoters of many phenylpropanoid and flavonoid pathway genes have been cloned from various species. The common elicitor responsive cis elements, in addition to Myb-binding regions, have been identified (Hartmann et al., 1998; Lesnick and Chandler, 1998; Leyva et al., 1992; Loake et al., 1992; Terauchi and Kahl, 2000). For example, the H-box (CCTACC) and G-box (CACGTG) originally discovered on the bean CHS15 promoter are found in promoters of PAL and other defense-induced genes as well (Arias et al., 1993; Hatton et al., 1995). In addition to Myb-like proteins, other transcription factors also regulate this pathway in response to diverse signal transduction pathways. A basic helix–loop–helix transcription factor (tt8) was shown to activate the DFR and anthocyanidin reductase (BAN) gene in Arabidopsis (Nesi et al., 2000). The LIM protein family transcription factor Ntlim1 controls lignin biosynthesis in tobacco (Kaothien et al., 2002). A Ku-like transcription factor was shown to regulate the CHS genes in bean (Lindsay et al., 2002), and a bZIP family factor G/HBF1 activates CHS in soybean through phosphorylation of the transcription factor (Droge-Laser et al., 1997). Several breakthroughs have been made in the study of signal transduction pathways that lead to the activation of aforementioned transcription factors. Regulatory proteins containing WD40 repeats that are related to the b subunit of heterotrimeric G protein clearly play major roles in activating flavonoid-related Myb-like transcription factors. These proteins include AN11 from petunia (Sompornpailin et al., 2002) and TTG1 from Arabidopsis (Walker et al., 1999). AN11 is a cytoplasmic protein that regulates the Myblike transcription factor ANA, which in turn regulates flavonoid synthesis
170
O. YU AND B. MCGONIGLE
(Quattrocchio et al., 1999). TTG1 regulates flavonoid biosynthesis, trichome development, and root epidermal cell patterning through the Myb-like transcription factor glabrous1 (Walker et al., 1999). Another group of novel regulators of the flavonoid pathway is the WRKY family of transcription factors. The recently cloned Arabidopsis ttg2 gene, which contains two zincfinger domains typical of the WRKY family, shows a similar phenotype of ttg1 and regulates Myb-like transcription factors through transcriptional activation (Johnson et al., 2002). Therefore, it is possible that developmental regulation of the phenylpropanoid pathway is regulated via a WD40-like G-protein signal transduction pathway, whereas defense-induced activation is mediated by WRKY family transcription factors (Winkel-Shirley, 2001). In contrast to our understanding of the transcriptional regulation of anthocyanin biosynthesis, the transcriptional regulation of isoflavonoid biosynthesis is poorly understood. The fungal-induced transcriptional activation of key enzymes in isoflavonoid synthesis, such as IFS, IOMT, and IFR, has been reported, mainly in alfalfa (He et al., 1998; Ni et al., 1996; Paiva et al., 1991; Shimada et al., 2000). Constructs that contain an alfalfa IFR promoter fused to an expression reporter were transformed into both alfalfa and tobacco. Fungal-induced expression in both species and developmental expression in alfalfa appeared to be determined by sequences downstream of −436 bp, whereas sequences up to −765 bp conferred a complex pattern of ectopic expression in a heterologous system (Oommen et al., 1994). An IOMT promoter from M. truncatula was cloned and has also exhibited extensive ectopic expression in heterologous systems (Dixon, personal communication). The promoters of soybean IFS1 and IFS2 genes have been cloned and characterized (Subramanian et al., 2004). The promoters are root and seed specific and respond differently to defense and nodulation signals. A unique xylem-specific expression was induced upon Bradyrhizobium innoculation, suggesting novel roles of isoflavones during legume–rhizobium interactions (Subramanian et al., 2004). The transcription factors specific to the isoflavonoid pathway have yet to be reported. Posttranscriptional regulation involves enzyme activation and inactivation and appears to be present at various steps of the pathway, although its general importance is not well understood. One notable exception concerns a hydroxylase that converts coumaroyl CoA to caffeoyl CoA in parsley cells. This enzyme has a very narrow pH optimum and is presumed to be inactive at the normal cellular pH; exposure of cells to a fungal elicitor results in a rapid decrease in intracellular pH, leading to increased enzyme activity and to the production of caffeoyl and feruloyl esters (Kneusel et al., 1989). The entry point enzyme PAL can be regulated posttranscriptionally as well (Bolwell, 1992). The product of PAL, cinnamic acid, not only inhibits PAL transcription, but also induces a proteinaceous inactivation of PAL enzyme (Barz and Mackenbrock, 1994), probably through increased
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
171
phosphorylation and turnover of the enzyme (Allwood et al., 1999). Few other enzymes have been investigated at the posttranscriptional level.
VIII.
METABOLIC ENGINEERING OF ISOFLAVONE ACCUMULATION IN LEGUMES
It has long been desired to enhance the nutritional value of the soybean by increasing the total isoflavone content in seeds while maintaining the quality and quantity of other output traits such as protein and oil. The most straightforward approach to metabolic engineering, the overexpression of a single rate-limiting enzyme, may result in a higher accumulation of the final product. For example, overexpression of stricotosidine synthase in Catharanthus roseus cell cultures leads to higher levels of alkaloids production (Hallard et al., 1997). There are not sufficient measurements of flux through the phenylpropanoid or isoflavone pathway to determine which enzyme catalyzes the rate-limiting reaction. PAL, CHS, and IFS are the entry-point enzymes of major branches of the phenylpropanoid pathway. However, overexpression of these three enzymes independently failed to alter the isoflavone content significantly (Zernova et al., 2002). PAL, CHS, and IFS were transformed into soybean via somatic embryo culture bombardment. Early results suggested that out of the three genes, only the lectin promoter-driven PAL gene showed up to a 30% increase in isoflavone contents in some of the transgenic seeds. Even that was not significantly different than natural variations of control lines (Zernova et al., 2002). These results agree with similar experiments carried out at DuPont (unpublished results). When PAL, CHS, and IFS were expressed in seed under the control of a storage protein b-conglycinin promoter, no significant alterations of the isoflavone level could be found, even when increased protein levels of these genes were detected by Western blots (unpublished results). There are several possibilities why this approach did not work. One is that even though the flux of substrates was increased by the overexpression of PAL and CHS, the substrates may not be targeted toward isoflavone biosynthesis. A second set of experiments further demonstrates that enhancing the flow of substrates through the metabolic pathway can cause a surprising outcome due to the complexity of metabolic networks. As mentioned earlier, the maize CRC chimeric gene encodes a chimeric transcription factor that transcriptionally activates the entire phenylpropanoid pathway in maize (Grotewold et al., 1998). When the CRC gene was transformed into soybean under a seed-specific phaseolin promoter, it activated the soybean phenylpropanoid and flavonoid pathway (Yu et al., 2003). In the phenylpropanoid pathway, RNA or protein levels of the four enzymes tested, PAL, C4H,
172
O. YU AND B. MCGONIGLE
CHS, and CHI, were increased by more than 10-fold. In the flavonoid pathway, expressions of the three tested genes, F3H, DFR, and FLS, were drastically increased as well (Yu et al., 2003). In contrast, transformation with CRC did not affect the expression of isoflavonoid branch enzymes, including IFS, IFR, and IOMT (Fig. 5). The heterologously expressed CRC is apparently binding to the consensus cis elements at the promoters of soybean phenylpropanoid and flavonoid pathway genes, interacting with the soybean transcription apparatus, which eventually leads to enhanced gene expression. However, the genes specific to the isoflavone pathway in soybean, and perhaps in other legumes, apparently do not share this particular mode of transcriptional activation. One exception is the CHR gene, which showed similar induction patterns as other phenylpropanoid pathway genes, even though CHR is legume specific and not present in maize. It is possible that the required coordinate expression of
Figure 5 Summary of phenotypes from CRC-transformed soybean seeds in a pathway background. CRC activated the phenylpropanoid and flavonoid pathway genes (broad arrows) and resulted in increased daidzein, isoliquiritigenin, and liquiritigenin levels (thin arrows). No changes in isoflavonoid pathway gene expression (dots) occurred, but genistein levels were reduced.
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
173
CHS and CHR forced the genes to adopt similar transcriptional regulation patterns during evolution. Overexpression of CRC not only perturbed the gene expression of related pathways, but also caused a dramatic change in the phenotype of the seed (Yu et al., 2003). First, the daidzein and glycitein content, as measured by HPLC, were increased by up to fourfold in transgenic seeds. Second, the genistein content was significantly decreased, in some cases, to almost undetectable levels. As a result, the daidzein composition as a percentage of the total isoflavone increased from an average of 40% to about 90%, whereas the genistein composition decreased from approximately 50% to less than 10% or lower. Third, all the transgenic seeds carrying a CRC gene showed a brownish coloration on the seed coat. While this brownish coloration was not condensed tannin (i.e., it was not stained with vanillin), condensed tannins were increased in the seed coats as compared to wild-type seed. The gene expression pattern might explain some of the observed phenotypes (Fig. 5). Because CRC increased the flux of the phenylpropanoid and flavonoid pathways but not the isoflavonoid pathway, the increased pool of naringenin substrate was driven toward flavonoid biosynthesis, such as the condensed tannin synthesis. At the same time, the IFS transcript or protein was not increased; therefore genistein accumulation was significantly reduced. Similarly, because CHR and CHS were overexpressed while the downstream enzymes IFR and IOMT were not, the enhanced daidzein and glycitein accumulations were to be expected. However, the location of the brown color cannot be explained because the seed coat is a maternal tissue and its color should not segregate with the transgene, unless an active transportation system exists that transports some flavonoids from the cotyledon to the seed coat during embryo development. Another surprising and contradictory outcome is the discovery of a pool of daidzein precursors, isoliquiritigenin and liquiritigenin, in immature seeds containing CRC (Yu et al., 2003). These compounds have never been reported in wild-type seed. This implies that CHI, which is a diffusionlimited enzyme, is a rate-limiting factor in daidzein formation. Alternatively, CRC might only induce the expression of a type I CHI in soybean (which does have homologs in maize), but fails to activate the type II CHI that is necessary for daidzein biosynthesis, and does not have a counterpart in maize. Under these circumstances, the entire isoflavonoid pathway, including at least type II CHI, IFS, IOMT, and IFR genes, seems to be regulated by a different set of transcription factors than the CRC type. This set of genes may respond to defense or nodulation signals that specifically call for elevated isoflavonoid biosynthesis. The interaction and competition between flavonoid and isoflavonoid pathways are areas ripe for further exploration. The CRC transcription factor could be used to generate high daidzein/low genistein soybeans suitable to make soy foods that would be useful in
174
O. YU AND B. MCGONIGLE
studying the role daidzein contributes to the health benefits observed from the consumption of soy foods. However, to increase the total isoflavone content, the genistein concentration needs to be increased in CRC lines. Because F3H directly competes with IFS for the naringenin substrate, silencing of the F3H gene in CRC lines will block a major portion of the flavonoid pathway and redirect the flow of substrate toward isoflavone biosynthesis. Indeed, when the CRC and F3H cosuppression constructs were cotransformed into soybean, the genistein was restored to the wildtype level, and the total isoflavone level was increased by at least sixfold (Yu et al., 2003). The physiological features (such as defense and nodulation phenotypes) and molecular pathway analyses (such as gene expression and downstream metabolite quantification) of these transgenic soybeans have not been reported. It will be interesting to see if the increasing isoflavone levels in soybean seed lead to differences in plant–microbe interactions. In conclusion, the combination of transcription factor-driven gene activation and suppression of a competing pathway provided a successful method to enhance the accumulation of isoflavones in soybean seed. Compared to projects aimed at increasing isoflavone levels, strategies to develop soybeans with decreased levels of isoflavones are relatively straightforward. There are at least two approaches that should be suitable for producing isoflavone-null soybeans: (1) using a gene-silencing approach to decrease IFS expression in seeds and (2) the combination of overexpression of the CRC transcription factor and gene silencing of CHR to simultaneously reduce both genistein and daidzein production. This may be a case in which traditional breeding is particularly useful, as an alternative strategy it is an extensive mutant isolation screen to identify isoflavone-depleted lines. In addition to being of use for potential applications such as infant formula or protein isolate for bodybuilders, these lines will also be of use in studying the function of isoflavones during plant–microbe interactions and the role that isoflavones play in providing beneficial health benefits derived from the consumption of soy foods.
IX. METABOLIC ENGINEERING OF ISOFLAVONE ACCUMULATION IN NONLEGUMES The introduction of isoflavones into widely consumed crops such as corn may offer new sources of dietary isoflavone, which will increase the nutritional value of the crop and bring the health benefits of isoflavones to more consumers. Aside from the nutritional benefits, the accumulation of isoflavones as a novel phytoalexin in nonlegume plants could potentially
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
175
enhance disease resistance and thus increase the agronomic value of the crop as well. To engineer the isoflavone pathway in nonlegume plants, heterologous expression of IFS is an essential step because it is required to convert the naringenin substrate (ubiquitous in higher plants) to isoflavone. Several experiments demonstrated that complex enzymatic interactions and pathway alterations occurred in nonlegume plants after expression of IFS. The soybean IFS1 cDNA was expressed under the control of a strong constitutive promoter in Arabidopsis, a nonlegume plant that does not synthesize isoflavonoids (Jung et al., 2000). Transgenic plants analyzed by HPLC showed that IFS diverted naringenin from the phenylpropanoid pathway to produce the isoflavone genistein. However, the level of genistein accumulated in Arabidopsis was low as compared to other flavonoids, even when the expression of IFS was high as confirmed by Northern and Western blot analyses. To increase naringenin levels, the expression of phenylpropanoid pathway genes was activated by a 12-h UV-light irradiation. This resulted in a threefold increase in genistein levels, but the portion of genistein in total flavonoid levels actually decreased, suggesting that the pool of naringenin substrate was not equally accessible to the flavonoid and isoflavonoid branch of the pathways. Similar experiments using transgenic tobacco also indicate differential partitioning of naringenin. When the aforementioned IFS construct was transformed into tobacco, the only tissue in which isoflavone accumulation could be detected was the flower, where the phenylpropanoid pathway was actively producing pink anthocyanin pigments (Yu et al., 2000). This was despite the fact that gene expression analysis and in vitro microsomal enzyme assays demonstrated a high level of functional expression of IFS gene in the leaf. The UV-light induction of isoflavone accumulation in Arabidopsis and the tissue-specific accumulation of isoflavone in tobacco indicate that isoflavone biosynthesis in nonlegume plants is dependent on phenylpropanoid path way activity. Additionally, unlike in Arabidopsis, UV-light treatment of tobacco leaves to increase naringenin levels resulted in elevated flavonol levels but failed to raise anthocyanin or isoflavone levels, suggesting that in this tissue, flux through the phenylpropanoid pathway was tightly channeled to flavonol production (unpublished results). Taken together, isoflavone biosynthesis is governed by both pathway activities and enzymatic interactions in heterologous systems. To delineate enzyme interactions between IFS and endogenous flavonoid pathway genes, a CHI from alfalfa, shown to be a type II by its ability to convert isoliquiritigenin into liquiritigenin, was transformed into Arabidopsis and then genetically crossed into a transgenic Arabidopsis carrying the IFS gene (Liu et al., 2002). The combination of genes enhanced genistein
176
O. YU AND B. MCGONIGLE
accumulation moderately. However, the flavonol accumulation was significantly reduced compared to transgenic plants carrying only the legumespecific CHI. The disproportionate reduction of flavonol biosynthesis caused by the presence of IFS further confirmed that the flux of substrate is preferentially channeled toward endogenous flavonoid biosynthesis. Using this information, the authors then went on to carry out experiments that led to significantly higher levels of isoflavone accumulation. Arabidopsis tt6/tt3 double mutant has structural defects in both F3H and DFR genes and is thus blocked in flavonol and anthocyanin production. When the IFS and type II CHI were introduced into a tt6/tt3 double mutant background, genistein accumulation was enhanced by up to 30-fold as compared to plants expressing IFS alone. Again, this suggests that the bottleneck for isoflavone production in Arabidopsis is competition for flavanone between IFS and endogenous flavonol synthesis. The levels of genistein measured approximately 50 mg per gram, the highest levels of isoflavone detected in an engineered nonlegume plant. It should be noted that is still 40- to 60-fold lower than the level of total isoflavones found in commodity soybean seeds grown under typical field conditions. For metabolic engineering of isoflavone production in a monocotyledonous plant, the IFS gene was cloned into monocot expression vectors and was transformed into Black Mexican Sweet (BMS) maize suspension cultures (Yu et al., 2000). Initially, no genistein could be detected in 32 independently transformed lines. To activate the phenylpropanoid pathway, the CRC gene was cotransformed with IFS into BMS. Activation of the phenylpropanoid pathway by the CRC transcription factor resulted in an intense red anthocyanin accumulation. In the IFS and CRC cotransformed lines, approximately half of the transgenic lines were red and the rest were colorless, probably caused by the lack of CRC expression, which was driven by a weak NOS promoter. Genistein accumulated to a detectable level only in the red lines where the phenylpropanoid pathway activity was visible by the pigments. Once again, isoflavone accumulation is correlated with phenylpropanoid pathway activity (Yu et al., 2003). CRC and IFS under the embryo-specific oleosin promoter were also introduced into corn callus cultures capable of regeneration into fertile plants. The regenerated plants had dark red kernels with genistein levels less than 0.1% of that found in soybean seeds. As observed in Arabidopsis and tobacco, the flavonoid branch of the pathway overwhelmed the engineered isoflavonoid branch in maize. One strategy for increasing genistein accumulation is to block anthocyanin production in isoflavone-producing maize. The a1 recessive mutant contains a transposon insertion at the DFR gene and blocks anthocyanin and 3-deoxyflavonoid accumulations throughout the plant. By genetic crossing, the a1 mutant was introduced to the CRC þ IFS maize, and F2 seeds containing CRC þ IFS and homozygous
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
177
a1 alleles were tested for isoflavone accumulations (Yu, unpublished data). As mentioned earlier, there are at least five enzymes using naringenin as a substrate in maize. Blocking DFR may not have a significant impact on overall flux of the substrate. To engineer high levels of isoflavone in maize, further investigations into the enzyme organization and pathway interactions may need to be carried out. To produce daidzein (5-deoxyisoflavone) in nonlegume plants, CHR must be introduced in addition to IFS. Experiments in BMS that introduced IFS, CRC, CHR, and a transformation selection marker showed only a minute amount of daidzein, which was detected by GC-MS (Yu et al., 2000). The detailed gene expression and metabolite distribution analyses were not reported. The conversion of isoliquiritigenin to liquiritigenin may also be a limiting factor for the production of daidzein due to the lack of type II CHI in maize and other nonlegume species.
X. THE BOTTLENECK OF ISOFLAVONE PATHWAY ENGINEERING: “METABOLIC CHANNELING?” Previous experiments repeatedly demonstrated that protein–protein interactions between the key enzymes regulate isoflavone biosynthesis in legume and nonlegume plants. In general, sequential or related enzymes in metabolic pathways sometimes maintain specific interactions and are even colocalized to defined regions in the cell to form dynamic complexes called “metabolons” (Dixon and Steele, 1999; Ovadi and Srere, 2000). These multienzyme assemblies localize the accumulation of pathway intermediates and regulate competition for metabolites among branch pathways. Therefore, protein–protein interactions give rise to a higher level of complexity for controlling metabolic pathways beyond simple kinetic parameters and transcriptional control. The formation of metabolons has been well documented in many organelles, including chloroplasts, mitochondria, and peroxisomes (for reviews, see Dixon and Steele, 1999; Winkel-Shirley, 1999). For example, enzymes related to the Calvin cycle, which converts carbon dioxide into fixed carbon, are colocalized to chloroplast thylakoid membranes and have direct physical contact to ensure the channeling of substrates and products and to maintain efficient carbon fixation and conversion (Suss et al., 1993). In the cytoplasm, enzymes of secondary metabolism pathways may form similar metabolons. Multiple evidences support the existence of such metabolic channels in the phenylpropanoid pathway. When radioisotope-labeled substrates were fed to isolated buckwheat microsomes, the majority of p-coumaric acid (the product of C4H) came from phenylalanine, instead of cinnamic acid,
178
O. YU AND B. MCGONIGLE
suggesting a tight channeling of the first two enzymes of the phenylpropanoid pathway, PAL and C4H (Hrazdina and Wagner, 1985a). More direct evidence came from the colocalization of “soluble” enzymes to the ER membrane. It has been proposed that soluble enzymes form metabolons on a membrane surface, adjacent to the cytochrome P450 monooxygenase enzymes. The cytochrome P450 monooxygenases are integral membrane proteins that have been shown to be associated with the ER (Hrazdina and Wagner, 1985b). In the phenylpropanoid pathway, PAL and CHS were cofractionated with membrane-bound C4H and an ER marker protein (Wagner and Hrazdina, 1984). Additionally, the direct protein–protein interactions of CHS, CHI, F3H, and DFR in Arabidopsis were demonstrated by Burbulis and Winkel-Shirley (1999) with a yeast twohybrid assay, affinity purification, and immunocoprecipitation. In the isoflavonoid pathway, both IFS and I2’H are cytochrome P450 monooxygenases and have been shown to localize on ER (Liu and Dixon, 2001). Therefore, the type II CHI, together with other isoflavone biosynthesis enzymes, may form its own metabolic channels, independent of the flavonoid channels described in Arabidopsis. In fact, Liu and Dixon (2001) demonstrated that a key isoflavonoid phytoalexin synthesis enzyme, IOMT, is localized to the cortical ER surface only after elicitor induction, and this transient ER localization may be important for the function of IOMT. They provided a model suggesting that intermediates of the isoflavonoid pathway could flow rapidly from one enzyme center (IFS) to the next enzyme center (I2´H), which may eventually fuse with vacuoles. However, the prenyltransferases involved in the synthesis of prenylated pterocarpans and furanocoumarins are associated with plastids, not just the ER, thus requiring the shuttling of compounds between membranes and compartments (Dhillon and Brown, 1976). Although three-dimensional structures are available for many of the phenylpropanoid pathway enzymes, including alfalfa CHS and CHI, no structural model of this proposed macromolecular complex has been published. However, mechanistic studies of CHS and CHI support the need for metabolic channeling in this pathway. The nonenzymatic cyclization of chalcones into flavanones occurs in solution, but yields an enantiomeric mix of biologically inactive and active isomers. Channeling between CHS and CHI would prevent the formation of mixed isomers. Interestingly, the catalytic efficiency of CHI (kcat/Km ¼ 109 M−1 min−1) approaches the diffusion limit, so why would CHI need metabolites directed toward it? The moderate lipophilic nature of chalcones/flavanones may require close contact to limit potential sequestration in cellular membranes. Moreover, in legumes, the metabolon may “slow” CHI to protect the chalcone pool from complete conversion into naringenin (Jez et al., 2002). In many ways CHI is a paradoxical enzyme: it catalyzes a reaction that is thought to occur spontaneously yet its presence is necessary for flavonol
METABOLIC ENGINEERING OF ISOFLAVONE BIOSYNTHESIS
179
production. In Arabidopsis, CHI is a single copy gene and mutations (tt5) have been isolated on the basis of their yellow seed color (Shirley et al., 1992). These mutants survive but are more sensitive to UV light (Li et al., 1992) and have been shown not to accumulate anthocyanidins or flavonols (Shirley et al., 1995). In tomato skins there is no CHI present and the naringenin-chalcone accumulates. Overexpression of CHI causes up to a 78-fold increase in flavonoid content (Muir et al., 2001; Verhoeyen et al., 2002), all of which suggests that although the reaction from chalcone to flavone should occur spontaneously, it is dependent on the presence of the enzyme CHI. Other recent evidence suggests that CHI may have functions other than those of a catalyst. The maize CHI is capable of complementing the Arabidopsis tt5 mutant (Dong et al., 2001). Interestingly, two mutant maize proteins with only approximately 20% or 3–5% of wild-type maize CHI activity, respectively, are able to complement the tt5 mutant as well (Irnai and Grotewold, 2003; personal communications). Either a small amount of CHI activity is sufficient to drive flux or the enzymatic activity is really not that relevant for CHI function. This is suggestive of a role for CHI extraneous to the catalytic activity, perhaps acting as a structural scaffold for the enzymes involved in the various branches of the pathway. Furthermore, a soybean CHI homolog (TC177628) with 81% similarity to L. japonicus CHI3 is expressed throughout the plants (based on EST distribution). Surprisingly, this “CHI” lacks a conserved serine residue at the reaction center and does
Figure 6 Proposed macromolecular complexes in the phenylpropanoid pathway. The isoflavonoid metabolon is shown on the left. The flavonoid metabolon after CHS is on the right. Arrows indicate metabolite flow.
180
O. YU AND B. MCGONIGLE
not have the enzyme activity in vitro (Yu, unpublished data), which may support the nonenzymatic function of CHI. Because metabolites flow through the phenylpropanoid pathway branches after CHI, this enzyme is likely a key component in interacting with downstream proteins in isoflavonoid (IFS) and flavonoid (F3H, FNS, DFR) biosynthesis (Fig. 6). Legumes contain both type I CHIs that are found in all plants and convert trihydroxy-chalcones into flavanones and type II CHIs, which convert both tetrahydroxy-chalcones and trihydroxy-chalcones into flavanones (Kimura et al., 2001; Shimada et al., 2003). Understanding the interactions between CHI and other pathway enzymes under the context of metabolic channeling may be crucial for achieving high levels of isoflavones in nonlegume plants. Taken together, the metabolic channeling in secondary metabolism presents additional challenges for metabolic engineering attempts.
XI. CONCLUSIONS Although there has been significant progress in metabolic engineering of isoflavone biosynthesis, the process remains a challenge. A better understanding of the health effects derived from consuming isoflavones would significantly increase the rewards for efficacious methods. To that end, some of the novel phenotypes produced by metabolic engineering will be useful in understanding the health benefits of consuming isoflavones. Separate from the goals to increase isoflavones for human consumption, it may be desirable to produce or increase the production of isoflavones in both legumes and nonlegume plants to alter plant–microbe interactions, which may lead to increased plant disease resistance or other desirable phenotypes. The temporal and spatial expression patterns, as well as the specific species of isoflavonoid(s) necessary for the